Merge feature/pet-aware-security into main

Pet detection (YOLOv8), pet ID classifier (MobileNetV3-Small),
wildlife threat-tiered alerting, pet dashboard, training UI,
and recording labeling overlay.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.python-version

@@ -0,0 +1 @@
+3.12.0

pyproject.toml

@@ -22,6 +22,8 @@ dependencies = [
     "pynut2>=0.1.0",
     "pywebpush>=2.0.0",
     "py-vapid>=1.9.0",
+    "ultralytics>=8.2.0",
+    "torchvision>=0.18.0",
 ]
 
 [project.optional-dependencies]

tests/conftest.py

@@ -23,7 +23,7 @@ def _create_test_engine(db_path: Path):
 def tmp_data_dir(tmp_path):
     """Temporary data directory for tests."""
     data_dir = tmp_path / "data"
-    data_dir.mkdir()
+    data_dir.mkdir(exist_ok=True)
     return data_dir
 
 

tests/unit/test_config.py

@@ -1,6 +1,7 @@
 """Tests for config loading and validation."""
 
 from vigilar.config import CameraConfig, VigilarConfig
+from vigilar.config import PetsConfig, WildlifeThreatMap, WildlifeSizeHeuristics, PetActivityConfig
 
 
 def test_default_config():
@@ -33,3 +34,61 @@ def test_camera_sensitivity_bounds():
     import pytest
     with pytest.raises(Exception):
         CameraConfig(id="test", display_name="Test", rtsp_url="rtsp://localhost", motion_sensitivity=1.5)
+
+
+class TestPetsConfig:
+    def test_defaults(self):
+        cfg = PetsConfig()
+        assert cfg.enabled is False
+        assert cfg.model == "yolov8s"
+        assert cfg.confidence_threshold == 0.5
+        assert cfg.pet_id_threshold == 0.7
+        assert cfg.pet_id_low_confidence == 0.5
+        assert cfg.min_training_images == 20
+        assert cfg.crop_retention_days == 7
+
+    def test_custom_values(self):
+        cfg = PetsConfig(enabled=True, model="yolov8m", confidence_threshold=0.6)
+        assert cfg.enabled is True
+        assert cfg.model == "yolov8m"
+        assert cfg.confidence_threshold == 0.6
+
+
+class TestWildlifeThreatMap:
+    def test_defaults(self):
+        tm = WildlifeThreatMap()
+        assert "bear" in tm.predator
+        assert "bird" in tm.passive
+
+    def test_custom_mapping(self):
+        tm = WildlifeThreatMap(predator=["bear", "wolf"], nuisance=["raccoon"])
+        assert "wolf" in tm.predator
+        assert "raccoon" in tm.nuisance
+
+
+class TestWildlifeSizeHeuristics:
+    def test_defaults(self):
+        sh = WildlifeSizeHeuristics()
+        assert sh.small == 0.02
+        assert sh.medium == 0.08
+        assert sh.large == 0.15
+
+
+class TestPetActivityConfig:
+    def test_defaults(self):
+        cfg = PetActivityConfig()
+        assert cfg.daily_digest is True
+        assert cfg.highlight_clips is True
+        assert cfg.zoomie_threshold == 0.8
+
+
+class TestCameraConfigLocation:
+    def test_default_location_is_interior(self):
+        from vigilar.config import CameraConfig
+        cfg = CameraConfig(id="test", display_name="Test", rtsp_url="rtsp://x")
+        assert cfg.location == "INTERIOR"
+
+    def test_exterior_location(self):
+        from vigilar.config import CameraConfig
+        cfg = CameraConfig(id="test", display_name="Test", rtsp_url="rtsp://x", location="EXTERIOR")
+        assert cfg.location == "EXTERIOR"

tests/unit/test_constants.py

@@ -0,0 +1,51 @@
+"""Tests for new pet/wildlife constants."""
+
+from vigilar.constants import (
+    CameraLocation,
+    EventType,
+    ThreatLevel,
+    Topics,
+)
+
+
+class TestThreatLevel:
+    def test_values(self):
+        assert ThreatLevel.PREDATOR == "PREDATOR"
+        assert ThreatLevel.NUISANCE == "NUISANCE"
+        assert ThreatLevel.PASSIVE == "PASSIVE"
+
+    def test_is_strenum(self):
+        assert isinstance(ThreatLevel.PREDATOR, str)
+
+
+class TestCameraLocation:
+    def test_values(self):
+        assert CameraLocation.EXTERIOR == "EXTERIOR"
+        assert CameraLocation.INTERIOR == "INTERIOR"
+        assert CameraLocation.TRANSITION == "TRANSITION"
+
+    def test_is_strenum(self):
+        assert isinstance(CameraLocation.EXTERIOR, str)
+
+
+class TestNewEventTypes:
+    def test_pet_events_exist(self):
+        assert EventType.PET_DETECTED == "PET_DETECTED"
+        assert EventType.PET_ESCAPE == "PET_ESCAPE"
+        assert EventType.UNKNOWN_ANIMAL == "UNKNOWN_ANIMAL"
+
+    def test_wildlife_events_exist(self):
+        assert EventType.WILDLIFE_PREDATOR == "WILDLIFE_PREDATOR"
+        assert EventType.WILDLIFE_NUISANCE == "WILDLIFE_NUISANCE"
+        assert EventType.WILDLIFE_PASSIVE == "WILDLIFE_PASSIVE"
+
+
+class TestPetTopics:
+    def test_pet_detected_topic(self):
+        assert Topics.camera_pet_detected("front") == "vigilar/camera/front/pet/detected"
+
+    def test_wildlife_detected_topic(self):
+        assert Topics.camera_wildlife_detected("front") == "vigilar/camera/front/wildlife/detected"
+
+    def test_pet_location_topic(self):
+        assert Topics.pet_location("angel") == "vigilar/pets/angel/location"

tests/unit/test_crop_manager.py

@@ -0,0 +1,48 @@
+"""Tests for detection crop saving and staging cleanup."""
+
+import time
+from pathlib import Path
+
+import numpy as np
+
+from vigilar.detection.crop_manager import CropManager
+
+
+class TestCropManager:
+    def test_save_crop(self, tmp_path):
+        manager = CropManager(staging_dir=str(tmp_path / "staging"),
+                              training_dir=str(tmp_path / "training"))
+        crop = np.zeros((100, 80, 3), dtype=np.uint8)
+        path = manager.save_staging_crop(crop, species="cat", camera_id="kitchen")
+        assert Path(path).exists()
+        assert "cat" in path
+        assert "kitchen" in path
+
+    def test_promote_to_training(self, tmp_path):
+        manager = CropManager(staging_dir=str(tmp_path / "staging"),
+                              training_dir=str(tmp_path / "training"))
+        crop = np.zeros((100, 80, 3), dtype=np.uint8)
+        staging_path = manager.save_staging_crop(crop, species="cat", camera_id="kitchen")
+        training_path = manager.promote_to_training(staging_path, pet_name="angel")
+        assert Path(training_path).exists()
+        assert "angel" in training_path
+        assert not Path(staging_path).exists()
+
+    def test_cleanup_old_crops(self, tmp_path):
+        staging = tmp_path / "staging"
+        staging.mkdir(parents=True)
+
+        old_file = staging / "old_crop.jpg"
+        old_file.write_bytes(b"fake")
+        old_time = time.time() - 10 * 86400
+        import os
+        os.utime(old_file, (old_time, old_time))
+
+        new_file = staging / "new_crop.jpg"
+        new_file.write_bytes(b"fake")
+
+        manager = CropManager(staging_dir=str(staging), training_dir=str(tmp_path / "training"))
+        deleted = manager.cleanup_expired(retention_days=7)
+        assert deleted == 1
+        assert not old_file.exists()
+        assert new_file.exists()

tests/unit/test_events.py

@@ -327,3 +327,105 @@ class TestEventHistory:
         assert len(rows_offset) == 2
         # Should be different events
         assert rows[0]["id"] != rows_offset[0]["id"]
+
+
+# ---------------------------------------------------------------------------
+# Pet / Wildlife Event Classification
+# ---------------------------------------------------------------------------
+
+class TestPetEventClassification:
+    def test_pet_detected_event(self):
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/kitchen/pet/detected",
+            {"pet_id": "p1", "pet_name": "Angel", "confidence": 0.92, "camera_location": "INTERIOR"},  # noqa: E501
+        )
+        assert etype == EventType.PET_DETECTED
+        assert sev == Severity.INFO
+        assert source == "kitchen"
+
+    def test_pet_escape_exterior(self):
+        """Known pet in exterior zone → PET_ESCAPE/ALERT."""
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/front/pet/detected",
+            {"pet_id": "p1", "pet_name": "Angel", "camera_location": "EXTERIOR"},
+        )
+        assert etype == EventType.PET_ESCAPE
+        assert sev == Severity.ALERT
+
+    def test_pet_escape_transition(self):
+        """Known pet in transition zone → PET_ESCAPE/ALERT."""
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/garage/pet/detected",
+            {"pet_id": "p1", "pet_name": "Milo", "camera_location": "TRANSITION"},
+        )
+        assert etype == EventType.PET_ESCAPE
+        assert sev == Severity.ALERT
+
+    def test_unknown_animal(self):
+        """No pet_id → UNKNOWN_ANIMAL/WARNING."""
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/kitchen/pet/detected",
+            {"species": "cat", "camera_location": "INTERIOR"},
+        )
+        assert etype == EventType.UNKNOWN_ANIMAL
+        assert sev == Severity.WARNING
+
+    def test_known_pet_interior(self):
+        """Known pet in interior → PET_DETECTED/INFO."""
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/kitchen/pet/detected",
+            {"pet_id": "p1", "pet_name": "Angel", "camera_location": "INTERIOR"},
+        )
+        assert etype == EventType.PET_DETECTED
+        assert sev == Severity.INFO
+
+    def test_wildlife_predator_event(self):
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/front/wildlife/detected",
+            {"species": "bear", "threat_level": "PREDATOR"},
+        )
+        assert etype == EventType.WILDLIFE_PREDATOR
+        assert sev == Severity.CRITICAL
+        assert source == "front"
+
+    def test_wildlife_nuisance_event(self):
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/back/wildlife/detected",
+            {"species": "raccoon", "threat_level": "NUISANCE"},
+        )
+        assert etype == EventType.WILDLIFE_NUISANCE
+        assert sev == Severity.WARNING
+        assert source == "back"
+
+    def test_wildlife_passive_event(self):
+        from vigilar.events.processor import EventProcessor
+        from vigilar.constants import EventType, Severity
+        processor = EventProcessor.__new__(EventProcessor)
+        etype, sev, source = processor._classify_event(
+            "vigilar/camera/front/wildlife/detected",
+            {"species": "deer", "threat_level": "PASSIVE"},
+        )
+        assert etype == EventType.WILDLIFE_PASSIVE
+        assert sev == Severity.INFO
+        assert source == "front"

tests/unit/test_health.py

@@ -5,8 +5,10 @@ import tempfile
 from pathlib import Path
 from unittest.mock import patch
 
+from vigilar.health.digest import build_digest, format_digest
 from vigilar.health.pruner import find_prunable_recordings, calculate_disk_usage_pct
 from vigilar.health.monitor import HealthCheck, HealthStatus, check_disk, check_mqtt_port
+from vigilar.storage.schema import pet_sightings, wildlife_sightings
 
 
 class TestDiskCheck:
@@ -40,6 +42,48 @@ class TestPruner:
         assert result == []
 
 
+class TestPetDigest:
+    def test_digest_includes_pet_sightings(self, test_db, tmp_data_dir):
+        import time
+        with test_db.begin() as conn:
+            conn.execute(pet_sightings.insert().values(
+                ts=time.time(), pet_id="p1", species="cat",
+                camera_id="kitchen", confidence=0.9, labeled=1,
+            ))
+            conn.execute(pet_sightings.insert().values(
+                ts=time.time(), pet_id="p2", species="dog",
+                camera_id="kitchen", confidence=0.85, labeled=1,
+            ))
+        data = build_digest(test_db, str(tmp_data_dir), since_hours=1)
+        assert data["pet_sightings"] == 2
+
+    def test_digest_includes_wildlife(self, test_db, tmp_data_dir):
+        import time
+        with test_db.begin() as conn:
+            conn.execute(wildlife_sightings.insert().values(
+                ts=time.time(), species="bear", threat_level="PREDATOR",
+                camera_id="front", confidence=0.9,
+            ))
+            conn.execute(wildlife_sightings.insert().values(
+                ts=time.time(), species="deer", threat_level="PASSIVE",
+                camera_id="back", confidence=0.8,
+            ))
+        data = build_digest(test_db, str(tmp_data_dir), since_hours=1)
+        assert data["wildlife_predators"] == 1
+        assert data["wildlife_passive"] == 1
+
+    def test_format_includes_pets(self):
+        data = {
+            "person_detections": 2, "unknown_vehicles": 0,
+            "recordings": 5, "disk_used_gb": 100.0, "disk_used_pct": 50,
+            "since_hours": 12, "pet_sightings": 15,
+            "wildlife_predators": 0, "wildlife_nuisance": 1, "wildlife_passive": 3,
+        }
+        text = format_digest(data)
+        assert "15 pet" in text
+        assert "wildlife" in text.lower() or "nuisance" in text.lower()
+
+
 class TestMQTTCheck:
     @patch("vigilar.health.monitor.socket.create_connection")
     def test_mqtt_reachable(self, mock_conn):

tests/unit/test_pet_id.py

@@ -0,0 +1,49 @@
+"""Tests for pet ID classifier."""
+
+import numpy as np
+import pytest
+
+from vigilar.detection.pet_id import PetIDClassifier, PetIDResult
+
+
+class TestPetIDResult:
+    def test_identified(self):
+        r = PetIDResult(pet_id="pet-1", pet_name="Angel", confidence=0.9)
+        assert r.is_identified
+        assert not r.is_low_confidence
+
+    def test_low_confidence(self):
+        r = PetIDResult(pet_id="pet-1", pet_name="Angel", confidence=0.6)
+        assert r.is_identified
+        assert r.is_low_confidence
+
+    def test_unknown(self):
+        r = PetIDResult(pet_id=None, pet_name=None, confidence=0.3)
+        assert not r.is_identified
+
+
+class TestPetIDClassifier:
+    def test_not_loaded_returns_unknown(self):
+        classifier = PetIDClassifier(model_path="nonexistent.pt")
+        assert not classifier.is_loaded
+        crop = np.zeros((224, 224, 3), dtype=np.uint8)
+        result = classifier.identify(crop, species="cat")
+        assert not result.is_identified
+
+    def test_no_pets_registered_returns_unknown(self):
+        classifier = PetIDClassifier(model_path="nonexistent.pt")
+        assert classifier.pet_count == 0
+
+    def test_register_pet(self):
+        classifier = PetIDClassifier(model_path="nonexistent.pt")
+        classifier.register_pet("pet-1", "Angel", "cat")
+        classifier.register_pet("pet-2", "Milo", "dog")
+        assert classifier.pet_count == 2
+
+    def test_species_filter(self):
+        classifier = PetIDClassifier(model_path="nonexistent.pt")
+        classifier.register_pet("pet-1", "Angel", "cat")
+        classifier.register_pet("pet-2", "Taquito", "cat")
+        classifier.register_pet("pet-3", "Milo", "dog")
+        assert len(classifier.get_pets_by_species("cat")) == 2
+        assert len(classifier.get_pets_by_species("dog")) == 1

tests/unit/test_pet_queries.py

@@ -0,0 +1,101 @@
+"""Tests for pet and wildlife query functions."""
+
+import time
+
+from vigilar.storage.queries import (
+    insert_pet,
+    get_pet,
+    get_all_pets,
+    insert_pet_sighting,
+    get_pet_sightings,
+    get_pet_last_location,
+    insert_wildlife_sighting,
+    get_wildlife_sightings,
+    insert_training_image,
+    get_training_images,
+    get_unlabeled_sightings,
+    label_sighting,
+)
+
+
+class TestPetCRUD:
+    def test_insert_and_get_pet(self, test_db):
+        pet_id = insert_pet(test_db, name="Angel", species="cat", breed="DSH",
+                            color_description="black")
+        pet = get_pet(test_db, pet_id)
+        assert pet is not None
+        assert pet["name"] == "Angel"
+        assert pet["species"] == "cat"
+        assert pet["training_count"] == 0
+
+    def test_get_all_pets(self, test_db):
+        insert_pet(test_db, name="Angel", species="cat")
+        insert_pet(test_db, name="Milo", species="dog")
+        all_pets = get_all_pets(test_db)
+        assert len(all_pets) == 2
+
+
+class TestPetSightings:
+    def test_insert_and_query(self, test_db):
+        pet_id = insert_pet(test_db, name="Angel", species="cat")
+        insert_pet_sighting(test_db, pet_id=pet_id, species="cat",
+                            camera_id="kitchen", confidence=0.92)
+        sightings = get_pet_sightings(test_db, limit=10)
+        assert len(sightings) == 1
+        assert sightings[0]["camera_id"] == "kitchen"
+
+    def test_last_location(self, test_db):
+        pet_id = insert_pet(test_db, name="Angel", species="cat")
+        insert_pet_sighting(test_db, pet_id=pet_id, species="cat",
+                            camera_id="kitchen", confidence=0.9)
+        insert_pet_sighting(test_db, pet_id=pet_id, species="cat",
+                            camera_id="living_room", confidence=0.95)
+        loc = get_pet_last_location(test_db, pet_id)
+        assert loc is not None
+        assert loc["camera_id"] == "living_room"
+
+    def test_unlabeled_sightings(self, test_db):
+        insert_pet_sighting(test_db, pet_id=None, species="cat",
+                            camera_id="kitchen", confidence=0.6, crop_path="/tmp/crop.jpg")
+        unlabeled = get_unlabeled_sightings(test_db, limit=10)
+        assert len(unlabeled) == 1
+        assert unlabeled[0]["labeled"] == 0
+
+    def test_label_sighting(self, test_db):
+        pet_id = insert_pet(test_db, name="Angel", species="cat")
+        insert_pet_sighting(test_db, pet_id=None, species="cat",
+                            camera_id="kitchen", confidence=0.6)
+        sightings = get_unlabeled_sightings(test_db, limit=10)
+        sighting_id = sightings[0]["id"]
+        label_sighting(test_db, sighting_id, pet_id)
+        updated = get_pet_sightings(test_db, pet_id=pet_id)
+        assert len(updated) == 1
+        assert updated[0]["labeled"] == 1
+
+
+class TestWildlifeSightings:
+    def test_insert_and_query(self, test_db):
+        insert_wildlife_sighting(test_db, species="bear", threat_level="PREDATOR",
+                                 camera_id="front", confidence=0.88)
+        sightings = get_wildlife_sightings(test_db, limit=10)
+        assert len(sightings) == 1
+        assert sightings[0]["threat_level"] == "PREDATOR"
+
+    def test_filter_by_threat(self, test_db):
+        insert_wildlife_sighting(test_db, species="bear", threat_level="PREDATOR",
+                                 camera_id="front", confidence=0.88)
+        insert_wildlife_sighting(test_db, species="deer", threat_level="PASSIVE",
+                                 camera_id="back", confidence=0.75)
+        predators = get_wildlife_sightings(test_db, threat_level="PREDATOR")
+        assert len(predators) == 1
+
+
+class TestTrainingImages:
+    def test_insert_and_query(self, test_db):
+        pet_id = insert_pet(test_db, name="Angel", species="cat")
+        insert_training_image(test_db, pet_id=pet_id,
+                              image_path="/var/vigilar/pets/training/angel/001.jpg",
+                              source="upload")
+        images = get_training_images(test_db, pet_id)
+        assert len(images) == 1
+        assert images[0]["source"] == "upload"

tests/unit/test_pets_api.py

@@ -0,0 +1,109 @@
+"""Tests for pets web blueprint API endpoints."""
+
+import pytest
+from sqlalchemy import create_engine
+
+from vigilar.config import VigilarConfig
+from vigilar.storage.schema import metadata
+from vigilar.web.app import create_app
+
+
+@pytest.fixture
+def client(tmp_path):
+    """Flask test client wired to an in-memory test DB."""
+    db_path = tmp_path / "pets_test.db"
+    engine = create_engine(f"sqlite:///{db_path}", echo=False)
+    metadata.create_all(engine)
+
+    cfg = VigilarConfig()
+    app = create_app(cfg)
+    app.config["DB_ENGINE"] = engine
+    app.config["TESTING"] = True
+
+    with app.test_client() as c:
+        yield c
+
+
+class TestPetsAPI:
+    def test_register_pet(self, client):
+        resp = client.post("/pets/register", json={
+            "name": "Angel", "species": "cat", "breed": "DSH",
+            "color_description": "black",
+        })
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert data["name"] == "Angel"
+        assert "id" in data
+
+    def test_get_pet_status(self, client):
+        client.post("/pets/register", json={"name": "Angel", "species": "cat"})
+        resp = client.get("/pets/api/status")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert len(data["pets"]) == 1
+        assert data["pets"][0]["name"] == "Angel"
+
+    def test_get_sightings_empty(self, client):
+        resp = client.get("/pets/api/sightings")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert data["sightings"] == []
+
+    def test_get_wildlife_empty(self, client):
+        resp = client.get("/pets/api/wildlife")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert data["sightings"] == []
+
+    def test_get_unlabeled_empty(self, client):
+        resp = client.get("/pets/api/unlabeled")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert data["crops"] == []
+
+    def test_pets_dashboard_loads(self, client):
+        resp = client.get("/pets/")
+        assert resp.status_code == 200
+
+    def test_register_pet_missing_fields(self, client):
+        resp = client.post("/pets/register", json={"name": "Buddy"})
+        assert resp.status_code == 400
+
+    def test_label_crop(self, client):
+        # Register a pet first
+        reg = client.post("/pets/register", json={"name": "Angel", "species": "cat"})
+        pet_id = reg.get_json()["id"]
+
+        # Insert an unlabeled sighting directly via DB
+
+        from vigilar.storage.queries import insert_pet_sighting
+        app_engine = client.application.config["DB_ENGINE"]
+        sighting_id = insert_pet_sighting(app_engine, species="cat", camera_id="cam1",
+                                          confidence=0.9)
+
+        resp = client.post(f"/pets/{sighting_id}/label", json={"pet_id": pet_id})
+        assert resp.status_code == 200
+        assert resp.get_json()["ok"] is True
+
+    def test_delete_pet(self, client):
+        reg = client.post("/pets/register", json={"name": "Ghost", "species": "cat"})
+        pet_id = reg.get_json()["id"]
+
+        resp = client.delete(f"/pets/{pet_id}/delete")
+        assert resp.status_code == 200
+        assert resp.get_json()["ok"] is True
+
+        status = client.get("/pets/api/status").get_json()
+        assert len(status["pets"]) == 0
+
+    def test_training_status(self, client):
+        resp = client.get("/pets/api/training-status")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert "status" in data
+
+    def test_highlights(self, client):
+        resp = client.get("/pets/api/highlights")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert "highlights" in data

tests/unit/test_profiles.py

@@ -75,3 +75,56 @@ class TestActionForEvent:
         action, recipients = get_action_for_event(profile, "person", "front_door")
         assert action == "quiet_log"
         assert recipients == "none"
+
+
+class TestPetAlertRouting:
+    def test_known_pet_exterior_gets_push(self):
+        rules = [
+            AlertProfileRule(detection_type="known_pet", camera_location="EXTERIOR",
+                             action="push_and_record", recipients="all"),
+            AlertProfileRule(detection_type="known_pet", camera_location="INTERIOR",
+                             action="quiet_log", recipients="none"),
+        ]
+        profile = _make_profile("Away", ["EMPTY"], rules=rules)
+        action, recipients = get_action_for_event(profile, "known_pet", "front_entrance",
+                                                  camera_location="EXTERIOR")
+        assert action == "push_and_record"
+
+    def test_known_pet_interior_gets_quiet_log(self):
+        rules = [
+            AlertProfileRule(detection_type="known_pet", camera_location="EXTERIOR",
+                             action="push_and_record", recipients="all"),
+            AlertProfileRule(detection_type="known_pet", camera_location="INTERIOR",
+                             action="quiet_log", recipients="none"),
+        ]
+        profile = _make_profile("Home", ["ALL_HOME"], rules=rules)
+        action, recipients = get_action_for_event(profile, "known_pet", "kitchen",
+                                                  camera_location="INTERIOR")
+        assert action == "quiet_log"
+
+    def test_wildlife_predator_gets_urgent(self):
+        rules = [
+            AlertProfileRule(detection_type="wildlife_predator",
+                             action="push_and_record", recipients="all"),
+        ]
+        profile = _make_profile("Always", ["EMPTY", "ALL_HOME"], rules=rules)
+        action, _ = get_action_for_event(profile, "wildlife_predator", "front",
+                                         camera_location="EXTERIOR")
+        assert action == "push_and_record"
+
+    def test_camera_location_any_matches_all(self):
+        rules = [
+            AlertProfileRule(detection_type="wildlife_predator", camera_location="any",
+                             action="push_and_record", recipients="all"),
+        ]
+        profile = _make_profile("Always", ["EMPTY"], rules=rules)
+        action, _ = get_action_for_event(profile, "wildlife_predator", "kitchen",
+                                         camera_location="INTERIOR")
+        assert action == "push_and_record"
+
+    def test_backward_compatible_without_camera_location(self):
+        """Existing calls without camera_location still work."""
+        rules = [AlertProfileRule(detection_type="person", action="push_and_record", recipients="all")]
+        profile = _make_profile("Away", ["EMPTY"], rules=rules)
+        action, _ = get_action_for_event(profile, "person", "front_door")
+        assert action == "push_and_record"

tests/unit/test_schema.py

@@ -2,7 +2,7 @@
 
 from sqlalchemy import create_engine, inspect
 
-from vigilar.storage.schema import metadata
+from vigilar.storage.schema import metadata, pets, pet_sightings, wildlife_sightings, pet_training_images
 
 
 def test_tables_created(tmp_path):
@@ -21,3 +21,52 @@ def test_tables_created(tmp_path):
     ]
     for name in expected:
         assert name in table_names, f"Missing table: {name}"
+
+
+class TestPetTables:
+    def test_pets_table_exists(self, test_db):
+        with test_db.connect() as conn:
+            result = conn.execute(pets.insert().values(
+                id="pet-1", name="Angel", species="cat", breed="DSH",
+                color_description="black", training_count=0, created_at=1000.0,
+            ))
+            row = conn.execute(pets.select().where(pets.c.id == "pet-1")).first()
+            assert row is not None
+            assert row.name == "Angel"
+            assert row.species == "cat"
+
+    def test_pet_sightings_table(self, test_db):
+        with test_db.begin() as conn:
+            conn.execute(pets.insert().values(
+                id="pet-1", name="Angel", species="cat", training_count=0, created_at=1000.0,
+            ))
+            conn.execute(pet_sightings.insert().values(
+                ts=1000.0, pet_id="pet-1", species="cat", camera_id="kitchen",
+                confidence=0.92, labeled=True,
+            ))
+            rows = conn.execute(pet_sightings.select()).fetchall()
+            assert len(rows) == 1
+            assert rows[0].camera_id == "kitchen"
+
+    def test_wildlife_sightings_table(self, test_db):
+        with test_db.begin() as conn:
+            conn.execute(wildlife_sightings.insert().values(
+                ts=1000.0, species="bear", threat_level="PREDATOR",
+                camera_id="front", confidence=0.88,
+            ))
+            rows = conn.execute(wildlife_sightings.select()).fetchall()
+            assert len(rows) == 1
+            assert rows[0].threat_level == "PREDATOR"
+
+    def test_pet_training_images_table(self, test_db):
+        with test_db.begin() as conn:
+            conn.execute(pets.insert().values(
+                id="pet-1", name="Angel", species="cat", training_count=0, created_at=1000.0,
+            ))
+            conn.execute(pet_training_images.insert().values(
+                pet_id="pet-1", image_path="/var/vigilar/pets/training/angel/001.jpg",
+                source="upload", created_at=1000.0,
+            ))
+            rows = conn.execute(pet_training_images.select()).fetchall()
+            assert len(rows) == 1
+            assert rows[0].source == "upload"

tests/unit/test_trainer.py

@@ -0,0 +1,55 @@
+"""Tests for pet ID model trainer."""
+
+import os
+from pathlib import Path
+
+import numpy as np
+import pytest
+
+from vigilar.detection.trainer import PetTrainer, TrainingStatus
+
+
+class TestTrainingStatus:
+    def test_initial_status(self):
+        status = TrainingStatus()
+        assert status.is_training is False
+        assert status.progress == 0.0
+        assert status.error is None
+
+
+class TestPetTrainer:
+    def test_check_readiness_no_pets(self, tmp_path):
+        trainer = PetTrainer(training_dir=str(tmp_path), model_output_path=str(tmp_path / "model.pt"))
+        ready, msg = trainer.check_readiness(min_images=20)
+        assert not ready
+        assert "No pet" in msg
+
+    def test_check_readiness_insufficient_images(self, tmp_path):
+        pet_dir = tmp_path / "angel"
+        pet_dir.mkdir()
+        for i in range(5):
+            (pet_dir / f"{i}.jpg").write_bytes(b"fake")
+
+        trainer = PetTrainer(training_dir=str(tmp_path), model_output_path=str(tmp_path / "model.pt"))
+        ready, msg = trainer.check_readiness(min_images=20)
+        assert not ready
+        assert "angel" in msg.lower()
+
+    def test_check_readiness_sufficient_images(self, tmp_path):
+        for name in ["angel", "taquito"]:
+            pet_dir = tmp_path / name
+            pet_dir.mkdir()
+            for i in range(25):
+                (pet_dir / f"{i}.jpg").write_bytes(b"fake")
+
+        trainer = PetTrainer(training_dir=str(tmp_path), model_output_path=str(tmp_path / "model.pt"))
+        ready, msg = trainer.check_readiness(min_images=20)
+        assert ready
+
+    def test_get_class_names(self, tmp_path):
+        for name in ["angel", "milo", "taquito"]:
+            (tmp_path / name).mkdir()
+
+        trainer = PetTrainer(training_dir=str(tmp_path), model_output_path=str(tmp_path / "model.pt"))
+        names = trainer.get_class_names()
+        assert names == ["angel", "milo", "taquito"]  # sorted

tests/unit/test_wildlife.py

@@ -0,0 +1,50 @@
+"""Tests for wildlife threat classification."""
+
+from vigilar.config import WildlifeConfig, WildlifeThreatMap, WildlifeSizeHeuristics
+from vigilar.detection.person import Detection
+from vigilar.detection.wildlife import classify_wildlife_threat
+
+
+def _make_config(**kwargs):
+    return WildlifeConfig(**kwargs)
+
+
+class TestWildlifeThreatClassification:
+    def test_bear_is_predator(self):
+        cfg = _make_config()
+        d = Detection(class_name="bear", class_id=21, confidence=0.9, bbox=(100, 100, 200, 300))
+        level, species = classify_wildlife_threat(d, cfg, frame_area=1920 * 1080)
+        assert level == "PREDATOR"
+        assert species == "bear"
+
+    def test_bird_is_passive(self):
+        cfg = _make_config()
+        d = Detection(class_name="bird", class_id=14, confidence=0.8, bbox=(10, 10, 30, 20))
+        level, species = classify_wildlife_threat(d, cfg, frame_area=1920 * 1080)
+        assert level == "PASSIVE"
+        assert species == "bird"
+
+    def test_unknown_small_is_nuisance(self):
+        cfg = _make_config()
+        d = Detection(class_name="unknown", class_id=99, confidence=0.7, bbox=(100, 100, 30, 30))
+        level, species = classify_wildlife_threat(d, cfg, frame_area=1920 * 1080)
+        assert level == "NUISANCE"
+        assert species == "unknown"
+
+    def test_unknown_medium_is_predator(self):
+        cfg = _make_config()
+        d = Detection(class_name="unknown", class_id=99, confidence=0.7, bbox=(100, 100, 300, 300))
+        level, species = classify_wildlife_threat(d, cfg, frame_area=1920 * 1080)
+        assert level == "PREDATOR"
+
+    def test_unknown_large_is_passive(self):
+        cfg = _make_config()
+        d = Detection(class_name="unknown", class_id=99, confidence=0.7, bbox=(100, 100, 600, 500))
+        level, species = classify_wildlife_threat(d, cfg, frame_area=1920 * 1080)
+        assert level == "PASSIVE"
+
+    def test_custom_threat_map(self):
+        cfg = _make_config(threat_map=WildlifeThreatMap(predator=["bear", "wolf"]))
+        d = Detection(class_name="wolf", class_id=99, confidence=0.85, bbox=(100, 100, 200, 200))
+        level, _ = classify_wildlife_threat(d, cfg, frame_area=1920 * 1080)
+        assert level == "PREDATOR"

tests/unit/test_yolo_detector.py

@@ -0,0 +1,52 @@
+"""Tests for YOLOv8 unified detector."""
+
+import numpy as np
+import pytest
+
+from vigilar.detection.person import Detection
+from vigilar.detection.yolo import YOLODetector, ANIMAL_CLASSES, WILDLIFE_CLASSES
+
+
+class TestYOLOConstants:
+    def test_animal_classes(self):
+        assert "cat" in ANIMAL_CLASSES
+        assert "dog" in ANIMAL_CLASSES
+
+    def test_wildlife_classes(self):
+        assert "bear" in WILDLIFE_CLASSES
+        assert "bird" in WILDLIFE_CLASSES
+
+    def test_no_overlap_animal_wildlife(self):
+        assert not ANIMAL_CLASSES.intersection(WILDLIFE_CLASSES)
+
+
+class TestYOLODetector:
+    def test_initializes_without_model(self):
+        detector = YOLODetector(model_path="nonexistent.pt", confidence_threshold=0.5)
+        assert not detector.is_loaded
+
+    def test_detect_returns_empty_when_not_loaded(self):
+        detector = YOLODetector(model_path="nonexistent.pt")
+        frame = np.zeros((480, 640, 3), dtype=np.uint8)
+        detections = detector.detect(frame)
+        assert detections == []
+
+    def test_classify_detection_person(self):
+        d = Detection(class_name="person", class_id=0, confidence=0.9, bbox=(10, 20, 100, 200))
+        assert YOLODetector.classify(d) == "person"
+
+    def test_classify_detection_vehicle(self):
+        d = Detection(class_name="car", class_id=2, confidence=0.85, bbox=(10, 20, 100, 200))
+        assert YOLODetector.classify(d) == "vehicle"
+
+    def test_classify_detection_domestic_animal(self):
+        d = Detection(class_name="cat", class_id=15, confidence=0.9, bbox=(10, 20, 100, 200))
+        assert YOLODetector.classify(d) == "domestic_animal"
+
+    def test_classify_detection_wildlife(self):
+        d = Detection(class_name="bear", class_id=21, confidence=0.8, bbox=(10, 20, 100, 200))
+        assert YOLODetector.classify(d) == "wildlife"
+
+    def test_classify_detection_other(self):
+        d = Detection(class_name="chair", class_id=56, confidence=0.7, bbox=(10, 20, 100, 200))
+        assert YOLODetector.classify(d) == "other"

vigilar/alerts/profiles.py

@@ -1,9 +1,8 @@
 """Smart alert profile matching engine."""
 
 import logging
-from datetime import datetime
 
-from vigilar.config import AlertProfileConfig, AlertProfileRule
+from vigilar.config import AlertProfileConfig
 from vigilar.constants import HouseholdState
 
 log = logging.getLogger(__name__)
@@ -46,11 +45,19 @@ def get_action_for_event(
     profile: AlertProfileConfig,
     detection_type: str,
     camera_id: str,
+    camera_location: str | None = None,
 ) -> tuple[str, str]:
     for rule in profile.rules:
         if rule.detection_type != detection_type:
             continue
-        if rule.camera_location not in ("any", camera_id):
-            continue
-        return rule.action, rule.recipients
+        # Check camera_location match
+        if rule.camera_location == "any":
+            return rule.action, rule.recipients
+        if rule.camera_location == camera_id:
+            return rule.action, rule.recipients
+        if camera_location and rule.camera_location == camera_location:
+            return rule.action, rule.recipients
+        if not camera_location and rule.camera_location == "any":
+            return rule.action, rule.recipients
+        continue
     return "quiet_log", "none"

vigilar/camera/worker.py

@@ -13,15 +13,18 @@ import signal
 import time
 
 import cv2
-import numpy as np
 
 from vigilar.bus import MessageBus
 from vigilar.camera.hls import HLSStreamer, RemoteHLSStreamer
 from vigilar.camera.motion import MotionDetector
 from vigilar.camera.recorder import AdaptiveRecorder
 from vigilar.camera.ring_buffer import RingBuffer
-from vigilar.config import CameraConfig, MQTTConfig, RemoteConfig
+from vigilar.config import CameraConfig, MQTTConfig, PetsConfig, RemoteConfig
 from vigilar.constants import Topics
+from vigilar.detection.crop_manager import CropManager
+from vigilar.detection.pet_id import PetIDClassifier
+from vigilar.detection.wildlife import classify_wildlife_threat
+from vigilar.detection.yolo import YOLODetector
 
 log = logging.getLogger(__name__)
 
@@ -46,6 +49,7 @@ def run_camera_worker(
     recordings_dir: str,
     hls_dir: str,
     remote_cfg: RemoteConfig | None = None,
+    pets_cfg: PetsConfig | None = None,
 ) -> None:
     """Main entry point for a camera worker process."""
     camera_id = camera_cfg.id
@@ -107,8 +111,29 @@ def run_camera_worker(
             bitrate_kbps=remote_cfg.remote_hls_bitrate_kbps,
         )
 
+    # Object detection (YOLOv8 unified detector)
+    yolo_detector = None
+    pet_classifier = None
+    crop_manager = None
+    if pets_cfg and pets_cfg.enabled:
+        yolo_detector = YOLODetector(
+            model_path=pets_cfg.model_path,
+            confidence_threshold=pets_cfg.confidence_threshold,
+        )
+        if pets_cfg.pet_id_enabled:
+            pet_classifier = PetIDClassifier(
+                model_path=pets_cfg.pet_id_model_path,
+                high_threshold=pets_cfg.pet_id_threshold,
+                low_threshold=pets_cfg.pet_id_low_confidence,
+            )
+        crop_manager = CropManager(
+            staging_dir=pets_cfg.crop_staging_dir,
+            training_dir=pets_cfg.training_dir,
+        )
+
     state = CameraState()
     shutdown = False
+    last_detection_time: float = 0
 
     def handle_signal(signum, frame):
         nonlocal shutdown
@@ -243,6 +268,74 @@ def run_camera_worker(
                 if state.frame_count % idle_skip_factor == 0:
                     recorder.write_frame(frame)
 
+        # Run object detection on motion frames — throttled to 1 inference/second
+        if (state.motion_active and yolo_detector and yolo_detector.is_loaded
+                and now - last_detection_time >= 1.0):
+            last_detection_time = now
+            detections = yolo_detector.detect(frame)
+            for det in detections:
+                category = YOLODetector.classify(det)
+                if category == "domestic_animal":
+                    # Crop for pet ID and staging
+                    x, y, w, h = det.bbox
+                    crop = frame[max(0, y):y + h, max(0, x):x + w]
+
+                    crop_path = None
+                    if crop_manager and crop.size > 0:
+                        crop_path = crop_manager.save_staging_crop(
+                            crop, species=det.class_name, camera_id=camera_id
+                        )
+
+                    pet_result = None
+                    if pet_classifier and pet_classifier.is_loaded and crop.size > 0:
+                        pet_result = pet_classifier.identify(crop, species=det.class_name)
+
+                    payload = {
+                        "species": det.class_name,
+                        "confidence": round(det.confidence, 3),
+                        "camera_location": camera_cfg.location,
+                        "crop_path": crop_path,
+                    }
+                    if pet_result and pet_result.is_identified:
+                        payload["pet_id"] = pet_result.pet_id
+                        payload["pet_name"] = pet_result.pet_name
+                        payload["pet_confidence"] = round(pet_result.confidence, 3)
+                        bus.publish_event(
+                            Topics.pet_location(pet_result.pet_name.lower()),
+                            camera_id=camera_id,
+                            camera_location=camera_cfg.location,
+                        )
+
+                    bus.publish_event(Topics.camera_pet_detected(camera_id), **payload)
+
+                elif category == "wildlife" and pets_cfg:
+                    frame_area = frame.shape[0] * frame.shape[1]
+                    threat_level, species = classify_wildlife_threat(
+                        det, pets_cfg.wildlife, frame_area,
+                    )
+                    bus.publish_event(
+                        Topics.camera_wildlife_detected(camera_id),
+                        species=species,
+                        threat_level=threat_level,
+                        confidence=round(det.confidence, 3),
+                        camera_location=camera_cfg.location,
+                        crop_path=None,
+                    )
+
+                elif category == "person":
+                    bus.publish_event(
+                        Topics.camera_motion_start(camera_id),
+                        detection="person",
+                        confidence=round(det.confidence, 3),
+                    )
+
+                elif category == "vehicle":
+                    bus.publish_event(
+                        Topics.camera_motion_start(camera_id),
+                        detection="vehicle",
+                        confidence=round(det.confidence, 3),
+                    )
+
         # Heartbeat every 10 seconds
         if now - last_heartbeat >= 10:
             last_heartbeat = now

vigilar/config.py

@@ -20,9 +20,9 @@ from vigilar.constants import (
     DEFAULT_RETENTION_DAYS,
     DEFAULT_UPS_POLL_INTERVAL_S,
     DEFAULT_WEB_PORT,
+    CameraLocation,
 )
 
-
 # --- Camera Config ---
 
 class CameraConfig(BaseModel):
@@ -43,6 +43,7 @@ class CameraConfig(BaseModel):
     resolution_capture: list[int] = Field(default_factory=lambda: [1920, 1080])
     resolution_motion: list[int] = Field(default_factory=lambda: [640, 360])
     zones: list["CameraZone"] = Field(default_factory=list)
+    location: CameraLocation = CameraLocation.INTERIOR
 
 
 # --- Sensor Config ---
@@ -239,6 +240,48 @@ class HealthConfig(BaseModel):
     daily_digest_time: str = "08:00"
 
 
+# --- Pet Detection Config ---
+
+class WildlifeThreatMap(BaseModel):
+    predator: list[str] = Field(default_factory=lambda: ["bear"])
+    nuisance: list[str] = Field(default_factory=list)
+    passive: list[str] = Field(default_factory=lambda: ["bird", "horse", "cow", "sheep"])
+
+
+class WildlifeSizeHeuristics(BaseModel):
+    small: float = 0.02   # < 2% of frame → nuisance
+    medium: float = 0.08  # 2-8% → predator
+    large: float = 0.15   # > 8% → passive (deer-sized)
+
+
+class WildlifeConfig(BaseModel):
+    threat_map: WildlifeThreatMap = Field(default_factory=WildlifeThreatMap)
+    size_heuristics: WildlifeSizeHeuristics = Field(default_factory=WildlifeSizeHeuristics)
+
+
+class PetActivityConfig(BaseModel):
+    daily_digest: bool = True
+    highlight_clips: bool = True
+    zoomie_threshold: float = 0.8
+
+
+class PetsConfig(BaseModel):
+    enabled: bool = False
+    model: str = "yolov8s"
+    model_path: str = "/var/vigilar/models/yolov8s.pt"
+    confidence_threshold: float = 0.5
+    pet_id_enabled: bool = True
+    pet_id_model_path: str = "/var/vigilar/models/pet_id.pt"
+    pet_id_threshold: float = 0.7
+    pet_id_low_confidence: float = 0.5
+    training_dir: str = "/var/vigilar/pets/training"
+    crop_staging_dir: str = "/var/vigilar/pets/staging"
+    crop_retention_days: int = 7
+    min_training_images: int = 20
+    wildlife: WildlifeConfig = Field(default_factory=WildlifeConfig)
+    activity: PetActivityConfig = Field(default_factory=PetActivityConfig)
+
+
 # --- Rule Config ---
 
 class RuleCondition(BaseModel):
@@ -284,6 +327,7 @@ class VigilarConfig(BaseModel):
     detection: DetectionConfig = Field(default_factory=DetectionConfig)
     vehicles: VehicleConfig = Field(default_factory=VehicleConfig)
     health: HealthConfig = Field(default_factory=HealthConfig)
+    pets: PetsConfig = Field(default_factory=PetsConfig)
     cameras: list[CameraConfig] = Field(default_factory=list)
     sensors: list[SensorConfig] = Field(default_factory=list)
     sensor_gpio: SensorGPIOConfig = Field(default_factory=SensorGPIOConfig, alias="sensors.gpio")

vigilar/constants.py

@@ -2,7 +2,6 @@
 
 from enum import StrEnum
 
-
 # --- Arm States ---
 
 class ArmState(StrEnum):
@@ -40,6 +39,12 @@ class EventType(StrEnum):
     VEHICLE_DETECTED = "VEHICLE_DETECTED"
     KNOWN_VEHICLE_ARRIVED = "KNOWN_VEHICLE_ARRIVED"
     UNKNOWN_VEHICLE_DETECTED = "UNKNOWN_VEHICLE_DETECTED"
+    PET_DETECTED = "PET_DETECTED"
+    PET_ESCAPE = "PET_ESCAPE"
+    UNKNOWN_ANIMAL = "UNKNOWN_ANIMAL"
+    WILDLIFE_PREDATOR = "WILDLIFE_PREDATOR"
+    WILDLIFE_NUISANCE = "WILDLIFE_NUISANCE"
+    WILDLIFE_PASSIVE = "WILDLIFE_PASSIVE"
 
 
 # --- Sensor Types ---
@@ -68,6 +73,8 @@ class RecordingTrigger(StrEnum):
     MANUAL = "MANUAL"
     PERSON = "PERSON"
     VEHICLE = "VEHICLE"
+    PET = "PET"
+    WILDLIFE = "WILDLIFE"
 
 
 # --- Alert Channels ---
@@ -103,6 +110,22 @@ class HouseholdState(StrEnum):
     ALL_HOME = "ALL_HOME"
 
 
+# --- Threat Levels (Wildlife) ---
+
+class ThreatLevel(StrEnum):
+    PREDATOR = "PREDATOR"
+    NUISANCE = "NUISANCE"
+    PASSIVE = "PASSIVE"
+
+
+# --- Camera Location ---
+
+class CameraLocation(StrEnum):
+    EXTERIOR = "EXTERIOR"
+    INTERIOR = "INTERIOR"
+    TRANSITION = "TRANSITION"
+
+
 # --- MQTT Topics ---
 
 class Topics:
@@ -148,6 +171,19 @@ class Topics:
 
     PRESENCE_STATUS = "vigilar/presence/status"
 
+    # Pet
+    @staticmethod
+    def camera_pet_detected(camera_id: str) -> str:
+        return f"vigilar/camera/{camera_id}/pet/detected"
+
+    @staticmethod
+    def camera_wildlife_detected(camera_id: str) -> str:
+        return f"vigilar/camera/{camera_id}/wildlife/detected"
+
+    @staticmethod
+    def pet_location(pet_name: str) -> str:
+        return f"vigilar/pets/{pet_name}/location"
+
     # System
     SYSTEM_ARM_STATE = "vigilar/system/arm_state"
     SYSTEM_ALERT = "vigilar/system/alert"

vigilar/detection/crop_manager.py

@@ -0,0 +1,48 @@
+"""Manage detection crop images for training and staging."""
+
+import logging
+import shutil
+import time
+from pathlib import Path
+
+import cv2
+import numpy as np
+
+log = logging.getLogger(__name__)
+
+
+class CropManager:
+    def __init__(self, staging_dir: str, training_dir: str):
+        self._staging_dir = Path(staging_dir)
+        self._training_dir = Path(training_dir)
+
+    def save_staging_crop(self, crop: np.ndarray, species: str, camera_id: str) -> str:
+        self._staging_dir.mkdir(parents=True, exist_ok=True)
+        timestamp = int(time.time() * 1000)
+        filename = f"{species}_{camera_id}_{timestamp}.jpg"
+        filepath = self._staging_dir / filename
+        cv2.imwrite(str(filepath), crop)
+        return str(filepath)
+
+    def promote_to_training(self, staging_path: str, pet_name: str) -> str:
+        pet_dir = self._training_dir / pet_name.lower()
+        pet_dir.mkdir(parents=True, exist_ok=True)
+        src = Path(staging_path)
+        dst = pet_dir / src.name
+        shutil.move(str(src), str(dst))
+        return str(dst)
+
+    def cleanup_expired(self, retention_days: int = 7) -> int:
+        if not self._staging_dir.exists():
+            return 0
+
+        cutoff = time.time() - retention_days * 86400
+        deleted = 0
+        for filepath in self._staging_dir.iterdir():
+            if filepath.is_file() and filepath.stat().st_mtime < cutoff:
+                filepath.unlink()
+                deleted += 1
+
+        if deleted:
+            log.info("Cleaned up %d expired staging crops", deleted)
+        return deleted

vigilar/detection/pet_id.py

@@ -0,0 +1,132 @@
+"""Pet identification classifier using MobileNetV3-Small."""
+
+import logging
+from dataclasses import dataclass
+from pathlib import Path
+
+import numpy as np
+
+log = logging.getLogger(__name__)
+
+DEFAULT_HIGH_THRESHOLD = 0.7
+DEFAULT_LOW_THRESHOLD = 0.5
+
+
+@dataclass
+class PetIDResult:
+    pet_id: str | None
+    pet_name: str | None
+    confidence: float
+    high_threshold: float = DEFAULT_HIGH_THRESHOLD
+    low_threshold: float = DEFAULT_LOW_THRESHOLD
+
+    @property
+    def is_identified(self) -> bool:
+        return self.pet_id is not None and self.confidence >= self.low_threshold
+
+    @property
+    def is_low_confidence(self) -> bool:
+        return (
+            self.pet_id is not None
+            and self.low_threshold <= self.confidence < self.high_threshold
+        )
+
+
+@dataclass
+class RegisteredPet:
+    pet_id: str
+    name: str
+    species: str
+    class_index: int
+
+
+class PetIDClassifier:
+    def __init__(
+        self,
+        model_path: str,
+        high_threshold: float = DEFAULT_HIGH_THRESHOLD,
+        low_threshold: float = DEFAULT_LOW_THRESHOLD,
+    ):
+        self._model_path = model_path
+        self._high_threshold = high_threshold
+        self._low_threshold = low_threshold
+        self._model = None
+        self.is_loaded = False
+        self._pets: list[RegisteredPet] = []
+
+        if Path(model_path).exists():
+            try:
+                import torch
+                self._model = torch.load(model_path, map_location="cpu", weights_only=False)
+                self._model.eval()
+                self.is_loaded = True
+                log.info("Pet ID model loaded from %s", model_path)
+            except Exception as e:
+                log.error("Failed to load pet ID model: %s", e)
+        else:
+            log.info(
+                "Pet ID model not found at %s — identification disabled until trained",
+                model_path,
+            )
+
+    @property
+    def pet_count(self) -> int:
+        return len(self._pets)
+
+    def register_pet(self, pet_id: str, name: str, species: str) -> None:
+        idx = len(self._pets)
+        self._pets.append(RegisteredPet(pet_id=pet_id, name=name, species=species,
+                                        class_index=idx))
+
+    def get_pets_by_species(self, species: str) -> list[RegisteredPet]:
+        return [p for p in self._pets if p.species == species]
+
+    def identify(self, crop: np.ndarray, species: str) -> PetIDResult:
+        if not self.is_loaded or self._model is None:
+            return PetIDResult(pet_id=None, pet_name=None, confidence=0.0)
+
+        candidates = self.get_pets_by_species(species)
+        if not candidates:
+            return PetIDResult(pet_id=None, pet_name=None, confidence=0.0)
+
+        try:
+            import cv2
+            import torch
+            from torchvision import transforms
+
+            resized = cv2.resize(crop, (224, 224))
+            rgb = cv2.cvtColor(resized, cv2.COLOR_BGR2RGB)
+            transform = transforms.Compose([
+                transforms.ToTensor(),
+                transforms.Normalize(mean=[0.485, 0.456, 0.406],
+                                     std=[0.229, 0.224, 0.225]),
+            ])
+            tensor = transform(rgb).unsqueeze(0)
+
+            with torch.no_grad():
+                output = self._model(tensor)
+                probs = torch.softmax(output, dim=1)[0]
+
+            best_conf = 0.0
+            best_pet = None
+            for pet in candidates:
+                if pet.class_index < len(probs):
+                    conf = float(probs[pet.class_index])
+                    if conf > best_conf:
+                        best_conf = conf
+                        best_pet = pet
+
+            if best_pet and best_conf >= self._low_threshold:
+                return PetIDResult(
+                    pet_id=best_pet.pet_id,
+                    pet_name=best_pet.name,
+                    confidence=best_conf,
+                    high_threshold=self._high_threshold,
+                    low_threshold=self._low_threshold,
+                )
+
+            return PetIDResult(pet_id=None, pet_name=None, confidence=best_conf)
+
+        except Exception as e:
+            log.error("Pet ID inference failed: %s", e)
+            return PetIDResult(pet_id=None, pet_name=None, confidence=0.0)

vigilar/detection/trainer.py

@@ -0,0 +1,143 @@
+"""Pet ID model trainer using MobileNetV3-Small with transfer learning."""
+
+import logging
+import shutil
+from dataclasses import dataclass
+from pathlib import Path
+
+log = logging.getLogger(__name__)
+
+
+@dataclass
+class TrainingStatus:
+    is_training: bool = False
+    progress: float = 0.0
+    epoch: int = 0
+    total_epochs: int = 0
+    accuracy: float = 0.0
+    error: str | None = None
+
+
+class PetTrainer:
+    def __init__(self, training_dir: str, model_output_path: str):
+        self._training_dir = Path(training_dir)
+        self._model_output_path = Path(model_output_path)
+        self.status = TrainingStatus()
+
+    def get_class_names(self) -> list[str]:
+        if not self._training_dir.exists():
+            return []
+        return sorted([
+            d.name for d in self._training_dir.iterdir()
+            if d.is_dir() and not d.name.startswith(".")
+        ])
+
+    def check_readiness(self, min_images: int = 20) -> tuple[bool, str]:
+        class_names = self.get_class_names()
+        if not class_names:
+            return False, "No pet directories found in training directory."
+
+        insufficient = []
+        for name in class_names:
+            pet_dir = self._training_dir / name
+            image_count = sum(1 for f in pet_dir.iterdir()
+                              if f.suffix.lower() in (".jpg", ".jpeg", ".png"))
+            if image_count < min_images:
+                insufficient.append(f"{name}: {image_count}/{min_images}")
+
+        if insufficient:
+            return False, f"Insufficient training images: {', '.join(insufficient)}"
+
+        return True, f"Ready to train with {len(class_names)} classes."
+
+    def train(self, epochs: int = 30, batch_size: int = 16) -> bool:
+        try:
+            import torch
+            import torch.nn as nn
+            from torch.utils.data import DataLoader
+            from torchvision import datasets, models, transforms
+
+            self.status = TrainingStatus(is_training=True, total_epochs=epochs)
+            class_names = self.get_class_names()
+            num_classes = len(class_names)
+
+            if num_classes < 2:
+                self.status.error = "Need at least 2 pets to train."
+                self.status.is_training = False
+                return False
+
+            train_transform = transforms.Compose([
+                transforms.Resize((256, 256)),
+                transforms.RandomCrop(224),
+                transforms.RandomHorizontalFlip(),
+                transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2),
+                transforms.ToTensor(),
+                transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
+            ])
+
+            dataset = datasets.ImageFolder(str(self._training_dir), transform=train_transform)
+            loader = DataLoader(dataset, batch_size=batch_size, shuffle=True, num_workers=2)
+
+            model = models.mobilenet_v3_small(weights=models.MobileNet_V3_Small_Weights.DEFAULT)
+            model.classifier[-1] = nn.Linear(model.classifier[-1].in_features, num_classes)
+
+            device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+            model = model.to(device)
+
+            for param in model.features.parameters():
+                param.requires_grad = False
+
+            optimizer = torch.optim.Adam(model.classifier.parameters(), lr=1e-3)
+            criterion = nn.CrossEntropyLoss()
+
+            for epoch in range(epochs):
+                model.train()
+                running_loss = 0.0
+                correct = 0
+                total = 0
+
+                if epoch == 5:
+                    for param in model.features.parameters():
+                        param.requires_grad = True
+                    optimizer = torch.optim.Adam(model.parameters(), lr=1e-4)
+
+                for inputs, labels in loader:
+                    inputs, labels = inputs.to(device), labels.to(device)
+                    optimizer.zero_grad()
+                    outputs = model(inputs)
+                    loss = criterion(outputs, labels)
+                    loss.backward()
+                    optimizer.step()
+
+                    running_loss += loss.item()
+                    _, predicted = outputs.max(1)
+                    total += labels.size(0)
+                    correct += predicted.eq(labels).sum().item()
+
+                accuracy = correct / total if total > 0 else 0
+                self.status.epoch = epoch + 1
+                self.status.progress = (epoch + 1) / epochs
+                self.status.accuracy = accuracy
+                log.info("Epoch %d/%d — loss: %.4f, accuracy: %.4f",
+                         epoch + 1, epochs, running_loss / len(loader), accuracy)
+
+            if self._model_output_path.exists():
+                backup_path = self._model_output_path.with_name(
+                    self._model_output_path.stem + "_backup.pt"
+                )
+                shutil.copy2(self._model_output_path, backup_path)
+                log.info("Backed up previous model to %s", backup_path)
+
+            model = model.to("cpu")
+            torch.save(model, self._model_output_path)
+            log.info("Pet ID model saved to %s (accuracy: %.2f%%)",
+                     self._model_output_path, self.status.accuracy * 100)
+
+            self.status.is_training = False
+            return True
+
+        except Exception as e:
+            log.exception("Training failed: %s", e)
+            self.status.error = str(e)
+            self.status.is_training = False
+            return False

vigilar/detection/wildlife.py

@@ -0,0 +1,39 @@
+"""Wildlife threat level classification."""
+
+from vigilar.config import WildlifeConfig
+from vigilar.constants import ThreatLevel
+from vigilar.detection.person import Detection
+
+
+def classify_wildlife_threat(
+    detection: Detection,
+    config: WildlifeConfig,
+    frame_area: int,
+) -> tuple[ThreatLevel, str]:
+    """Classify a wildlife detection into threat level and species.
+
+    Returns (threat_level, species_name).
+    """
+    species = detection.class_name
+    threat_map = config.threat_map
+
+    # Direct COCO class mapping first
+    if species in threat_map.predator:
+        return ThreatLevel.PREDATOR, species
+    if species in threat_map.nuisance:
+        return ThreatLevel.NUISANCE, species
+    if species in threat_map.passive:
+        return ThreatLevel.PASSIVE, species
+
+    # Fallback to size heuristics for unknown species
+    _, _, w, h = detection.bbox
+    bbox_area = w * h
+    area_ratio = bbox_area / frame_area if frame_area > 0 else 0
+
+    heuristics = config.size_heuristics
+    if area_ratio < heuristics.small:
+        return ThreatLevel.NUISANCE, species
+    elif area_ratio < heuristics.medium:
+        return ThreatLevel.PREDATOR, species
+    else:
+        return ThreatLevel.PASSIVE, species

vigilar/detection/yolo.py

@@ -0,0 +1,78 @@
+"""Unified object detection using YOLOv8 via ultralytics."""
+
+import logging
+from pathlib import Path
+
+import numpy as np
+
+from vigilar.detection.person import Detection
+
+log = logging.getLogger(__name__)
+
+# COCO class names for domestic animals
+ANIMAL_CLASSES = {"cat", "dog"}
+
+# COCO class names for wildlife (subset that YOLO can detect)
+WILDLIFE_CLASSES = {"bear", "bird", "horse", "cow", "sheep", "elephant", "zebra", "giraffe"}
+
+# Vehicle class names from COCO
+VEHICLE_CLASSES = {"car", "motorcycle", "bus", "truck", "boat"}
+
+
+class YOLODetector:
+    def __init__(self, model_path: str, confidence_threshold: float = 0.5):
+        self._threshold = confidence_threshold
+        self._model = None
+        self.is_loaded = False
+
+        if Path(model_path).exists():
+            try:
+                from ultralytics import YOLO
+                self._model = YOLO(model_path)
+                self.is_loaded = True
+                log.info("YOLO model loaded from %s", model_path)
+            except Exception as e:
+                log.error("Failed to load YOLO model: %s", e)
+        else:
+            log.warning("YOLO model not found at %s — detection disabled", model_path)
+
+    def detect(self, frame: np.ndarray) -> list[Detection]:
+        if not self.is_loaded or self._model is None:
+            return []
+
+        results = self._model(frame, conf=self._threshold, verbose=False)
+        detections = []
+
+        for result in results:
+            for box in result.boxes:
+                class_id = int(box.cls[0])
+                confidence = float(box.conf[0])
+                class_name = result.names[class_id]
+
+                x1, y1, x2, y2 = box.xyxy[0].tolist()
+                x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
+                bw, bh = x2 - x1, y2 - y1
+                if bw <= 0 or bh <= 0:
+                    continue
+
+                detections.append(Detection(
+                    class_name=class_name,
+                    class_id=class_id,
+                    confidence=confidence,
+                    bbox=(x1, y1, bw, bh),
+                ))
+
+        return detections
+
+    @staticmethod
+    def classify(detection: Detection) -> str:
+        name = detection.class_name
+        if name == "person":
+            return "person"
+        if name in VEHICLE_CLASSES:
+            return "vehicle"
+        if name in ANIMAL_CLASSES:
+            return "domestic_animal"
+        if name in WILDLIFE_CLASSES:
+            return "wildlife"
+        return "other"

vigilar/events/processor.py

@@ -1,6 +1,5 @@
 """Event processor — subscribes to MQTT, logs events, evaluates rules."""
 
-import json
 import logging
 import signal
 import time
@@ -10,7 +9,7 @@ from sqlalchemy.engine import Engine
 
 from vigilar.bus import MessageBus
 from vigilar.config import VigilarConfig
-from vigilar.constants import ArmState, EventType, Severity, Topics
+from vigilar.constants import EventType, Severity, Topics
 from vigilar.events.rules import RuleEngine
 from vigilar.events.state import ArmStateFSM
 from vigilar.storage.db import get_db_path, init_db
@@ -102,6 +101,34 @@ class EventProcessor:
                 payload=payload,
             )
 
+            # Insert pet/wildlife sightings
+            if event_type in (
+                EventType.PET_DETECTED, EventType.PET_ESCAPE, EventType.UNKNOWN_ANIMAL
+            ):
+                from vigilar.storage.queries import insert_pet_sighting
+                insert_pet_sighting(
+                    engine,
+                    pet_id=payload.get("pet_id"),
+                    species=payload.get("species", "unknown"),
+                    camera_id=source_id or "",
+                    confidence=payload.get("confidence", 0.0),
+                    crop_path=payload.get("crop_path"),
+                    event_id=event_id,
+                )
+            elif event_type in (
+                EventType.WILDLIFE_PREDATOR, EventType.WILDLIFE_NUISANCE, EventType.WILDLIFE_PASSIVE
+            ):
+                from vigilar.storage.queries import insert_wildlife_sighting
+                insert_wildlife_sighting(
+                    engine,
+                    species=payload.get("species", "unknown"),
+                    threat_level=payload.get("threat_level", "PASSIVE"),
+                    camera_id=source_id or "",
+                    confidence=payload.get("confidence", 0.0),
+                    crop_path=payload.get("crop_path"),
+                    event_id=event_id,
+                )
+
             # Evaluate rules
             actions = rule_engine.evaluate(topic, payload, fsm.state)
 
@@ -127,6 +154,30 @@ class EventProcessor:
             if suffix in _TOPIC_EVENT_MAP:
                 etype, sev = _TOPIC_EVENT_MAP[suffix]
                 return etype, sev, camera_id
+
+            # Pet detection
+            if suffix == "pet/detected":
+                pet_id = payload.get("pet_id")
+                camera_location = payload.get("camera_location", "INTERIOR")
+
+                if not pet_id:
+                    return EventType.UNKNOWN_ANIMAL, Severity.WARNING, camera_id
+
+                if camera_location in ("EXTERIOR", "TRANSITION"):
+                    return EventType.PET_ESCAPE, Severity.ALERT, camera_id
+
+                return EventType.PET_DETECTED, Severity.INFO, camera_id
+
+            # Wildlife detection — severity depends on threat_level in payload
+            if suffix == "wildlife/detected":
+                threat = payload.get("threat_level", "PASSIVE")
+                if threat == "PREDATOR":
+                    return EventType.WILDLIFE_PREDATOR, Severity.CRITICAL, camera_id
+                elif threat == "NUISANCE":
+                    return EventType.WILDLIFE_NUISANCE, Severity.WARNING, camera_id
+                else:
+                    return EventType.WILDLIFE_PASSIVE, Severity.INFO, camera_id
+
             # Ignore heartbeats etc.
             return None, None, None
 

vigilar/health/digest.py

@@ -7,7 +7,7 @@ import time
 from sqlalchemy import func, select
 from sqlalchemy.engine import Engine
 
-from vigilar.storage.schema import events, recordings
+from vigilar.storage.schema import events, pet_sightings, recordings, wildlife_sightings
 
 log = logging.getLogger(__name__)
 
@@ -36,6 +36,31 @@ def build_digest(engine: Engine, data_dir: str, since_hours: int = 12) -> dict:
             .where(recordings.c.started_at >= since_ts // 1000)
         ).scalar() or 0
 
+        # pet_sightings/wildlife_sightings use float seconds (time.time()),
+        # while events table uses integer milliseconds (time.time() * 1000)
+        pet_count = conn.execute(
+            select(func.count()).select_from(pet_sightings)
+            .where(pet_sightings.c.ts >= since_ts / 1000)
+        ).scalar() or 0
+
+        wildlife_predator_count = conn.execute(
+            select(func.count()).select_from(wildlife_sightings)
+            .where(wildlife_sightings.c.ts >= since_ts / 1000,
+                   wildlife_sightings.c.threat_level == "PREDATOR")
+        ).scalar() or 0
+
+        wildlife_nuisance_count = conn.execute(
+            select(func.count()).select_from(wildlife_sightings)
+            .where(wildlife_sightings.c.ts >= since_ts / 1000,
+                   wildlife_sightings.c.threat_level == "NUISANCE")
+        ).scalar() or 0
+
+        wildlife_passive_count = conn.execute(
+            select(func.count()).select_from(wildlife_sightings)
+            .where(wildlife_sightings.c.ts >= since_ts / 1000,
+                   wildlife_sightings.c.threat_level == "PASSIVE")
+        ).scalar() or 0
+
     usage = shutil.disk_usage(data_dir)
     disk_pct = usage.used / usage.total * 100
     disk_gb = usage.used / (1024**3)
@@ -47,15 +72,31 @@ def build_digest(engine: Engine, data_dir: str, since_hours: int = 12) -> dict:
         "disk_used_gb": round(disk_gb, 1),
         "disk_used_pct": round(disk_pct, 0),
         "since_hours": since_hours,
+        "pet_sightings": pet_count,
+        "wildlife_predators": wildlife_predator_count,
+        "wildlife_nuisance": wildlife_nuisance_count,
+        "wildlife_passive": wildlife_passive_count,
     }
 
 
 def format_digest(data: dict) -> str:
-    return (
-        f"Vigilar Daily Summary\n"
+    lines = [
+        "Vigilar Daily Summary",
         f"Last {data['since_hours']}h: "
         f"{data['person_detections']} person detections, "
         f"{data['unknown_vehicles']} unknown vehicles, "
-        f"{data['recordings']} recordings\n"
-        f"Storage: {data['disk_used_gb']} GB ({data['disk_used_pct']:.0f}%)"
-    )
+        f"{data['recordings']} recordings",
+    ]
+    if data.get("pet_sightings", 0) > 0:
+        lines.append(f"Pets: {data['pet_sightings']} pet sightings")
+    wildlife_parts = []
+    if data.get("wildlife_predators", 0) > 0:
+        wildlife_parts.append(f"{data['wildlife_predators']} predator")
+    if data.get("wildlife_nuisance", 0) > 0:
+        wildlife_parts.append(f"{data['wildlife_nuisance']} nuisance")
+    if data.get("wildlife_passive", 0) > 0:
+        wildlife_parts.append(f"{data['wildlife_passive']} passive")
+    if wildlife_parts:
+        lines.append(f"Wildlife: {', '.join(wildlife_parts)}")
+    lines.append(f"Storage: {data['disk_used_gb']} GB ({data['disk_used_pct']:.0f}%)")
+    return "\n".join(lines)

vigilar/storage/queries.py

@@ -11,13 +11,16 @@ from vigilar.storage.schema import (
     alert_log,
     arm_state_log,
     events,
+    pet_sightings,
+    pet_training_images,
+    pets,
     push_subscriptions,
     recordings,
     sensor_states,
     system_events,
+    wildlife_sightings,
 )
 
-
 # --- Events ---
 
 def insert_event(
@@ -271,3 +274,186 @@ def delete_push_subscription(engine: Engine, endpoint: str) -> bool:
             push_subscriptions.delete().where(push_subscriptions.c.endpoint == endpoint)
         )
         return result.rowcount > 0
+
+
+# --- Pets ---
+
+def insert_pet(
+    engine: Engine,
+    name: str,
+    species: str,
+    breed: str | None = None,
+    color_description: str | None = None,
+    photo_path: str | None = None,
+) -> str:
+    import uuid
+    pet_id = str(uuid.uuid4())
+    with engine.begin() as conn:
+        conn.execute(pets.insert().values(
+            id=pet_id, name=name, species=species, breed=breed,
+            color_description=color_description, photo_path=photo_path,
+            training_count=0, created_at=time.time(),
+        ))
+    return pet_id
+
+
+def get_pet(engine: Engine, pet_id: str) -> dict[str, Any] | None:
+    with engine.connect() as conn:
+        row = conn.execute(pets.select().where(pets.c.id == pet_id)).first()
+        return dict(row._mapping) if row else None
+
+
+def get_all_pets(engine: Engine) -> list[dict[str, Any]]:
+    with engine.connect() as conn:
+        rows = conn.execute(pets.select().order_by(pets.c.name)).fetchall()
+        return [dict(r._mapping) for r in rows]
+
+
+# --- Pet Sightings ---
+
+def insert_pet_sighting(
+    engine: Engine,
+    species: str,
+    camera_id: str,
+    confidence: float,
+    pet_id: str | None = None,
+    crop_path: str | None = None,
+    event_id: int | None = None,
+) -> int:
+    with engine.begin() as conn:
+        result = conn.execute(pet_sightings.insert().values(
+            ts=time.time(), pet_id=pet_id, species=species,
+            camera_id=camera_id, confidence=confidence,
+            crop_path=crop_path, labeled=1 if pet_id else 0,
+            event_id=event_id,
+        ))
+        return result.inserted_primary_key[0]
+
+
+def get_pet_sightings(
+    engine: Engine,
+    pet_id: str | None = None,
+    camera_id: str | None = None,
+    since_ts: float | None = None,
+    limit: int = 100,
+) -> list[dict[str, Any]]:
+    query = select(pet_sightings).order_by(desc(pet_sightings.c.ts)).limit(limit)
+    if pet_id:
+        query = query.where(pet_sightings.c.pet_id == pet_id)
+    if camera_id:
+        query = query.where(pet_sightings.c.camera_id == camera_id)
+    if since_ts:
+        query = query.where(pet_sightings.c.ts >= since_ts)
+    with engine.connect() as conn:
+        rows = conn.execute(query).fetchall()
+        return [dict(r._mapping) for r in rows]
+
+
+def get_pet_last_location(engine: Engine, pet_id: str) -> dict[str, Any] | None:
+    with engine.connect() as conn:
+        row = conn.execute(
+            select(pet_sightings)
+            .where(pet_sightings.c.pet_id == pet_id)
+            .order_by(desc(pet_sightings.c.ts))
+            .limit(1)
+        ).first()
+        return dict(row._mapping) if row else None
+
+
+def get_unlabeled_sightings(
+    engine: Engine,
+    species: str | None = None,
+    limit: int = 50,
+) -> list[dict[str, Any]]:
+    query = (
+        select(pet_sightings)
+        .where(pet_sightings.c.labeled == 0)
+        .order_by(desc(pet_sightings.c.ts))
+        .limit(limit)
+    )
+    if species:
+        query = query.where(pet_sightings.c.species == species)
+    with engine.connect() as conn:
+        rows = conn.execute(query).fetchall()
+        return [dict(r._mapping) for r in rows]
+
+
+def label_sighting(engine: Engine, sighting_id: int, pet_id: str) -> None:
+    with engine.begin() as conn:
+        conn.execute(
+            pet_sightings.update()
+            .where(pet_sightings.c.id == sighting_id)
+            .values(pet_id=pet_id, labeled=1)
+        )
+
+
+# --- Wildlife Sightings ---
+
+def insert_wildlife_sighting(
+    engine: Engine,
+    species: str,
+    threat_level: str,
+    camera_id: str,
+    confidence: float,
+    crop_path: str | None = None,
+    event_id: int | None = None,
+) -> int:
+    with engine.begin() as conn:
+        result = conn.execute(wildlife_sightings.insert().values(
+            ts=time.time(), species=species, threat_level=threat_level,
+            camera_id=camera_id, confidence=confidence,
+            crop_path=crop_path, event_id=event_id,
+        ))
+        return result.inserted_primary_key[0]
+
+
+def get_wildlife_sightings(
+    engine: Engine,
+    threat_level: str | None = None,
+    camera_id: str | None = None,
+    since_ts: float | None = None,
+    limit: int = 100,
+) -> list[dict[str, Any]]:
+    query = select(wildlife_sightings).order_by(desc(wildlife_sightings.c.ts)).limit(limit)
+    if threat_level:
+        query = query.where(wildlife_sightings.c.threat_level == threat_level)
+    if camera_id:
+        query = query.where(wildlife_sightings.c.camera_id == camera_id)
+    if since_ts:
+        query = query.where(wildlife_sightings.c.ts >= since_ts)
+    with engine.connect() as conn:
+        rows = conn.execute(query).fetchall()
+        return [dict(r._mapping) for r in rows]
+
+
+# --- Training Images ---
+
+def insert_training_image(
+    engine: Engine,
+    pet_id: str,
+    image_path: str,
+    source: str,
+) -> int:
+    with engine.begin() as conn:
+        result = conn.execute(pet_training_images.insert().values(
+            pet_id=pet_id, image_path=image_path,
+            source=source, created_at=time.time(),
+        ))
+        conn.execute(
+            pets.update().where(pets.c.id == pet_id)
+            .values(training_count=pets.c.training_count + 1)
+        )
+        return result.inserted_primary_key[0]
+
+
+def get_training_images(
+    engine: Engine,
+    pet_id: str,
+) -> list[dict[str, Any]]:
+    with engine.connect() as conn:
+        rows = conn.execute(
+            select(pet_training_images)
+            .where(pet_training_images.c.pet_id == pet_id)
+            .order_by(desc(pet_training_images.c.created_at))
+        ).fetchall()
+        return [dict(r._mapping) for r in rows]

vigilar/storage/schema.py

@@ -123,3 +123,58 @@ push_subscriptions = Table(
     Column("last_used_at", Integer),
     Column("user_agent", String),
 )
+
+pets = Table(
+    "pets",
+    metadata,
+    Column("id", String, primary_key=True),
+    Column("name", String, nullable=False),
+    Column("species", String, nullable=False),
+    Column("breed", String),
+    Column("color_description", String),
+    Column("photo_path", String),
+    Column("training_count", Integer, nullable=False, default=0),
+    Column("created_at", Float, nullable=False),
+)
+
+pet_sightings = Table(
+    "pet_sightings",
+    metadata,
+    Column("id", Integer, primary_key=True, autoincrement=True),
+    Column("ts", Float, nullable=False),
+    Column("pet_id", String),
+    Column("species", String, nullable=False),
+    Column("camera_id", String, nullable=False),
+    Column("confidence", Float),
+    Column("crop_path", String),
+    Column("labeled", Integer, nullable=False, default=0),
+    Column("event_id", Integer),
+)
+Index("idx_pet_sightings_ts", pet_sightings.c.ts.desc())
+Index("idx_pet_sightings_pet", pet_sightings.c.pet_id, pet_sightings.c.ts.desc())
+Index("idx_pet_sightings_camera", pet_sightings.c.camera_id, pet_sightings.c.ts.desc())
+
+wildlife_sightings = Table(
+    "wildlife_sightings",
+    metadata,
+    Column("id", Integer, primary_key=True, autoincrement=True),
+    Column("ts", Float, nullable=False),
+    Column("species", String, nullable=False),
+    Column("threat_level", String, nullable=False),
+    Column("camera_id", String, nullable=False),
+    Column("confidence", Float),
+    Column("crop_path", String),
+    Column("event_id", Integer),
+)
+Index("idx_wildlife_ts", wildlife_sightings.c.ts.desc())
+Index("idx_wildlife_threat", wildlife_sightings.c.threat_level, wildlife_sightings.c.ts.desc())
+
+pet_training_images = Table(
+    "pet_training_images",
+    metadata,
+    Column("id", Integer, primary_key=True, autoincrement=True),
+    Column("pet_id", String, nullable=False),
+    Column("image_path", String, nullable=False),
+    Column("source", String, nullable=False),
+    Column("created_at", Float, nullable=False),
+)

vigilar/web/app.py

@@ -26,6 +26,7 @@ def create_app(cfg: VigilarConfig | None = None) -> Flask:
     from vigilar.web.blueprints.cameras import cameras_bp
     from vigilar.web.blueprints.events import events_bp
     from vigilar.web.blueprints.kiosk import kiosk_bp
+    from vigilar.web.blueprints.pets import pets_bp
     from vigilar.web.blueprints.recordings import recordings_bp
     from vigilar.web.blueprints.sensors import sensors_bp
     from vigilar.web.blueprints.system import system_bp
@@ -33,6 +34,7 @@ def create_app(cfg: VigilarConfig | None = None) -> Flask:
     app.register_blueprint(cameras_bp)
     app.register_blueprint(events_bp)
     app.register_blueprint(kiosk_bp)
+    app.register_blueprint(pets_bp)
     app.register_blueprint(recordings_bp)
     app.register_blueprint(sensors_bp)
     app.register_blueprint(system_bp)

vigilar/web/blueprints/pets.py

@@ -0,0 +1,307 @@
+"""Pets blueprint — pet registration, sightings, labeling, training."""
+
+import os
+import time
+from pathlib import Path
+
+from flask import Blueprint, current_app, jsonify, render_template, request
+from werkzeug.utils import secure_filename
+
+pets_bp = Blueprint("pets", __name__, url_prefix="/pets")
+
+ALLOWED_IMAGE_EXTENSIONS = {".jpg", ".jpeg", ".png", ".webp"}
+
+
+def _engine():
+    return current_app.config.get("DB_ENGINE")
+
+
+@pets_bp.route("/")
+def dashboard():
+    engine = _engine()
+    all_pets = []
+    if engine:
+        from vigilar.storage.queries import get_all_pets
+        all_pets = get_all_pets(engine)
+    return render_template("pets/dashboard.html", pets=all_pets)
+
+
+@pets_bp.route("/register", methods=["POST"])
+def register_pet():
+    data = request.get_json(silent=True) or {}
+    name = data.get("name")
+    species = data.get("species")
+    if not name or not species:
+        return jsonify({"error": "name and species required"}), 400
+
+    engine = _engine()
+    if engine is None:
+        return jsonify({"error": "database not available"}), 503
+
+    from vigilar.storage.queries import insert_pet
+    pet_id = insert_pet(
+        engine,
+        name=name,
+        species=species,
+        breed=data.get("breed"),
+        color_description=data.get("color_description"),
+        photo_path=data.get("photo_path"),
+    )
+    return jsonify({"id": pet_id, "name": name, "species": species})
+
+
+@pets_bp.route("/api/status")
+def pet_status():
+    engine = _engine()
+    if engine is None:
+        return jsonify({"pets": []})
+
+    from vigilar.storage.queries import get_all_pets, get_pet_last_location
+    pets_list = get_all_pets(engine)
+    now = time.time()
+    result = []
+    for pet in pets_list:
+        last = get_pet_last_location(engine, pet["id"])
+        if last:
+            last_seen_ago = now - last["ts"]
+            status = "safe" if last_seen_ago < 300 else "unknown"  # seen in last 5 min
+        else:
+            status = "unknown"
+        result.append({
+            **pet,
+            "last_seen_ts": last["ts"] if last else None,
+            "last_camera": last["camera_id"] if last else None,
+            "status": status,
+        })
+    return jsonify({"pets": result})
+
+
+@pets_bp.route("/api/sightings")
+def pet_sightings():
+    engine = _engine()
+    if engine is None:
+        return jsonify({"sightings": []})
+
+    pet_id = request.args.get("pet_id")
+    camera_id = request.args.get("camera_id")
+    since_ts = request.args.get("since", type=float)
+    limit = min(request.args.get("limit", 100, type=int), 500)
+
+    from vigilar.storage.queries import get_pet_sightings
+    sightings = get_pet_sightings(engine, pet_id=pet_id, camera_id=camera_id,
+                                   since_ts=since_ts, limit=limit)
+    return jsonify({"sightings": sightings})
+
+
+@pets_bp.route("/api/wildlife")
+def wildlife_sightings():
+    engine = _engine()
+    if engine is None:
+        return jsonify({"sightings": []})
+
+    threat_level = request.args.get("threat_level")
+    camera_id = request.args.get("camera_id")
+    since_ts = request.args.get("since", type=float)
+    limit = min(request.args.get("limit", 100, type=int), 500)
+
+    from vigilar.storage.queries import get_wildlife_sightings
+    sightings = get_wildlife_sightings(engine, threat_level=threat_level, camera_id=camera_id,
+                                        since_ts=since_ts, limit=limit)
+    return jsonify({"sightings": sightings})
+
+
+@pets_bp.route("/api/unlabeled")
+def unlabeled_crops():
+    engine = _engine()
+    if engine is None:
+        return jsonify({"crops": []})
+
+    species = request.args.get("species")
+    limit = request.args.get("limit", 50, type=int)
+
+    from vigilar.storage.queries import get_unlabeled_sightings
+    crops = get_unlabeled_sightings(engine, species=species, limit=limit)
+    return jsonify({"crops": crops})
+
+
+@pets_bp.route("/<sighting_id>/label", methods=["POST"])
+def label_crop(sighting_id: str):
+    data = request.get_json(silent=True) or {}
+    pet_id = data.get("pet_id")
+    if not pet_id:
+        return jsonify({"error": "pet_id required"}), 400
+
+    engine = _engine()
+    if engine is None:
+        return jsonify({"error": "database not available"}), 503
+
+    from vigilar.storage.queries import label_sighting
+    try:
+        label_sighting(engine, int(sighting_id), pet_id)
+    except (ValueError, TypeError):
+        return jsonify({"error": "invalid sighting_id"}), 400
+    return jsonify({"ok": True})
+
+
+@pets_bp.route("/<pet_id>/upload", methods=["POST"])
+def upload_photo(pet_id: str):
+    engine = _engine()
+    if engine is None:
+        return jsonify({"error": "database not available"}), 503
+
+    from vigilar.storage.queries import get_pet, insert_training_image
+    pet = get_pet(engine, pet_id)
+    if not pet:
+        return jsonify({"error": "Pet not found"}), 404
+
+    files = request.files.getlist("photos")
+    if not files:
+        return jsonify({"error": "no files provided"}), 400
+
+    cfg = current_app.config.get("VIGILAR_CONFIG")
+    pets_cfg = current_app.config.get("PETS_CONFIG")
+    if pets_cfg:
+        training_dir = pets_cfg.training_dir
+    else:
+        training_dir = os.path.join(cfg.system.data_dir if cfg else "/tmp/vigilar", "training")
+
+    pet_dir = Path(training_dir) / pet["name"].lower()
+    pet_dir.mkdir(parents=True, exist_ok=True)
+
+    saved = []
+    for f in files:
+        if not f.filename:
+            continue
+        safe_name = secure_filename(f.filename)
+        if not safe_name:
+            continue
+        if Path(safe_name).suffix.lower() not in ALLOWED_IMAGE_EXTENSIONS:
+            continue
+        dest = pet_dir / f"{int(time.time() * 1000)}_{safe_name}"
+        f.save(str(dest))
+        insert_training_image(engine, pet_id=pet_id, image_path=str(dest), source="upload")
+        saved.append(str(dest))
+
+    return jsonify({"ok": True, "saved": len(saved)})
+
+
+@pets_bp.route("/<pet_id>/update", methods=["POST"])
+def update_pet(pet_id: str):
+    engine = _engine()
+    if engine is None:
+        return jsonify({"error": "database not available"}), 503
+
+    from vigilar.storage.queries import get_pet
+    pet = get_pet(engine, pet_id)
+    if not pet:
+        return jsonify({"error": "Pet not found"}), 404
+
+    data = request.get_json(silent=True) or {}
+    from sqlalchemy import update as sa_update
+
+    from vigilar.storage.schema import pets as pets_table
+
+    allowed = {"name", "species", "breed", "color_description", "photo_path"}
+    values = {k: v for k, v in data.items() if k in allowed}
+    if not values:
+        return jsonify({"error": "no valid fields to update"}), 400
+
+    with engine.begin() as conn:
+        conn.execute(sa_update(pets_table).where(pets_table.c.id == pet_id).values(**values))
+    return jsonify({"ok": True})
+
+
+@pets_bp.route("/<pet_id>/delete", methods=["DELETE"])
+def delete_pet(pet_id: str):
+    engine = _engine()
+    if engine is None:
+        return jsonify({"error": "database not available"}), 503
+
+    from vigilar.storage.schema import pets as pets_table
+    with engine.begin() as conn:
+        conn.execute(pets_table.delete().where(pets_table.c.id == pet_id))
+    return jsonify({"ok": True})
+
+
+@pets_bp.route("/train", methods=["POST"])
+def train_model():
+    pets_cfg = current_app.config.get("PETS_CONFIG")
+    if not pets_cfg:
+        return jsonify({"error": "Pet detection not configured"}), 400
+
+    from vigilar.detection.trainer import PetTrainer
+    trainer = PetTrainer(
+        training_dir=pets_cfg.training_dir,
+        model_output_path=pets_cfg.pet_id_model_path,
+    )
+
+    ready, msg = trainer.check_readiness(min_images=pets_cfg.min_training_images)
+    if not ready:
+        return jsonify({"error": msg}), 400
+
+    # Store trainer on app for status polling
+    current_app.extensions["pet_trainer"] = trainer
+
+    import threading
+    thread = threading.Thread(target=trainer.train, daemon=True)
+    thread.start()
+
+    return jsonify({"ok": True, "status": "training_started", "message": msg})
+
+
+@pets_bp.route("/api/training-status")
+def training_status():
+    trainer = current_app.extensions.get("pet_trainer")
+    if not trainer:
+        return jsonify({"status": "idle"})
+    s = trainer.status
+    return jsonify({
+        "status": "training" if s.is_training else "complete" if s.progress >= 1.0 else "idle",
+        "progress": round(s.progress, 2),
+        "epoch": s.epoch,
+        "total_epochs": s.total_epochs,
+        "accuracy": round(s.accuracy, 4),
+        "error": s.error,
+    })
+
+
+@pets_bp.route("/api/highlights")
+def highlights():
+    engine = _engine()
+    if not engine:
+        return jsonify({"highlights": []})
+
+    import time as time_mod
+    since = time_mod.time() - 86400
+
+    from vigilar.storage.queries import get_pet_sightings, get_wildlife_sightings
+
+    result = []
+
+    # High-confidence pet sightings
+    sightings = get_pet_sightings(engine, since_ts=since, limit=50)
+    for s in sightings:
+        if s.get("confidence", 0) > 0.8 and s.get("pet_id"):
+            result.append({
+                "type": "pet_sighting",
+                "pet_id": s.get("pet_id"),
+                "camera_id": s["camera_id"],
+                "timestamp": s["ts"],
+                "confidence": s.get("confidence"),
+            })
+
+    # Wildlife events
+    wildlife = get_wildlife_sightings(engine, since_ts=since, limit=20)
+    for w in wildlife:
+        result.append({
+            "type": "wildlife",
+            "species": w["species"],
+            "threat_level": w["threat_level"],
+            "camera_id": w["camera_id"],
+            "timestamp": w["ts"],
+            "confidence": w.get("confidence"),
+        })
+
+    # Sort by timestamp descending, limit to 20
+    result.sort(key=lambda x: x["timestamp"], reverse=True)
+    return jsonify({"highlights": result[:20]})

vigilar/web/static/css/pet-labeling.css

@@ -0,0 +1,205 @@
+/* Vigilar — Pet labeling overlay for recording playback */
+
+/* Container: sits on top of the video element, sized to match it */
+.detection-overlay {
+    position: absolute;
+    top: 0;
+    left: 0;
+    width: 100%;
+    height: 100%;
+    pointer-events: none; /* pass-through by default; bboxes re-enable */
+    z-index: 10;
+}
+
+/* Individual bounding box */
+.detection-bbox {
+    position: absolute;
+    pointer-events: auto;
+    cursor: pointer;
+    box-sizing: border-box;
+    border: 2px solid #00d4d4; /* cyan default for animals */
+    border-radius: 3px;
+    transition: border-color 0.15s ease, box-shadow 0.15s ease;
+}
+
+.detection-bbox:hover {
+    border-color: #00ffff;
+    box-shadow: 0 0 0 2px rgba(0, 255, 255, 0.25);
+}
+
+.detection-bbox.bbox-person {
+    border-color: #dc3545;
+}
+
+.detection-bbox.bbox-person:hover {
+    border-color: #ff4d5e;
+    box-shadow: 0 0 0 2px rgba(220, 53, 69, 0.25);
+}
+
+.detection-bbox.bbox-vehicle {
+    border-color: #0d6efd;
+}
+
+.detection-bbox.bbox-vehicle:hover {
+    border-color: #3d8bfd;
+    box-shadow: 0 0 0 2px rgba(13, 110, 253, 0.25);
+}
+
+.detection-bbox.bbox-labeled {
+    border-style: solid;
+    opacity: 0.85;
+}
+
+/* Species + confidence tag shown on top-left corner of bbox */
+.detection-bbox-label {
+    position: absolute;
+    top: -1px;
+    left: -1px;
+    padding: 1px 6px;
+    font-size: 0.7rem;
+    font-weight: 600;
+    line-height: 1.4;
+    background-color: #00d4d4;
+    color: #0d1117;
+    border-radius: 2px 0 2px 0;
+    white-space: nowrap;
+    pointer-events: none;
+    user-select: none;
+    transition: background-color 0.15s ease;
+}
+
+.detection-bbox.bbox-person .detection-bbox-label {
+    background-color: #dc3545;
+    color: #fff;
+}
+
+.detection-bbox.bbox-vehicle .detection-bbox-label {
+    background-color: #0d6efd;
+    color: #fff;
+}
+
+/* "Who is this?" popup card */
+.label-popup {
+    position: fixed; /* positioned by JS via clientX/clientY */
+    z-index: 9999;
+    min-width: 180px;
+    max-width: 240px;
+    background-color: #161b22;
+    border: 1px solid #30363d;
+    border-radius: 8px;
+    box-shadow: 0 8px 24px rgba(0, 0, 0, 0.6);
+    padding: 10px;
+    animation: popup-appear 0.12s ease-out;
+    pointer-events: auto;
+}
+
+@keyframes popup-appear {
+    from {
+        opacity: 0;
+        transform: scale(0.92) translateY(-4px);
+    }
+    to {
+        opacity: 1;
+        transform: scale(1) translateY(0);
+    }
+}
+
+.label-popup-title {
+    font-size: 0.75rem;
+    font-weight: 600;
+    color: #8b949e;
+    text-transform: uppercase;
+    letter-spacing: 0.05em;
+    margin-bottom: 8px;
+    padding-bottom: 6px;
+    border-bottom: 1px solid #30363d;
+}
+
+.label-popup-pets {
+    display: flex;
+    flex-direction: column;
+    gap: 4px;
+    margin-bottom: 6px;
+}
+
+.label-popup-pet-btn {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+    width: 100%;
+    text-align: left;
+    padding: 5px 8px;
+    border: 1px solid #30363d;
+    border-radius: 6px;
+    background: transparent;
+    color: #e6edf3;
+    font-size: 0.85rem;
+    cursor: pointer;
+    transition: background-color 0.1s ease, border-color 0.1s ease;
+}
+
+.label-popup-pet-btn:hover {
+    background-color: #21262d;
+    border-color: #58a6ff;
+}
+
+.label-popup-pet-btn .pet-species-dot {
+    width: 8px;
+    height: 8px;
+    border-radius: 50%;
+    background-color: #00d4d4;
+    flex-shrink: 0;
+}
+
+.label-popup-divider {
+    border: none;
+    border-top: 1px solid #30363d;
+    margin: 6px 0;
+}
+
+.label-popup-action-btn {
+    display: flex;
+    align-items: center;
+    gap: 6px;
+    width: 100%;
+    text-align: left;
+    padding: 5px 8px;
+    border: 1px dashed #30363d;
+    border-radius: 6px;
+    background: transparent;
+    color: #8b949e;
+    font-size: 0.8rem;
+    cursor: pointer;
+    transition: color 0.1s ease, border-color 0.1s ease;
+}
+
+.label-popup-action-btn:hover {
+    color: #e6edf3;
+    border-color: #8b949e;
+}
+
+/* Spinner shown while fetching pets */
+.label-popup-loading {
+    text-align: center;
+    color: #8b949e;
+    font-size: 0.8rem;
+    padding: 8px 0;
+}
+
+/* Confirmation flash on successful label */
+.label-confirm-flash {
+    position: absolute;
+    top: 0;
+    left: 0;
+    right: 0;
+    bottom: 0;
+    border-radius: 3px;
+    background-color: rgba(0, 212, 212, 0.2);
+    pointer-events: none;
+    animation: confirm-flash 0.5s ease-out forwards;
+}
+
+@keyframes confirm-flash {
+    0% { opacity: 1; }
+    100% { opacity: 0; }
+}

vigilar/web/static/js/pet-labeling.js

@@ -0,0 +1,377 @@
+/* Vigilar — Pet labeling overlay for recording playback
+ *
+ * Integration:
+ *   1. Add to the page:
+ *        <link rel="stylesheet" href="{{ url_for('static', filename='css/pet-labeling.css') }}">
+ *        <script src="{{ url_for('static', filename='js/pet-labeling.js') }}"></script>
+ *
+ *   2. Wrap the <video> element in a position:relative container and add
+ *      `data-pet-labeling="true"` to that container, along with a
+ *      `data-detections` attribute holding the JSON array of detections:
+ *
+ *        <div class="position-relative" data-pet-labeling="true"
+ *             data-detections='[{"id":1,"species":"cat","confidence":0.92,
+ *                                "bbox":[0.1,0.15,0.35,0.6]}]'>
+ *          <video id="player-video" class="w-100" controls></video>
+ *        </div>
+ *
+ *      bbox values are fractional [x1, y1, x2, y2] relative to video dimensions
+ *      (same convention as YOLO normalised coords).
+ *
+ *   3. The module self-initialises on DOMContentLoaded and also exposes
+ *      window.PetLabeling for manual calls:
+ *
+ *        PetLabeling.init(containerEl, detectionsArray);
+ *        PetLabeling.updateDetections(containerEl, detectionsArray);
+ */
+
+(function () {
+    'use strict';
+
+    // -------------------------------------------------------------------------
+    // Constants
+    // -------------------------------------------------------------------------
+
+    const ANIMAL_SPECIES = new Set(['cat', 'dog', 'bird', 'rabbit', 'hamster', 'fish', 'animal']);
+
+    // Map species to bbox CSS class (animals get default cyan)
+    function bboxClass(species) {
+        if (species === 'person') return 'bbox-person';
+        if (species === 'vehicle' || species === 'car' || species === 'truck') return 'bbox-vehicle';
+        return ''; // cyan default (animals / unknown)
+    }
+
+    // -------------------------------------------------------------------------
+    // Pets cache — fetched once per page load, filtered by species on demand
+    // -------------------------------------------------------------------------
+
+    let _petsCache = null; // null = not fetched yet; [] = fetched, none found
+
+    async function fetchPets() {
+        if (_petsCache !== null) return _petsCache;
+        try {
+            const resp = await fetch('/pets/api/status');
+            if (!resp.ok) throw new Error(`HTTP ${resp.status}`);
+            const data = await resp.json();
+            _petsCache = Array.isArray(data.pets) ? data.pets : [];
+        } catch (e) {
+            _petsCache = [];
+        }
+        return _petsCache;
+    }
+
+    function petsForSpecies(pets, species) {
+        if (!species) return pets;
+        // For generic "animal" detections show all pets; otherwise filter by species
+        if (ANIMAL_SPECIES.has(species) && species !== 'animal') {
+            return pets.filter(p => p.species && p.species.toLowerCase() === species.toLowerCase());
+        }
+        return pets;
+    }
+
+    // -------------------------------------------------------------------------
+    // Active popup — only one at a time
+    // -------------------------------------------------------------------------
+
+    let _activePopup = null;
+
+    function dismissActivePopup() {
+        if (_activePopup) {
+            _activePopup.remove();
+            _activePopup = null;
+        }
+    }
+
+    // -------------------------------------------------------------------------
+    // Label popup
+    // -------------------------------------------------------------------------
+
+    async function showLabelPopup(bbox, detection, anchorEl) {
+        dismissActivePopup();
+
+        const popup = document.createElement('div');
+        popup.className = 'label-popup';
+        popup.innerHTML = `
+            <div class="label-popup-title">
+                <i class="bi bi-tag-fill me-1"></i>Who is this?
+            </div>
+            <div class="label-popup-loading">
+                <span class="spinner-border spinner-border-sm me-1" role="status"></span>
+                Loading…
+            </div>`;
+        document.body.appendChild(popup);
+        _activePopup = popup;
+
+        // Position near the clicked bbox, keep on screen
+        positionPopup(popup, anchorEl);
+
+        // Fetch pets (cached after first call)
+        const allPets = await fetchPets();
+        if (_activePopup !== popup) return; // dismissed while loading
+
+        const pets = petsForSpecies(allPets, detection.species);
+
+        let petsHtml = '';
+        if (pets.length > 0) {
+            petsHtml = pets.map(p => `
+                <button class="label-popup-pet-btn" data-pet-id="${p.id}" data-pet-name="${escHtml(p.name)}">
+                    <span class="pet-species-dot"></span>
+                    ${escHtml(p.name)}
+                    <small class="ms-auto text-muted">${escHtml(p.species || '')}</small>
+                </button>`).join('');
+        }
+
+        popup.innerHTML = `
+            <div class="label-popup-title">
+                <i class="bi bi-tag-fill me-1"></i>Who is this?
+                <small class="float-end text-muted text-capitalize">${escHtml(detection.species || 'animal')}</small>
+            </div>
+            <div class="label-popup-pets">
+                ${petsHtml}
+            </div>
+            ${pets.length > 0 ? '<hr class="label-popup-divider">' : ''}
+            <button class="label-popup-action-btn" data-action="unknown">
+                <i class="bi bi-question-circle"></i>Unknown
+            </button>
+            <button class="label-popup-action-btn mt-1" data-action="new-pet">
+                <i class="bi bi-plus-circle"></i>+ New Pet
+            </button>`;
+
+        // Re-position after content change (height may differ)
+        positionPopup(popup, anchorEl);
+
+        // Pet selection
+        popup.querySelectorAll('.label-popup-pet-btn').forEach(btn => {
+            btn.addEventListener('click', (e) => {
+                e.stopPropagation();
+                const petId = btn.dataset.petId;
+                const petName = btn.dataset.petName;
+                applyLabel(detection, petId, petName, bbox);
+                dismissActivePopup();
+            });
+        });
+
+        // Unknown
+        const unknownBtn = popup.querySelector('[data-action="unknown"]');
+        if (unknownBtn) {
+            unknownBtn.addEventListener('click', (e) => {
+                e.stopPropagation();
+                applyLabel(detection, 'unknown', 'Unknown', bbox);
+                dismissActivePopup();
+            });
+        }
+
+        // New Pet — open the pets dashboard in a new tab
+        const newPetBtn = popup.querySelector('[data-action="new-pet"]');
+        if (newPetBtn) {
+            newPetBtn.addEventListener('click', (e) => {
+                e.stopPropagation();
+                window.open('/pets/', '_blank');
+                dismissActivePopup();
+            });
+        }
+    }
+
+    function positionPopup(popup, anchorEl) {
+        const rect = anchorEl.getBoundingClientRect();
+        const popW = popup.offsetWidth || 200;
+        const popH = popup.offsetHeight || 160;
+        const margin = 8;
+
+        // Prefer right of bbox; fall back to left
+        let left = rect.right + margin;
+        if (left + popW > window.innerWidth - margin) {
+            left = rect.left - popW - margin;
+        }
+        if (left < margin) left = margin;
+
+        // Align top with bbox; clamp to viewport
+        let top = rect.top;
+        if (top + popH > window.innerHeight - margin) {
+            top = window.innerHeight - popH - margin;
+        }
+        if (top < margin) top = margin;
+
+        popup.style.left = `${left}px`;
+        popup.style.top = `${top}px`;
+    }
+
+    // -------------------------------------------------------------------------
+    // Label API call
+    // -------------------------------------------------------------------------
+
+    async function applyLabel(detection, petId, petName, bboxEl) {
+        // Optimistic UI: update the label tag immediately
+        const labelTag = bboxEl.querySelector('.detection-bbox-label');
+        if (labelTag) {
+            labelTag.textContent = petId === 'unknown' ? 'Unknown' : petName;
+        }
+        bboxEl.classList.add('bbox-labeled');
+
+        // Confirmation flash
+        const flash = document.createElement('div');
+        flash.className = 'label-confirm-flash';
+        bboxEl.appendChild(flash);
+        flash.addEventListener('animationend', () => flash.remove());
+
+        if (!detection.id || petId === 'unknown') return;
+
+        try {
+            await fetch(`/pets/${detection.id}/label`, {
+                method: 'POST',
+                headers: { 'Content-Type': 'application/json' },
+                body: JSON.stringify({ pet_id: petId }),
+            });
+        } catch (e) {
+            // Silently fail — offline mode
+        }
+    }
+
+    // -------------------------------------------------------------------------
+    // Overlay rendering
+    // -------------------------------------------------------------------------
+
+    function renderOverlay(container, detections) {
+        // Remove any existing overlay
+        const existing = container.querySelector('.detection-overlay');
+        if (existing) existing.remove();
+
+        if (!detections || detections.length === 0) return;
+
+        const video = container.querySelector('video');
+        if (!video) return;
+
+        const overlay = document.createElement('div');
+        overlay.className = 'detection-overlay';
+        container.appendChild(overlay);
+
+        // Draw bboxes sized to the video's rendered dimensions
+        function drawBboxes() {
+            overlay.innerHTML = '';
+
+            const vRect = video.getBoundingClientRect();
+            const cRect = container.getBoundingClientRect();
+
+            // Offset from container origin
+            const offX = vRect.left - cRect.left;
+            const offY = vRect.top - cRect.top;
+            const vW = vRect.width;
+            const vH = vRect.height;
+
+            if (vW === 0 || vH === 0) return;
+
+            detections.forEach(det => {
+                const [x1f, y1f, x2f, y2f] = det.bbox || [0, 0, 1, 1];
+
+                const left = offX + x1f * vW;
+                const top = offY + y1f * vH;
+                const width = (x2f - x1f) * vW;
+                const height = (y2f - y1f) * vH;
+
+                const bbox = document.createElement('div');
+                bbox.className = `detection-bbox ${bboxClass(det.species)}`.trimEnd();
+                bbox.style.cssText = `
+                    left: ${left}px;
+                    top: ${top}px;
+                    width: ${width}px;
+                    height: ${height}px;
+                `;
+
+                const confidence = det.confidence !== undefined
+                    ? ` ${Math.round(det.confidence * 100)}%`
+                    : '';
+                const labelText = det.pet_name
+                    ? det.pet_name
+                    : `${det.species || 'animal'}${confidence}`;
+
+                const label = document.createElement('div');
+                label.className = 'detection-bbox-label';
+                label.textContent = labelText;
+                bbox.appendChild(label);
+
+                // Only show "Who is this?" popup for animal detections
+                if (!det.species || ANIMAL_SPECIES.has(det.species.toLowerCase())) {
+                    bbox.addEventListener('click', (e) => {
+                        e.stopPropagation();
+                        showLabelPopup(bbox, det, bbox);
+                    });
+                }
+
+                overlay.appendChild(bbox);
+            });
+        }
+
+        // Draw immediately and on video resize/metadata
+        drawBboxes();
+        video.addEventListener('loadedmetadata', drawBboxes);
+        const ro = new ResizeObserver(drawBboxes);
+        ro.observe(video);
+
+        // Store cleanup handle
+        overlay._roCleanup = () => ro.disconnect();
+    }
+
+    // -------------------------------------------------------------------------
+    // Public API
+    // -------------------------------------------------------------------------
+
+    function init(container, detections) {
+        if (!container) return;
+        renderOverlay(container, detections);
+    }
+
+    function updateDetections(container, detections) {
+        if (!container) return;
+        const old = container.querySelector('.detection-overlay');
+        if (old && old._roCleanup) old._roCleanup();
+        renderOverlay(container, detections);
+    }
+
+    // -------------------------------------------------------------------------
+    // Auto-init: find all [data-pet-labeling="true"] containers on page load
+    // -------------------------------------------------------------------------
+
+    function autoInit() {
+        document.querySelectorAll('[data-pet-labeling="true"]').forEach(container => {
+            let detections = [];
+            const raw = container.dataset.detections;
+            if (raw) {
+                try { detections = JSON.parse(raw); } catch (e) { detections = []; }
+            }
+            init(container, detections);
+        });
+    }
+
+    if (document.readyState === 'loading') {
+        document.addEventListener('DOMContentLoaded', autoInit);
+    } else {
+        autoInit();
+    }
+
+    // Dismiss popup when clicking outside
+    document.addEventListener('click', (e) => {
+        if (_activePopup && !_activePopup.contains(e.target)) {
+            dismissActivePopup();
+        }
+    });
+
+    document.addEventListener('keydown', (e) => {
+        if (e.key === 'Escape') dismissActivePopup();
+    });
+
+    // -------------------------------------------------------------------------
+    // Helpers
+    // -------------------------------------------------------------------------
+
+    function escHtml(str) {
+        if (!str) return '';
+        return String(str)
+            .replace(/&/g, '&amp;')
+            .replace(/</g, '&lt;')
+            .replace(/>/g, '&gt;')
+            .replace(/"/g, '&quot;');
+    }
+
+    // Expose public API
+    window.PetLabeling = { init, updateDetections };
+})();

vigilar/web/templates/pets/dashboard.html

@@ -0,0 +1,513 @@
+{% extends "base.html" %}
+{% block title %}Pets — Vigilar{% endblock %}
+
+{% block content %}
+<div class="d-flex justify-content-between align-items-center mb-3">
+    <h5 class="mb-0"><i class="bi bi-heart-fill me-2 text-danger"></i>Pet &amp; Wildlife Monitor</h5>
+    <div class="d-flex gap-2">
+        <button class="btn btn-sm btn-outline-secondary" id="btn-train" title="Train model">
+            <i class="bi bi-cpu me-1"></i>Train Model
+        </button>
+        <button class="btn btn-sm btn-outline-primary" data-bs-toggle="modal" data-bs-target="#registerPetModal">
+            <i class="bi bi-plus-lg me-1"></i>Register Pet
+        </button>
+    </div>
+</div>
+
+<!-- Per-pet status cards -->
+<div class="row g-3 mb-3" id="pet-cards">
+    {% for pet in pets %}
+    <div class="col-sm-6 col-lg-4 col-xl-3">
+        <div class="card bg-dark border-secondary h-100 pet-card" data-pet-id="{{ pet.id }}">
+            <div class="card-body d-flex flex-column gap-1">
+                <div class="d-flex justify-content-between align-items-start">
+                    <div>
+                        <span class="fs-4 me-2">{{ '🐈\u200d⬛' if pet.species == 'cat' else '🐕' if pet.species == 'dog' else '🐾' }}</span>
+                        <span class="fw-semibold">{{ pet.name }}</span>
+                        {% if pet.breed %}
+                        <small class="text-muted ms-1">{{ pet.breed }}</small>
+                        {% endif %}
+                    </div>
+                    <span class="badge pet-status-badge bg-secondary" id="status-{{ pet.id }}">
+                        <i class="bi bi-circle-fill me-1"></i>—
+                    </span>
+                </div>
+                <div class="small text-muted mt-1">
+                    <div><i class="bi bi-camera me-1"></i><span class="pet-location" id="loc-{{ pet.id }}">—</span></div>
+                    <div><i class="bi bi-clock me-1"></i><span class="pet-last-seen" id="seen-{{ pet.id }}">—</span></div>
+                </div>
+            </div>
+        </div>
+    </div>
+    {% else %}
+    <div class="col-12">
+        <div class="card bg-dark border-secondary">
+            <div class="card-body text-center text-muted py-4">
+                <i class="bi bi-heart me-2"></i>No pets registered yet.
+                <a href="#" class="ms-2" data-bs-toggle="modal" data-bs-target="#registerPetModal">Register your first pet</a>
+            </div>
+        </div>
+    </div>
+    {% endfor %}
+</div>
+
+<!-- Wildlife summary bar -->
+<div class="card bg-dark border-secondary mb-3">
+    <div class="card-header border-secondary d-flex justify-content-between align-items-center py-2">
+        <span class="small fw-semibold"><i class="bi bi-binoculars me-2"></i>Wildlife Today</span>
+        <small class="text-muted" id="wildlife-updated">—</small>
+    </div>
+    <div class="card-body py-2">
+        <div class="d-flex flex-wrap gap-3" id="wildlife-summary">
+            <span class="text-muted small">Loading…</span>
+        </div>
+    </div>
+</div>
+
+<!-- Activity timeline -->
+<div class="card bg-dark border-secondary mb-3">
+    <div class="card-header border-secondary py-2">
+        <span class="small fw-semibold"><i class="bi bi-bar-chart-line me-2"></i>24-Hour Activity</span>
+    </div>
+    <div class="card-body p-2" id="activity-timeline">
+        {% for pet in pets %}
+        <div class="mb-2 activity-row" data-pet-id="{{ pet.id }}">
+            <div class="d-flex align-items-center mb-1">
+                <span class="small text-muted" style="min-width:8rem">
+                    {{ '🐈\u200d⬛' if pet.species == 'cat' else '🐕' if pet.species == 'dog' else '🐾' }}
+                    {{ pet.name }}
+                </span>
+                <div class="flex-grow-1 position-relative" style="height:20px;background:rgba(255,255,255,.05);border-radius:3px;">
+                    <canvas class="activity-canvas w-100 h-100" data-pet-id="{{ pet.id }}" style="display:block;"></canvas>
+                </div>
+            </div>
+        </div>
+        {% else %}
+        <div class="text-muted small text-center py-2">No pets registered</div>
+        {% endfor %}
+        <!-- Hour labels -->
+        {% if pets %}
+        <div class="d-flex justify-content-between small text-muted px-1" style="margin-left:8rem;">
+            <span>00:00</span><span>06:00</span><span>12:00</span><span>18:00</span><span>24:00</span>
+        </div>
+        {% endif %}
+    </div>
+</div>
+
+<!-- Unlabeled detection queue -->
+<div class="card bg-dark border-secondary mb-3">
+    <div class="card-header border-secondary d-flex justify-content-between align-items-center py-2">
+        <span class="small fw-semibold"><i class="bi bi-question-circle me-2"></i>Unlabeled Detections</span>
+        <span class="badge bg-secondary" id="unlabeled-count">0</span>
+    </div>
+    <div class="card-body">
+        <div class="row g-2" id="unlabeled-grid">
+            <div class="col-12 text-muted small text-center py-2">Loading…</div>
+        </div>
+    </div>
+</div>
+
+<!-- Highlight reel -->
+<div class="card bg-dark border-secondary mb-3">
+    <div class="card-header border-secondary py-2">
+        <span class="small fw-semibold"><i class="bi bi-stars me-2"></i>Today's Highlights</span>
+    </div>
+    <div class="card-body p-0">
+        <div class="table-responsive">
+            <table class="table table-dark table-hover mb-0 small">
+                <thead>
+                    <tr>
+                        <th>Pet</th>
+                        <th>Camera</th>
+                        <th>Time</th>
+                        <th>Description</th>
+                        <th></th>
+                    </tr>
+                </thead>
+                <tbody id="highlights-list">
+                    <tr><td colspan="5" class="text-center text-muted py-3">Loading…</td></tr>
+                </tbody>
+            </table>
+        </div>
+    </div>
+</div>
+
+<!-- Register Pet Modal -->
+<div class="modal fade" id="registerPetModal" tabindex="-1" aria-labelledby="registerPetLabel" aria-hidden="true">
+    <div class="modal-dialog">
+        <div class="modal-content bg-dark border-secondary">
+            <div class="modal-header border-secondary">
+                <h6 class="modal-title" id="registerPetLabel">
+                    <i class="bi bi-plus-circle me-2"></i>Register Pet
+                </h6>
+                <button type="button" class="btn-close btn-close-white" data-bs-dismiss="modal"></button>
+            </div>
+            <div class="modal-body">
+                <form id="register-pet-form">
+                    <div class="mb-3">
+                        <label class="form-label small">Name <span class="text-danger">*</span></label>
+                        <input type="text" class="form-control form-control-sm bg-dark text-light border-secondary"
+                               id="pet-name" name="name" required placeholder="e.g. Mochi">
+                    </div>
+                    <div class="mb-3">
+                        <label class="form-label small">Species <span class="text-danger">*</span></label>
+                        <select class="form-select form-select-sm bg-dark text-light border-secondary"
+                                id="pet-species" name="species" required>
+                            <option value="">Select…</option>
+                            <option value="cat">Cat</option>
+                            <option value="dog">Dog</option>
+                            <option value="other">Other</option>
+                        </select>
+                    </div>
+                    <div class="mb-3">
+                        <label class="form-label small">Breed <span class="text-muted">(optional)</span></label>
+                        <input type="text" class="form-control form-control-sm bg-dark text-light border-secondary"
+                               id="pet-breed" name="breed" placeholder="e.g. Tabby">
+                    </div>
+                    <div class="mb-3">
+                        <label class="form-label small">Color / Description <span class="text-muted">(optional)</span></label>
+                        <input type="text" class="form-control form-control-sm bg-dark text-light border-secondary"
+                               id="pet-color" name="color_description" placeholder="e.g. orange with white socks">
+                    </div>
+                    <div class="mb-3">
+                        <label class="form-label small">Upload Photos <span class="text-muted">(optional)</span></label>
+                        <input type="file" class="form-control form-control-sm bg-dark text-light border-secondary"
+                               id="pet-photos" name="photos" accept="image/*" multiple>
+                        <div class="form-text text-muted small">Upload clear photos to improve detection accuracy.</div>
+                    </div>
+                    <div id="register-error" class="alert alert-danger py-2 small d-none"></div>
+                </form>
+            </div>
+            <div class="modal-footer border-secondary">
+                <button type="button" class="btn btn-sm btn-secondary" data-bs-dismiss="modal">Cancel</button>
+                <button type="button" class="btn btn-sm btn-primary" id="btn-register-submit">
+                    <i class="bi bi-check-lg me-1"></i>Register
+                </button>
+            </div>
+        </div>
+    </div>
+</div>
+
+<!-- Label crop modal -->
+<div class="modal fade" id="labelModal" tabindex="-1" aria-labelledby="labelModalLabel" aria-hidden="true">
+    <div class="modal-dialog modal-sm">
+        <div class="modal-content bg-dark border-secondary">
+            <div class="modal-header border-secondary">
+                <h6 class="modal-title" id="labelModalLabel">Label Detection</h6>
+                <button type="button" class="btn-close btn-close-white" data-bs-dismiss="modal"></button>
+            </div>
+            <div class="modal-body text-center">
+                <img id="label-crop-img" src="" alt="Detection crop"
+                     class="img-fluid rounded mb-3" style="max-height:200px;">
+                <div class="d-grid gap-2" id="label-pet-buttons">
+                    <!-- Populated by JS -->
+                </div>
+                <button class="btn btn-sm btn-outline-secondary mt-2 w-100" id="btn-label-unknown">
+                    Not a registered pet
+                </button>
+            </div>
+        </div>
+    </div>
+</div>
+{% endblock %}
+
+{% block scripts %}
+<script>
+(function () {
+    'use strict';
+
+    const PET_SPECIES_ICON = { cat: '🐈\u200d⬛', dog: '🐕' };
+    const THREAT_COLORS = { low: 'success', medium: 'warning', high: 'danger' };
+
+    // ---- pet status -------------------------------------------------------
+    function relativeTime(isoStr) {
+        if (!isoStr) return '—';
+        const diff = Math.floor((Date.now() - new Date(isoStr)) / 1000);
+        if (diff < 60) return `${diff}s ago`;
+        if (diff < 3600) return `${Math.floor(diff / 60)}m ago`;
+        if (diff < 86400) return `${Math.floor(diff / 3600)}h ago`;
+        return new Date(isoStr).toLocaleDateString();
+    }
+
+    function statusBadgeClass(status) {
+        if (!status) return 'bg-secondary';
+        const s = status.toLowerCase();
+        if (s === 'safe' || s === 'interior') return 'bg-success';
+        if (s === 'check' || s === 'transition') return 'bg-warning text-dark';
+        if (s === 'alert' || s === 'exterior') return 'bg-danger';
+        return 'bg-secondary';
+    }
+
+    function loadPetStatus() {
+        fetch('/pets/api/status')
+            .then(r => r.ok ? r.json() : Promise.reject(r.status))
+            .then(data => {
+                (data.pets || []).forEach(pet => {
+                    const badge = document.getElementById('status-' + pet.id);
+                    const loc   = document.getElementById('loc-'    + pet.id);
+                    const seen  = document.getElementById('seen-'   + pet.id);
+                    if (badge) {
+                        badge.className = 'badge pet-status-badge ' + statusBadgeClass(pet.status);
+                        badge.innerHTML = `<i class="bi bi-circle-fill me-1"></i>${pet.status || '—'}`;
+                    }
+                    if (loc)  loc.textContent  = pet.last_camera || '—';
+                    if (seen) seen.textContent = relativeTime(pet.last_seen);
+                });
+            })
+            .catch(() => {});
+    }
+
+    // ---- wildlife summary -------------------------------------------------
+    function loadWildlife() {
+        fetch('/pets/api/wildlife')
+            .then(r => r.ok ? r.json() : Promise.reject(r.status))
+            .then(data => {
+                const bar = document.getElementById('wildlife-summary');
+                const upd = document.getElementById('wildlife-updated');
+                const counts = {};
+                (data.sightings || []).forEach(s => {
+                    const key = s.species || 'unknown';
+                    if (!counts[key]) counts[key] = { total: 0, tier: s.threat_tier || 'low' };
+                    counts[key].total++;
+                });
+                if (Object.keys(counts).length === 0) {
+                    bar.innerHTML = '<span class="text-muted small">No wildlife detected today</span>';
+                } else {
+                    bar.innerHTML = Object.entries(counts).map(([sp, v]) =>
+                        `<span class="badge bg-${THREAT_COLORS[v.tier] || 'secondary'}">
+                            ${sp} <span class="fw-bold">${v.total}</span>
+                        </span>`
+                    ).join('');
+                }
+                upd.textContent = 'Updated ' + new Date().toLocaleTimeString();
+            })
+            .catch(() => {
+                document.getElementById('wildlife-summary').innerHTML =
+                    '<span class="text-muted small">Unavailable</span>';
+            });
+    }
+
+    // ---- activity timeline ------------------------------------------------
+    function loadActivityTimeline() {
+        fetch('/pets/api/sightings')
+            .then(r => r.ok ? r.json() : Promise.reject(r.status))
+            .then(data => {
+                const sightings = data.sightings || [];
+                document.querySelectorAll('.activity-canvas').forEach(canvas => {
+                    const petId = canvas.dataset.petId;
+                    const petSightings = sightings.filter(s => String(s.pet_id) === String(petId));
+                    drawActivityBar(canvas, petSightings);
+                });
+            })
+            .catch(() => {});
+    }
+
+    function drawActivityBar(canvas, sightings) {
+        const ctx = canvas.getContext('2d');
+        const W = canvas.offsetWidth || 400;
+        const H = canvas.offsetHeight || 20;
+        canvas.width  = W;
+        canvas.height = H;
+        ctx.clearRect(0, 0, W, H);
+
+        const now = new Date();
+        const dayStart = new Date(now.getFullYear(), now.getMonth(), now.getDate()).getTime();
+        const dayMs = 86400000;
+
+        sightings.forEach(s => {
+            const t = new Date(s.ts).getTime();
+            const x = Math.floor(((t - dayStart) / dayMs) * W);
+            const w = Math.max(2, Math.floor(((s.duration_s || 60) * 1000 / dayMs) * W));
+            ctx.fillStyle = s.location === 'exterior' ? '#dc3545' :
+                            s.location === 'transition' ? '#ffc107' : '#198754';
+            ctx.fillRect(x, 1, w, H - 2);
+        });
+    }
+
+    // ---- unlabeled crops --------------------------------------------------
+    let pendingLabelCropId = null;
+
+    function loadUnlabeled() {
+        fetch('/pets/api/unlabeled')
+            .then(r => r.ok ? r.json() : Promise.reject(r.status))
+            .then(data => {
+                const crops = data.crops || [];
+                const grid  = document.getElementById('unlabeled-grid');
+                const badge = document.getElementById('unlabeled-count');
+                badge.textContent = crops.length;
+
+                if (crops.length === 0) {
+                    grid.innerHTML = '<div class="col-12 text-muted small text-center py-2">No unlabeled detections</div>';
+                    return;
+                }
+
+                grid.innerHTML = crops.map(crop =>
+                    `<div class="col-6 col-sm-4 col-md-3 col-lg-2">
+                        <div class="card bg-dark border-secondary h-100 crop-thumb" data-crop-id="${crop.id}"
+                             style="cursor:pointer;" data-img="${crop.thumb_url || ''}">
+                            <img src="${crop.thumb_url || ''}" class="card-img-top" alt="Detection crop"
+                                 style="object-fit:cover;height:80px;"
+                                 onerror="this.src='/static/img/no-preview.svg'">
+                            <div class="card-body p-1 small text-muted text-center">
+                                ${crop.camera_id || '?'}<br>
+                                <span class="x-small">${relativeTime(crop.ts)}</span>
+                            </div>
+                        </div>
+                    </div>`
+                ).join('');
+
+                grid.querySelectorAll('.crop-thumb').forEach(el => {
+                    el.addEventListener('click', () => openLabelModal(el.dataset.cropId, el.dataset.img));
+                });
+            })
+            .catch(() => {
+                document.getElementById('unlabeled-grid').innerHTML =
+                    '<div class="col-12 text-muted small text-center py-2">Unavailable</div>';
+            });
+    }
+
+    function openLabelModal(cropId, imgUrl) {
+        pendingLabelCropId = cropId;
+        document.getElementById('label-crop-img').src = imgUrl;
+
+        const btnContainer = document.getElementById('label-pet-buttons');
+        const petCards = document.querySelectorAll('.pet-card');
+        btnContainer.innerHTML = '';
+        petCards.forEach(card => {
+            const petId = card.dataset.petId;
+            const name  = card.querySelector('.fw-semibold')?.textContent?.trim() || petId;
+            const btn   = document.createElement('button');
+            btn.className = 'btn btn-sm btn-outline-primary w-100';
+            btn.textContent = name;
+            btn.addEventListener('click', () => submitLabel(petId));
+            btnContainer.appendChild(btn);
+        });
+
+        if (petCards.length === 0) {
+            btnContainer.innerHTML = '<p class="text-muted small">No pets registered.</p>';
+        }
+
+        bootstrap.Modal.getOrCreateInstance(document.getElementById('labelModal')).show();
+    }
+
+    function submitLabel(petId) {
+        if (!pendingLabelCropId) return;
+        fetch(`/pets/${pendingLabelCropId}/label`, {
+            method: 'POST',
+            headers: { 'Content-Type': 'application/json' },
+            body: JSON.stringify({ sighting_id: pendingLabelCropId, pet_id: petId }),
+        }).finally(() => {
+            bootstrap.Modal.getOrCreateInstance(document.getElementById('labelModal')).hide();
+            loadUnlabeled();
+        });
+    }
+
+    document.getElementById('btn-label-unknown')?.addEventListener('click', () => {
+        submitLabel(null);
+    });
+
+    // ---- highlights -------------------------------------------------------
+    function loadHighlights() {
+        fetch('/pets/api/highlights')
+            .then(r => r.ok ? r.json() : Promise.reject(r.status))
+            .then(data => {
+                const list = document.getElementById('highlights-list');
+                const rows = data.highlights || [];
+                if (rows.length === 0) {
+                    list.innerHTML = '<tr><td colspan="5" class="text-center text-muted py-3">No highlights yet today</td></tr>';
+                    return;
+                }
+                list.innerHTML = rows.map(h =>
+                    `<tr>
+                        <td>${h.pet_name || '—'}</td>
+                        <td>${h.camera_id || '—'}</td>
+                        <td>${relativeTime(h.ts)}</td>
+                        <td>${h.description || '—'}</td>
+                        <td>
+                            ${h.recording_id
+                                ? `<a href="/recordings/${h.recording_id}/download"
+                                      class="btn btn-sm btn-outline-light" title="Download">
+                                       <i class="bi bi-play-circle"></i>
+                                   </a>`
+                                : ''}
+                        </td>
+                    </tr>`
+                ).join('');
+            })
+            .catch(() => {
+                document.getElementById('highlights-list').innerHTML =
+                    '<tr><td colspan="5" class="text-center text-muted py-3">Unavailable</td></tr>';
+            });
+    }
+
+    // ---- register pet form ------------------------------------------------
+    document.getElementById('btn-register-submit').addEventListener('click', () => {
+        const form = document.getElementById('register-pet-form');
+        const errBox = document.getElementById('register-error');
+        errBox.classList.add('d-none');
+
+        const name    = document.getElementById('pet-name').value.trim();
+        const species = document.getElementById('pet-species').value;
+        if (!name || !species) {
+            errBox.textContent = 'Name and species are required.';
+            errBox.classList.remove('d-none');
+            return;
+        }
+
+        const body = { name, species };
+        const breed = document.getElementById('pet-breed').value.trim();
+        const color = document.getElementById('pet-color').value.trim();
+        if (breed) body.breed = breed;
+        if (color) body.color_description = color;
+
+        fetch('/pets/register', {
+            method: 'POST',
+            headers: { 'Content-Type': 'application/json' },
+            body: JSON.stringify(body),
+        })
+            .then(r => r.ok ? r.json() : r.json().then(e => Promise.reject(e.error || 'Error')))
+            .then(() => {
+                bootstrap.Modal.getOrCreateInstance(document.getElementById('registerPetModal')).hide();
+                form.reset();
+                window.location.reload();
+            })
+            .catch(err => {
+                errBox.textContent = typeof err === 'string' ? err : 'Registration failed.';
+                errBox.classList.remove('d-none');
+            });
+    });
+
+    // ---- train model button -----------------------------------------------
+    document.getElementById('btn-train').addEventListener('click', () => {
+        const btn = document.getElementById('btn-train');
+        btn.disabled = true;
+        btn.innerHTML = '<span class="spinner-border spinner-border-sm me-1"></span>Training…';
+        fetch('/pets/train', { method: 'POST' })
+            .then(r => r.ok ? r.json() : Promise.reject())
+            .then(() => {
+                btn.innerHTML = '<i class="bi bi-check-lg me-1"></i>Queued';
+                setTimeout(() => {
+                    btn.disabled = false;
+                    btn.innerHTML = '<i class="bi bi-cpu me-1"></i>Train Model';
+                }, 3000);
+            })
+            .catch(() => {
+                btn.disabled = false;
+                btn.innerHTML = '<i class="bi bi-cpu me-1"></i>Train Model';
+            });
+    });
+
+    // ---- initial load + polling ------------------------------------------
+    function refreshAll() {
+        loadPetStatus();
+        loadWildlife();
+        loadActivityTimeline();
+        loadUnlabeled();
+        loadHighlights();
+    }
+
+    refreshAll();
+    setInterval(refreshAll, 30000);
+})();
+</script>
+{% endblock %}

vigilar/web/templates/recordings.html

@@ -1,6 +1,10 @@
 {% extends "base.html" %}
 {% block title %}Vigilar — Recordings{% endblock %}
 
+{% block head %}
+<link rel="stylesheet" href="{{ url_for('static', filename='css/pet-labeling.css') }}">
+{% endblock %}
+
 {% block content %}
 <div class="d-flex justify-content-between align-items-center mb-3">
     <h5 class="mb-0"><i class="bi bi-camera-video me-2"></i>Recordings</h5>
@@ -38,7 +42,10 @@
         </button>
     </div>
     <div class="card-body p-0">
-        <video id="player-video" class="w-100" controls></video>
+        <!-- data-pet-labeling and data-detections are updated by JS when a recording is loaded -->
+        <div class="position-relative" id="player-video-wrap" data-pet-labeling="true" data-detections="[]">
+            <video id="player-video" class="w-100" controls></video>
+        </div>
     </div>
 </div>
 
@@ -90,6 +97,7 @@
 </div>
 
 <script src="{{ url_for('static', filename='js/timeline.js') }}"></script>
+<script src="{{ url_for('static', filename='js/pet-labeling.js') }}"></script>
 <script>
 (function() {
     'use strict';
@@ -112,11 +120,35 @@
         playerVideo.src = '';
     });
 
+    const playerVideoWrap = document.getElementById('player-video-wrap');
+
+    async function loadDetections(recordingId) {
+        try {
+            const resp = await fetch(`/recordings/${recordingId}/detections`);
+            if (!resp.ok) return [];
+            return await resp.json();
+        } catch (e) {
+            return [];
+        }
+    }
+
     function handleSegmentClick(seg) {
         playerCard.classList.remove('d-none');
         playerTitle.textContent = `Recording #${seg.id} (${seg.type || 'motion'})`;
         playerVideo.src = `/recordings/${seg.id}/download`;
         playerVideo.play().catch(() => {});
+
+        // Reset overlay while detections load
+        if (window.PetLabeling) {
+            PetLabeling.updateDetections(playerVideoWrap, []);
+        }
+
+        // Fetch detections for this recording and render overlay
+        loadDetections(seg.id).then(detections => {
+            if (window.PetLabeling) {
+                PetLabeling.updateDetections(playerVideoWrap, detections);
+            }
+        });
     }
 
     function loadTimeline() {