API working

Move to new db setup
2026-09-09 11:19:00 -04:00 · 2026-09-08 20:31:06 -04:00
18 changed files with 478 additions and 2043 deletions
@@ -6,6 +6,11 @@
 # -----------------------------------------------------
 DB_ROOT_PASSWORD=changeme_root
 DB_PASSWORD=changeme_videodetect
 DB_USER=videodetect
 DB_NAME=videodetect
 DB_HOST=mariadb
 API_HOST=https://api:3000
 # -----------------------------------------------------
 # Storage paths (host-side mounts)
@@ -14,7 +19,3 @@ NAS_OUTPUT_PATH=${PWD}/output
 MODELS_PATH=${PWD}/models
 TRAINING_PATH=${PWD}/training
 # -----------------------------------------------------
 # Grafana
 # -----------------------------------------------------
 GRAFANA_PASSWORD=changeme_grafana
@@ -1,170 +0,0 @@
 # STORY-01: Foundation & Infrastructure
 ## Epic
 **E6: Infrastructure** — As a DevOps engineer, I can deploy the entire stack via Docker Compose.
 ## Related Requirements
 | ID | Requirement |
 |----|-------------|
 | TC-01 | Hardware: 2× Tesla P40 24GB (compute capability 5.2, PCIe 3.0, no Tensor Cores) |
 | TC-02 | CUDA/Torch Compatibility: CUDA ≤ 11.8, PyTorch ≤ 2.1.0, FP32 inference only |
 | TC-03 | Storage I/O: Fast local NVMe/SSD for temp frame cache; shared NAS/SMB for video input/output |
 | TC-04 | Framework Stack: PyTorch → ONNX → TensorRT FP32; FFmpeg/OpenCV; MariaDB |
 | TC-05 | Deployment Model: Docker Compose orchestrates all services; GPUs passed via nvidia-container-toolkit |
 | TC-06 | Network Security: Internal LAN only; no reverse proxy, SSL, or auth |
 | NFR-04 | Determinism & Reproducibility: Config-seeded randomness, versioned models |
 ## Description
 Establish the Docker environment, database schema, and basic connectivity. No video processing logic yet — just the operational skeleton that all subsequent stories depend on.
 ## Scope
 ### In Scope
 - Docker Compose multi-service orchestration
 - Custom Worker Dockerfile with CUDA 11.8 + PyTorch 2.1.0 + TensorRT FP32
 - MariaDB schema design and migration scripts
 - Volume mounts for NVMe scratch, NAS input/output, and model persistence
 - Configuration management (config.yaml + environment variables)
 - Structured logging setup (JSON format to stdout)
 - Database connection layer with connection pooling
 - Network connectivity verification between all services
 ### Out of Scope
 - Video processing logic (covered in STORY-02 through STORY-05)
 - Model loading or inference (covered in STORY-04)
 - Review UI functionality (covered in STORY-06)
 - Monitoring dashboards (covered in STORY-08)
 - Active learning pipeline (covered in STORY-07)
 ## Deliverables
 ### 1.1 Docker Compose Structure
 **File:** `docker-compose.yml`
 Services defined:
 - `mariadb`: MariaDB 10.11+ with persistent volume
 - `worker`: PyTorch/TensorRT inference worker with GPU passthrough
 - `ui`: Placeholder service (nginx serving static page) for network verification
 Key configurations:
 - `nvidia-container-toolkit` runtime configuration for GPU passthrough
 - Volume mounts:
  - NVMe → `/scratch` (tmpfs for speed)
  - NAS/SMB → `/data/input` and `/data/output`
  - Persistent → `/models` and `/data/training`
 - Network bridge for inter-service communication
 - Resource limits (GPU memory caps per NFR-03)
 ### 1.2 Worker Dockerfile
 **File:** `worker/Dockerfile`
 Base image: `nvidia/cuda:11.8.0-runtime-ubuntu22.04`
 Installed packages:
 - PyTorch 2.1.0 (CUDA 11.8, FP32 only)
 - TensorRT 8.6+ (FP32)
 - OpenCV 4.8+
 - FFmpeg 5.x + ffprobe
 - ONNX Runtime
 - Python 3.10+
 - Required system libraries (libcudnn8, libglib2.0, etc.)
 ### 1.3 Database Schema
 **File:** `db/schema.sql`
 Tables:
 - `videos`: `id` (BIGINT PK), `file_path` (VARCHAR), `file_hash` (CHAR(64)), `resolution_w` (INT), `resolution_h` (INT), `codec` (VARCHAR), `duration` (FLOAT), `status` (ENUM: NEW, PENDING, PROCESSING, COMPLETED, UNSCANNABLE, ERROR), `last_scan_time` (DATETIME), `last_processed_time` (DATETIME), `created_at` (DATETIME), `updated_at` (DATETIME)
 - `processing_logs`: `id` (BIGINT PK), `video_id` (BIGINT FK), `model_version` (VARCHAR), `frame_count` (INT), `confidence_score` (FLOAT), `routing_decision` (ENUM: MATCH, REVIEW, SKIP), `processed_at` (DATETIME), `error_message` (TEXT)
 - `models`: `version` (VARCHAR PK), `status` (ENUM: ACTIVE, CANDIDATE, ARCHIVED), `path` (VARCHAR), `calibration_temp` (FLOAT), `f1_score` (FLOAT), `ece_score` (FLOAT), `deployed_at` (DATETIME), `created_at` (DATETIME)
 - `review_queue`: `id` (BIGINT PK), `video_id` (BIGINT FK), `confidence_score` (FLOAT), `routing_decision` (ENUM: REVIEW), `annotated` (BOOLEAN), `ground_truth` (BOOLEAN), `annotated_at` (DATETIME), `created_at` (DATETIME)
 Indexes:
 - `idx_videos_status` on `videos(status)`
 - `idx_videos_file_hash` on `videos(file_hash)` (UNIQUE)
 - `idx_videos_last_scan` on `videos(last_scan_time)`
 - `idx_processing_logs_video` on `processing_logs(video_id)`
 - `idx_models_status` on `models(status)`
 - `idx_review_queue_annotated` on `review_queue(annotated)`
 ### 1.4 Database Connection Layer
 **File:** `src/db_connector.py`
 Features:
 - Connection pooling (DBUtils + PyMySQL)
 - Configurable pool size (min=5, max=20)
 - Automatic reconnection on disconnect
 - Context manager support
 - Prepared statements for all queries
 - Transaction support for atomic state transitions
 ### 1.5 Configuration Management
 **File:** `config.yaml`
 Contents:
 - `sampling`: `interval_seconds: 30`, `override_per_job: true`
 - `thresholds`: `T_high: 0.75`, `T_low: 0.45`
 - `gpu`: `max_memory_gb: 18`, `batch_size: auto`, `device: cuda`
 - `storage`: `scratch_path: /scratch`, `input_path: /data/input`, `output_path: /data/output`
 - `database`: `host: mariadb`, `port: 3306`, `pool_size: 20`
 - `logging`: `format: json`, `level: INFO`
 - `model`: `face_detector: yolo8n`, `classifier: mobilenetv3`, `input_size: 224`
 **File:** `src/config_loader.py`
 Features:
 - Load config.yaml with environment variable overrides
 - Validate all required fields
 - Provide typed accessors (e.g., `config.thresholds.T_high`)
 - Hot-reload support for config changes
 ### 1.6 Logging Setup
 **File:** `src/logging_config.py`
 - JSON structured logging via `python-json-logger`
 - Log levels: DEBUG, INFO, WARNING, ERROR, CRITICAL
 - Fields: timestamp, level, service, video_id, message, metadata (key-value pairs)
 - Log rotation: 100MB per file, 10 files max
 - All logs to stdout for Docker capture
 ## Acceptance Criteria
 ### Functional
 - [ ] `docker-compose up` starts MariaDB, Worker, and UI containers successfully
 - [ ] Worker container can connect to MariaDB and execute schema migrations
 - [ ] GPU is visible inside Worker container (`nvidia-smi` shows Tesla P40)
 - [ ] Volume mounts are accessible and writable in all containers
 - [ ] Config.yaml loads correctly with all required fields validated
 - [ ] Structured logging produces valid JSON output in all services
 - [ ] Database connection pool handles concurrent connections (test with 20 simultaneous)
 ### Non-Functional
 - [ ] Worker container starts within 30 seconds
 - [ ] MariaDB starts within 15 seconds
 - [ ] GPU memory usage in Worker is < 2GB at idle (before model loading)
 - [ ] All services communicate over Docker internal network (no host network exposure except UI port)
 - [ ] Schema migration is idempotent (running twice produces same result)
 ### Technical Constraints
 - [ ] CUDA version in Worker is 11.8 (verified via `torch.version.cuda`)
 - [ ] PyTorch version ≤ 2.1.0 (verified via `torch.__version__`)
 - [ ] TensorRT runs in FP32 mode only
 - [ ] No Tensor Cores used (compute capability 5.2 constraint respected)
 - [ ] No SSL, auth, or reverse proxy configured (TC-06)
 ## Dependencies
 - **Prerequisites:** NVIDIA Container Toolkit installed on host, Docker Compose v2+, NAS/SMB mounts configured
 - **Depends on:** None (this is the foundational story)
 - **Enables:** STORY-02 (Ingestion), STORY-03 (Orchestration), STORY-04 (Inference), STORY-05 (Results), STORY-06 (Review UI), STORY-07 (Active Learning), STORY-08 (Monitoring)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|-----------|
 | Tesla P40 (CC 5.2) incompatible with newer TensorRT | Use TensorRT 8.6 which supports CC 5.x; test early |
 | CUDA 11.8 + PyTorch 2.1.0 dependency conflicts | Pin all versions in Dockerfile; use nvidia base image |
 | NAS/SMB mount latency affects processing | Use local tmpfs for scratch; only read from NAS |
 | MariaDB connection pool exhaustion | Monitor pool metrics; tune pool_size based on worker count |
 ## Estimated Effort
 - **Sprint:** 1-2
 - **Story Points:** 13
 - **Dependencies:** None
@@ -1,182 +0,0 @@
 # STORY-02: Core Ingestion & Codec Handling
 ## Epic
 **E5: Data Management** — As a system, I can scan directories and sync file state to MariaDB.
 **E1: Core Pipeline** — As a system, I can detect video resolution and codec.
 ## Related Requirements
 | ID | Requirement |
 |----|---------|
 | FR-09 | Directory Scanning & Sync: Process to scan input directories, detect new/removed files, and sync state to MariaDB |
 | FR-10 | Codec & Resolution Detection & Handling: Detect video properties; handle unsupported codecs by flagging files as UNSCANNABLE |
 | NFR-07 | Data Volume Handling: Efficient indexing for ~30TB dataset |
 | NFR-08 | Codec Agnosticism: Handle H.264, H.265, VP8, VP9, AV1, MJPEG, etc. |
 | NFR-05 | Fault Tolerance: Auto-retry on transient failures; skip & log on fatal errors |
 ## Description
 Implement the directory scanner, video probing, and robust error handling for codecs and resolutions. This story enables the system to discover new files in the 30TB corpus, extract their metadata, validate codec support, and maintain accurate database state.
 ## Scope
 ### In Scope
 - Background directory scanner service that walks `/data/input`
 - File detection and deduplication via SHA-256 hashing
 - Video probing via ffprobe (codec, resolution, duration)
 - Codec whitelist/blacklist validation
 - UNSCANNABLE status for unsupported or corrupt files
 - Priority queue / DB-based locking for job assignment
 - Atomic state transitions (PENDING → PROCESSING)
 - Efficient handling of 30TB directory structure (incremental scanning)
 ### Out of Scope
 - Frame extraction (covered in STORY-04)
 - Face detection and classification (covered in STORY-04)
 - Confidence scoring and routing (covered in STORY-05)
 - Review UI (covered in STORY-06)
 - Active learning pipeline (covered in STORY-07)
 - Monitoring dashboards (covered in STORY-08)
 ## Deliverables
 ### 2.1 Directory Scanner Service
 **File:** `src/scanner.py`
 Core components:
 - **Walker:** Recursive directory walker with configurable depth and path filters
 - **Incremental Sync:** Compare current filesystem state against DB `last_scan_time` — only process new/modified files
 - **File Detection:** Identify files not in DB or with status NEW/PENDING
 - **Hash Computation:** SHA-256 of first 1MB (or full file if < 1MB) for deduplication
 - **Scan Scheduler:** Configurable interval (default: every 60 seconds) via cron-like scheduler
 - **Concurrency:** Multi-threaded walker with configurable worker count (default: 8 threads)
 - **Error Handling:** Per-file error isolation — scanner continues on individual file failures
 Key behaviors:
 - On scan start: query DB for files with `last_scan_time < now()` or `status IN ('NEW', 'PENDING')`
 - For each file: compute hash → check DB for duplicate → if new, insert with status PENDING
 - For removed files: mark as REMOVED in DB (optional, configurable)
 - Update `last_scan_time` on `videos` table after successful scan
 ### 2.2 Video Probing Module
 **File:** `src/prober.py`
 Features:
 - **ffprobe wrapper:** Execute ffprobe with optimized arguments for speed
  ```bash
  ffprobe -v error -select_streams v:0 -show_entries stream=codec_name,width,height,duration,r_frame_rate -show_entries format=duration -of json <file>
  ```
 - **Metadata extraction:** Parse JSON output for:
  - `codec_name`: Video codec identifier
  - `width`, `height`: Resolution
  - `duration`: Video duration in seconds
  - `r_frame_rate`: Frame rate (for sampling calculations)
 - **Error handling:** Catch ffprobe failures (corrupt files, unsupported formats)
 - **Timeout:** ffprobe execution limited to 10 seconds per file
 ### 2.3 Codec Validation
 **File:** `src/codec_validator.py`
 Codec whitelist (supported):
 - H.264 (avc1)
 - H.265 (hevc)
 - VP8 (vp8)
 - VP9 (vp9)
 - AV1 (av01)
 - MJPEG (mjpeg)
 - MPEG-4 (mp4v)
 Codec blacklist (unsupported):
 - Theora
 - DivX/Xvid (legacy)
 - ProRes (requires special handling)
 - Any codec not in whitelist
 Behavior:
 - If codec not in whitelist → set status UNSCANNABLE, log reason
 - If ffprobe fails → set status UNSCANNABLE, log error code
 - If file is not a valid video → set status UNSCANNABLE, log reason
 - UNSCANNABLE files are excluded from processing pipeline
 ### 2.4 Batch Orchestration Skeleton
 **File:** `src/orchestrator.py`
 Components:
 - **Job Queue:** Priority queue based on file modification time (newest first)
 - **DB-based Locking:** Atomic state transition using UPDATE ... WHERE status = 'PENDING'
 - **Worker Pool:** Configurable number of worker processes (default: 2, one per GPU)
 - **State Machine:**
  ```
  NEW → PENDING → PROCESSING → COMPLETED
                         → UNSCANNABLE
                         → ERROR
  ```
 - **Concurrency Control:** Max concurrent processing per GPU (default: 1 video at a time per worker)
 - **Idempotency:** Re-processing a file does not duplicate DB entries or outputs
 ### 2.5 Configuration Updates
 **File:** `config.yaml` (updates to STORY-01)
 New fields:
 ```yaml
 scanner:
  scan_interval_seconds: 60
  walker_threads: 8
  ffprobe_timeout_seconds: 10
  hash_algorithm: sha256
  hash_chunk_size_mb: 1
 codec:
  whitelist: [avc1, hevc, vp8, vp9, av01, mjpeg, mp4v]
  default_status_on_error: UNSCANNABLE
 queue:
  priority: modification_time  # newest first
  max_concurrent_per_gpu: 1
  lock_timeout_seconds: 300
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] New files in `/data/input` appear in DB with correct metadata within 60 seconds of placement
 - [ ] File hash deduplication prevents re-processing identical files
 - [ ] ffprobe correctly extracts codec, resolution, and duration for all supported codecs
 - [ ] Unsupported codec files are marked UNSCANNABLE without crashing the scanner
 - [ ] Corrupt/unreadable files are marked UNSCANNABLE with appropriate error logged
 - [ ] Atomic state transitions: PENDING → PROCESSING succeeds only once (no duplicate processing)
 - [ ] Removed files are detected and marked REMOVED (if configured)
 - [ ] Scanner handles 30TB directory structure without OOM (memory < 500MB during scan)
 - [ ] Re-running scanner is idempotent — no duplicate entries or states
 ### Non-Functional
 - [ ] Scanner completes full 30TB directory walk in < 4 hours (incremental: < 10 minutes for typical day)
 - [ ] ffprobe timeout (10s) is enforced — does not hang on corrupt files
 - [ ] Scanner memory usage stays < 500MB regardless of directory depth
 - [ ] Hash computation for 100MB file completes in < 5 seconds on NVMe
 - [ ] Worker pool respects GPU count (2 workers for 2 GPUs)
 ### Technical Constraints
 - [ ] SHA-256 hash is deterministic and reproducible
 - [ ] Codec detection matches ffprobe output exactly
 - [ ] Resolution stored as integers (width, height)
 - [ ] Duration stored as float (seconds)
 - [ ] All scanner errors are logged with video_id, error_code, and file_path
 - [ ] Scanner does not modify video files (read-only operation)
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation & Infrastructure) — DB schema, connection layer, config
 - **Depends on:** None (runs in parallel with STORY-03 if needed)
 - **Enables:** STORY-04 (Inference Pipeline), STORY-05 (Results & Export)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | 30TB directory walk is extremely slow | Incremental scanning using last_scan_time; only walk new directories |
 | ffprobe hangs on corrupt files | Enforce 10-second timeout via subprocess timeout |
 | Hash computation on slow NAS is bottleneck | Hash only first 1MB; full hash optional for verification |
 | Concurrent scanner + worker conflicts | DB-based locking; scanner only writes PENDING, workers read PENDING |
 | Network mount latency | Cache directory listings; batch DB operations |
 ## Estimated Effort
 - **Sprint:** 3-4
 - **Story Points:** 21
 - **Dependencies:** STORY-01
@@ -1,152 +0,0 @@
 # STORY-03: Frame Sampling
 ## Epic
 **E1: Core Pipeline** — As an engineer, I can configure frame sampling interval and extract frames uniformly.
 ## Related Requirements
 | ID | Requirement |
 |----|---------|
 | FR-01 | Configurable frame sampling interval (default: 1 frame per 30 seconds); Must support override per job/batch; Uniform temporal sampling preferred |
 | NFR-02 | Latency per video: ≤ 45 seconds end-to-end (15-min avg video) |
 | NFR-03 | GPU Memory Safety: ≤ 18GB per GPU sustained |
 | NFR-05 | Fault Tolerance: Auto-retry on transient failures; skip & log on fatal errors |
 | TC-03 | Storage I/O: Fast local NVMe/SSD for temp frame cache |
 ## Description
 Implement frame extraction from processed videos using FFmpeg/OpenCV. Extract one frame per configurable interval (default: 30 seconds) uniformly across the video duration. Handle variable FPS, high-resolution frames, and manage scratch space efficiently.
 ## Scope
 ### In Scope
 - Uniform temporal frame sampling via FFmpeg `-ss` timestamp extraction
 - Configurable sampling interval (default: 30 seconds, override per job)
 - Variable FPS handling (calculate correct timestamps)
 - Frame extraction to JPEG format for storage efficiency
 - Scratch space management (NVMe tmpfs with auto-cleanup)
 - High-resolution frame handling (downscale 4K to fit VRAM constraints)
 - Frame naming convention: `{video_id}_{timestamp}.jpg`
 - Memory management (delete frames after processing)
 ### Out of Scope
 - Face detection (covered in STORY-04)
 - Classification (covered in STORY-05)
 - Confidence scoring (covered in STORY-05)
 - Results export (covered in STORY-06)
 - Review UI (covered in STORY-07)
 ## Deliverables
 ### 3.1 Frame Sampler Module
 **File:** `src/frame_sampler.py`
 Core functionality:
 - **Timestamp Calculator:** Compute uniform sampling timestamps
  ```python
  def calculate_timestamps(duration, interval):
      """Return list of timestamps for frame extraction."""
      count = max(1, int(duration / interval))
      step = duration / count
      return [i * step for i in range(count)]
  ```
 - **FFmpeg Extraction:** Extract frames at computed timestamps
  ```bash
  ffmpeg -ss {timestamp} -i {video_path} -vframes 1 -q:v 2 -f jpeg {output_path}
  ```
 - **Variable FPS Handling:** Adjust timestamps for videos with variable frame rates
 - **Quality Control:** JPEG quality factor 2 (high quality, reasonable size)
 - **Error Handling:** Per-frame error isolation — skip failed frames, log errors, continue
 ### 3.2 Resolution Handling
 **File:** `src/frame_sampler.py` (resolution handling section)
 Features:
 - **Resolution Detection:** Use metadata from prober (STORY-02)
 - **Downscale Logic:** Auto-downscale frames > 1080p to fit VRAM
  - 4K (3840x2160) → 1080p (1920x1080)
  - 2K (2560x1440) → 720p (1280x720)
  - ≤ 1080p → no change
 - **FFmpeg Scale Filter:** Apply during extraction
  ```bash
  ffmpeg -ss {ts} -i {video} -vf "scale=1920:1080:force_original_aspect_ratio=decrease,pad=1920:1080:(ow-iw)/2:(oh-ih)/2" -vframes 1 -q:v 2 -f jpeg {output}
  ```
 - **Aspect Ratio Preservation:** Letterbox/pillarbox to maintain original aspect ratio
 ### 3.3 Scratch Space Management
 **File:** `src/scratch_manager.py`
 Features:
 - **Directory Structure:** `/scratch/{video_id}/frames/` for each video
 - **Naming Convention:** `{video_id}_{timestamp_ms}.jpg`
 - **Auto-Cleanup:** Delete all frames after video processing completes
 - **Space Monitoring:** Alert if scratch usage exceeds 80% of tmpfs capacity
 - **Cleanup on Error:** Ensure frames are cleaned up even if processing fails
 - **Concurrent Access:** Per-video directory isolation (no conflicts between workers)
 ### 3.4 Configuration Updates
 **File:** `config.yaml` (updates to STORY-01/02)
 New fields:
 ```yaml
 sampling:
  interval_seconds: 30
  override_per_job: true
  quality: 2  # JPEG quality (1-31, lower=better)
  format: jpeg
 resolution:
  max_height: 1080
  preserve_aspect_ratio: true
  pad_to_square: false  # Will be done in face detection if needed
 scratch:
  path: /scratch
  max_usage_percent: 80
  auto_cleanup: true
  cleanup_on_error: true
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] Frames are extracted at uniform intervals (±1 frame tolerance) across video duration
 - [ ] Default interval of 30 seconds produces correct number of frames for test videos
 - [ ] Per-job interval override works (e.g., 10-second interval for specific videos)
 - [ ] Variable FPS videos produce evenly spaced timestamps (not evenly spaced frames)
 - [ ] 4K frames are downsampled to 1080p without crashing
 - [ ] Aspect ratio is preserved during downscaling (no distortion)
 - [ ] Frames are saved as high-quality JPEGs (quality factor 2)
 - [ ] Frame naming follows convention: `{video_id}_{timestamp_ms}.jpg`
 - [ ] Scratch space is cleaned up after video processing (no leftover frames)
 - [ ] Failed frame extraction does not stop processing of other frames
 ### Non-Functional
 - [ ] Frame extraction for a 15-minute video completes in < 10 seconds on NVMe
 - [ ] Scratch space usage per video < 50MB (typical case)
 - [ ] Memory usage during extraction < 500MB per worker
 - [ ] FFmpeg process timeout enforced (30 seconds per frame)
 - [ ] No frames left in scratch after processing completes (verified by directory check)
 ### Technical Constraints
 - [ ] FFmpeg uses `-ss` for accurate timestamp seeking (not frame-by-frame)
 - [ ] Downscaling uses FFmpeg scale filter (not post-processing resize)
 - [ ] JPEG quality is consistent across all extracted frames
 - [ ] Timestamps are in seconds with millisecond precision
 - [ ] Scratch directory is created per-video and isolated from other workers
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation), STORY-02 (Ingestion — provides metadata)
 - **Depends on:** None (runs after ingestion, before STORY-04)
 - **Enables:** STORY-04 (Face Detection — provides frames), STORY-05 (Classification)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | FFmpeg seeking is inaccurate for some codecs | Use `-accurate_seek` flag; fall back to frame-by-frame if needed |
 | 4K downscaling increases processing time | Use FFmpeg scale filter (GPU-accelerated if available) |
 | Scratch space fills up with many concurrent videos | Monitor usage; implement cleanup queue; alert at 80% |
 | Variable FPS causes uneven frame distribution | Calculate timestamps based on duration, not frame count |
 ## Estimated Effort
 - **Sprint:** 5 (first half)
 - **Story Points:** 13
 - **Dependencies:** STORY-01, STORY-02
@@ -1,168 +0,0 @@
 # STORY-04: Face Detection
 ## Epic
 **E1: Core Pipeline** — As an engineer, I can run face detection on sampled frames in batch.
 ## Related Requirements
 | ID | Requirement |
 |----|-------|
 | FR-02 | Face detection on all sampled frames: Lightweight detector only; no full-body or scene analysis |
 | NFR-03 | GPU Memory Safety: ≤ 18GB per GPU sustained |
 | NFR-01 | Throughput: ≥ 30 videos/hour/GPU |
 | TC-01 | Hardware: 2× Tesla P40 24GB (compute capability 5.2, PCIe 3.0, no Tensor Cores) |
 | TC-02 | CUDA/Torch Compatibility: CUDA ≤ 11.8, PyTorch ≤ 2.1.0, FP32 inference only |
 | TC-04 | Framework Stack: PyTorch → ONNX → TensorRT FP32 |
 ## Description
 Implement face detection on all sampled frames using a lightweight detector (YOLOv8n or equivalent). Convert the model to ONNX and TensorRT FP32 for inference. Implement dynamic batching to maximize GPU utilization while staying within VRAM constraints. Generate face crops for classification.
 ## Scope
 ### In Scope
 - YOLOv8n face detection model (or equivalent lightweight detector)
 - ONNX export and TensorRT FP32 engine conversion
 - Dynamic batching for face detection inference
 - Face crop extraction and resizing to inference input size (224×224)
 - GPU memory management (stay within 18GB per GPU)
 - Batch size auto-tuning based on available VRAM
 - Face bounding box output with confidence scores
 - Handling frames with no detected faces
 ### Out of Scope
 - Frame sampling (covered in STORY-03)
 - Demographic classification (covered in STORY-05)
 - Confidence aggregation (covered in STORY-05)
 - Results export (covered in STORY-06)
 - Review UI (covered in STORY-07)
 ## Deliverables
 ### 4.1 Face Detection Model
 **File:** `models/face_detector/`
 Components:
 - **Source Model:** YOLOv8n (Ultralytics) trained on face detection dataset (WIDER Face or equivalent)
 - **ONNX Export:** `face_detector.onnx` with proper input/output shapes
  - Input: `[1, 3, 640, 640]` (RGB, BGR depending on model)
  - Output: `[1, num_anchors, 4+1]` (bbox coordinates + confidence)
 - **TensorRT Engine:** `face_detector.trt` (FP32, built for Tesla P40 CC 5.2)
  - Builder config: max_batch_size=32, max_workspace_size=4GB
  - Serialization for runtime loading
 - **Model Metadata:** `model.json` with input shape, normalization params, calibration data
 ### 4.2 Face Detection Runner
 **File:** `src/face_detector.py`
 Core components:
 - **Model Loader:** Load TensorRT engine at startup
  ```python
  class FaceDetector:
      def __init__(self, engine_path, device='cuda'):
          self.engine = load_trt_engine(engine_path)
          self.context = self.engine.create_execution_context()
          self.input_shape = (1, 3, 640, 640)
          self.device = device
  ```
 - **Preprocessing:** Convert frame to model input format
  - Resize to 640×640
  - Normalize (mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
  - Convert to NHWC/NCHW as required by TensorRT
 - **Inference:** Run batched detection on multiple frames
  - Dynamic batching: accumulate frames until batch_size reached or timeout
  - Execute TensorRT engine
  - Parse outputs (bounding boxes + confidence scores)
 - **Post-processing:** NMS (Non-Maximum Suppression) to remove duplicate detections
  - IoU threshold: 0.45
  - Confidence threshold: 0.25 (low to avoid missing faces)
 - **Crop Generation:** Extract face crops from original frames
  - Resize crops to 224×224 for classification input
  - Save crops to `/scratch/{video_id}/crops/`
 ### 4.3 Dynamic Batching
 **File:** `src/batcher.py`
 Features:
 - **Batch Accumulator:** Collect face detections across frames
  - Max batch size: auto-tuned based on available VRAM (start at 16, adjust dynamically)
  - Timeout: 100ms (process batch even if not full)
 - **VRAM Monitoring:** Track GPU memory usage and adjust batch size
  - If VRAM > 16GB: reduce batch size by 25%
  - If VRAM < 10GB: increase batch size by 25%
 - **Frame Ordering:** Maintain frame-to-batch mapping for correct crop assignment
 ### 4.4 GPU Memory Management
 **File:** `src/gpu_manager.py`
 Features:
 - **Memory Tracking:** Monitor GPU memory usage continuously
 - **Peak Allocation:** Ensure total GPU memory < 18GB (leaving 6GB headroom)
 - **Cleanup:** Explicitly free TensorRT buffers after each batch
 - **Per-Video Limits:** Max faces per video (e.g., 100) to prevent memory issues
 - **Fallback:** If VRAM is critically low, reduce batch size and retry
 ### 4.5 Configuration Updates
 **File:** `config.yaml` (updates)
 New fields:
 ```yaml
 face_detection:
  model: yolo8n
  model_path: /models/face_detector/face_detector.trt
  input_size: 640
  confidence_threshold: 0.25
  iou_threshold: 0.45
  max_faces_per_frame: 10
  max_faces_per_video: 100
 batching:
  max_batch_size: 16
  batch_timeout_ms: 100
  vram_target_gb: 16
  vram_reduce_threshold_gb: 16
  vram_increase_threshold_gb: 10
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] YOLOv8n model loads and runs inference via TensorRT FP32
 - [ ] Face detection correctly identifies faces in test frames (mAP > 0.50 on WIDER Face val)
 - [ ] No faces detected in frames without faces (false positive rate < 5%)
 - [ ] Face crops are correctly extracted and resized to 224×224
 - [ ] Dynamic batching works correctly (multiple frames processed in single batch)
 - [ ] Batch size adjusts based on available VRAM
 - [ ] Frames with no detected faces are handled gracefully (no crashes, no crops generated)
 - [ ] NMS correctly removes duplicate detections (IoU > 0.45)
 - [ ] Face bounding boxes are accurate (center coordinates within 10 pixels of ground truth)
 ### Non-Functional
 - [ ] GPU memory usage stays ≤ 18GB per GPU during face detection
 - [ ] Face detection for 10 frames (batched) completes in < 2 seconds on Tesla P40
 - [ ] Batch processing throughput: ≥ 50 faces/second/GPU
 - [ ] Model loading time < 5 seconds at startup
 - [ ] No GPU OOM errors during extended processing
 ### Technical Constraints
 - [ ] TensorRT engine is FP32 only (no FP16, no INT8)
 - [ ] No Tensor Cores used (CC 5.2 constraint)
 - [ ] CUDA 11.8 compatible (verified via torch.version.cuda)
 - [ ] ONNX model exports without errors (verified via onnx.checker)
 - [ ] Face crops are saved in JPEG format at high quality
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation), STORY-03 (Frame Sampling — provides frames)
 - **Depends on:** None (runs after frame sampling)
 - **Enables:** STORY-05 (Classification — provides face crops)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | Tesla P40 (CC 5.2) has no Tensor Cores — slower inference | Use TensorRT engine caching; maximize batch size to compensate |
 | YOLOv8n may be too large for VRAM with classification | Use YOLOv8n (smallest variant); monitor VRAM closely |
 | Many faces per frame could cause VRAM spike | Limit max faces per frame; reduce batch size if needed |
 | TensorRT engine build time is long | Pre-build engines; cache on disk; skip rebuild if model unchanged |
 ## Estimated Effort
 - **Sprint:** 5-6 (second half)
 - **Story Points:** 21
 - **Dependencies:** STORY-01, STORY-03
@@ -1,187 +0,0 @@
 # STORY-05: Classification & Confidence Aggregation
 ## Epic
 **E1: Core Pipeline** — I can classify face crops and aggregate video confidence.
 ## Related Requirements
 | ID | Requirement |
 |----|------|
 | FR-03 | Binary demographic classification of detected face crops: Outputs probability p ∈ [0,1] for target class |
 | FR-04 | Video-level confidence aggregation & threshold routing: Aggregates frame-level scores → video confidence C; Routes to MATCH, REVIEW, or SKIP |
 | NFR-01 | Throughput: ≥ 30 videos/hour/GPU |
 | NFR-03 | GPU Memory Safety: ≤ 18GB per GPU sustained |
 | NFR-04 | Determinism & Reproducibility: Config-seeded randomness, versioned models |
 | TC-01 | Hardware: 2× Tesla P40 24GB (compute capability 5.2, PCIe 3.0, no Tensor Cores) |
 | TC-02 | CUDA/Torch Compatibility: CUDA ≤ 11.8, PyTorch ≤ 2.1.0, FP32 inference only |
 ## Description
 Implement demographic classification of detected face crops using a lightweight classifier (MobileNetV3 or equivalent). Apply temperature scaling for calibrated confidence scores. Aggregate frame-level confidence to video-level confidence using configurable aggregation logic. Route videos to MATCH, REVIEW, or SKIP based on threshold comparison.
 ## Scope
 ### In Scope
 - MobileNetV3 (or equivalent) classification model via TensorRT FP32
 - Temperature scaling for confidence calibration
 - Frame-level confidence computation (softmax with temperature)
 - Video-level confidence aggregation (max or weighted mean, configurable)
 - Threshold-based routing (T_high, T_low)
 - Routing decision assignment (MATCH, REVIEW, SKIP)
 - Confidence score persistence to DB and output files
 ### Out of Scope
 - Frame sampling (covered in STORY-03)
 - Face detection (covered in STORY-04)
 - Results export format (covered in STORY-06)
 - Review UI (covered in STORY-07)
 - Active learning / model retraining (covered in STORY-08)
 ## Deliverables
 ### 5.1 Classification Model
 **File:** `models/classifier/`
 Components:
 - **Source Model:** MobileNetV3-small (or equivalent lightweight classifier)
  - Trained on demographic dataset (fairness considerations documented)
  - Binary classification: Black male subject (class 1) vs. not (class 0)
 - **ONNX Export:** `classifier.onnx`
  - Input: `[1, 3, 224, 224]` (RGB, resized face crop)
  - Output: `[1, 2]` (logits for class 0 and class 1)
 - **TensorRT Engine:** `classifier.trt` (FP32)
  - Builder config: max_batch_size=32, max_workspace_size=2GB
 - **Calibration Data:** Temperature parameter T stored with model
 - **Model Metadata:** `model.json` with architecture, training dataset, validation metrics
 ### 5.2 Classification Runner
 **File:** `src/classifier.py`
 Core components:
 - **Model Loader:** Load TensorRT engine at startup
  ```python
  class FaceClassifier:
      def __init__(self, engine_path, temperature=1.0, device='cuda'):
          self.engine = load_trt_engine(engine_path)
          self.context = self.engine.create_execution_context()
          self.temperature = temperature
          self.input_shape = (1, 3, 224, 224)
  ```
 - **Preprocessing:** Convert face crop to model input
  - Resize to 224×224
  - Normalize (ImageNet statistics)
  - Convert to tensor (NCHW format)
 - **Inference:** Run classification on face crops
  - Dynamic batching (shared with face detection batcher)
  - Execute TensorRT engine
  - Get raw logits output
 - **Temperature Scaling:** Apply temperature to logits before softmax
  ```python
  def calibrated_softmax(logits, temperature):
      scaled_logits = logits / temperature
      return softmax(scaled_logits, axis=-1)
  ```
 - **Frame-level Confidence:** Extract probability for target class (class 1)
  ```python
  p_i = calibrated_probs[:, 1]  # probability of target class
  ```
 ### 5.3 Confidence Aggregation
 **File:** `src/aggregator.py`
 Features:
 - **Aggregation Strategies (configurable):**
  1. **Max:** `C = max(p_i)` — use highest confidence frame
  2. **Weighted Mean:** `C = softmax(α·mean(p_i) + β·var(p_i))`
  3. **Top-K Mean:** `C = mean(top_k(p_i))` — average of top K confidences
 - **Default Strategy:** Max (most conservative, aligns with FR-04)
 - **Parameter Configuration:** α, β, K configurable in config.yaml
 - **Variance Calculation:** Compute variance of frame-level confidences (for uncertainty estimation)
 ### 5.4 Threshold Routing
 **File:** `src/router.py`
 Features:
 - **Threshold Comparison:**
  ```python
  if C >= T_high:  # default 0.75
      routing = 'MATCH'
  elif C >= T_low:  # default 0.45
      routing = 'REVIEW'
  else:
      routing = 'SKIP'
  ```
 - **Configurable Thresholds:** T_high and T_low in config.yaml
 - **Routing Decision Logging:** Log routing decision with confidence score
 - **Edge Case Handling:**
  - No faces detected → routing = SKIP (with confidence = 0.0)
  - All frames have same confidence → routing based on threshold comparison
  - Confidence exactly at threshold → use >= comparison (inclusive)
 ### 5.5 Configuration Updates
 **File:** `config.yaml` (updates)
 New fields:
 ```yaml
 classifier:
  model: mobilenetv3-small
  model_path: /models/classifier/classifier.trt
  input_size: 224
  temperature: 1.0  # calibration temperature
  default_strategy: max  # max, weighted_mean, top_k_mean
 aggregation:
  strategy: max
  alpha: 1.0   # for weighted_mean
  beta: 0.1    # for weighted_mean
  top_k: 3     # for top_k_mean
 routing:
  T_high: 0.75
  T_low: 0.45
  no_faces_decision: SKIP
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] MobileNetV3 model loads and runs inference via TensorRT FP32
 - [ ] Classification outputs calibrated probability p ∈ [0,1] for target class
 - [ ] Temperature scaling is applied correctly (verified on test set)
 - [ ] Frame-level confidence scores are deterministic (same input → same output)
 - [ ] Max aggregation produces C = max(p_i) correctly
 - [ ] Weighted mean aggregation produces correct result with configurable α, β
 - [ ] Threshold routing assigns exactly one of: MATCH, REVIEW, SKIP
 - [ ] Videos with no detected faces are routed to SKIP with confidence 0.0
 - [ ] Confidence exactly at threshold uses >= comparison (inclusive)
 ### Non-Functional
 - [ ] Classification for 16 face crops (batched) completes in < 1 second on Tesla P40
 - [ ] GPU memory usage stays ≤ 18GB per GPU during classification
 - [ ] Aggregation computation is negligible (< 10ms per video)
 - [ ] Temperature parameter is stored with model checkpoint
 - [ ] All routing decisions are logged with confidence score and threshold values
 ### Technical Constraints
 - [ ] TensorRT engine is FP32 only
 - [ ] No Tensor Cores used (CC 5.2 constraint)
 - [ ] CUDA 11.8 compatible
 - [ ] ONNX model exports without errors
 - [ ] Confidence scores are reproducible (deterministic inference)
 - [ ] Aggregation strategy is configurable without code changes
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation), STORY-03 (Frame Sampling), STORY-04 (Face Detection — provides crops)
 - **Depends on:** None (runs after face detection)
 - **Enables:** STORY-06 (Results Persistence), STORY-07 (Review UI — provides routing decisions)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | Temperature scaling parameters need tuning | Evaluate on held-out calibration set during training cycle |
 | Max aggregation may be overly conservative | Offer weighted mean as alternative; allow per-job strategy selection |
 | Model bias concerns with demographic classification | Document training dataset; audit fairness metrics; legal review required |
 | TensorRT engine build time | Pre-build and cache engines; version control model files |
 ## Estimated Effort
 - **Sprint:** 5-6 (second half)
 - **Story Points:** 21
 - **Dependencies:** STORY-01, STORY-03, STORY-04
@@ -1,181 +0,0 @@
 # STORY-06: Results Persistence & Export
 ## Epic
 **E1: Core Pipeline** — As a system, I can persist processing results and export data for analytics.
 ## Related Requirements
 | ID | Requirement |
 |----|------|
 | FR-07 | Metadata logging & audit trail: Stores video ID, timestamps, frame counts, confidence scores, routing decision, model version |
 | FR-04 | Video-level confidence aggregation & threshold routing: Results saved to Parquet |
 | NFR-05 | Fault Tolerance: Metadata persisted before cleanup |
 | NFR-07 | Data Volume Handling: Efficient indexing for ~30TB dataset |
 | TC-03 | Storage I/O: Shared NAS/SMB for video input/output |
 ## Description
 Implement result persistence, metadata logging, and data export. Update the database with processing results, insert processing log entries, and export summary data in Parquet/JSONL format for analytics. Ensure all persistence happens before scratch space cleanup to prevent data loss.
 ## Scope
 ### In Scope
 - Database update for processed videos (status, confidence, routing, timestamps)
 - Processing log entry insertion (atomic transaction)
 - Parquet/JSONL export to `/data/output`
 - Confidence score persistence (frame-level and video-level)
 - Model version tracking in DB and export
 - Idempotent result persistence (re-processing doesn't duplicate entries)
 - Transaction-safe updates (all-or-nothing)
 ### Out of Scope
 - Frame sampling (covered in STORY-03)
 - Face detection (covered in STORY-04)
 - Classification (covered in STORY-05)
 - Review UI (covered in STORY-07)
 - Active learning pipeline (covered in STORY-08)
 ## Deliverables
 ### 6.1 Result Updater
 **File:** `src/result_updater.py`
 Features:
 - **Video Record Update:** Update `videos` table with processing results
  ```sql
  UPDATE videos SET
      status = 'COMPLETED',
      last_processed_time = NOW(),
      confidence_score = %s,
      routing_decision = %s,
      model_version = %s,
      frame_count = %s,
      updated_at = NOW()
  WHERE id = %s AND status = 'PROCESSING'
  ```
 - **Atomic Transaction:** Wrap video update + log insert in single transaction
 - **State Guard:** Only update if status is PROCESSING (prevents duplicate processing)
 - **Error Handling:** Rollback transaction on any error; re-queue video as PENDING
 ### 6.2 Processing Logger
 **File:** `src/processing_logger.py`
 Features:
 - **Log Entry Insertion:** Insert record into `processing_logs` table
  ```python
  def insert_processing_log(video_id, model_version, frame_count,
                            confidence_score, routing_decision):
      cursor.execute("""
          INSERT INTO processing_logs
              (video_id, model_version, frame_count, confidence_score,
               routing_decision, processed_at)
          VALUES (%s, %s, %s, %s, %s, NOW())
      """, (video_id, model_version, frame_count,
            confidence_score, routing_decision))
  ```
 - **Frame-Level Confidence Storage:** Store all frame-level confidence scores
  - Option A: JSON array in processing_logs.confidence_scores (TEXT column)
  - Option B: Separate `frame_confidences` table (for large datasets)
  - Default: JSON array in processing_logs (simpler, sufficient for typical frame counts)
 - **Audit Trail:** All log entries include timestamp, model version, and routing decision
 ### 6.3 Data Export
 **File:** `src/data_export.py`
 Features:
 - **Parquet Export:** Write summary data to Parquet format
  ```python
  # Schema:
  # video_id (int64), file_path (string), file_hash (string),
  # model_version (string), sample_count (int32),
  # confidence_scores (list<float>), video_confidence (float),
  # routing (string), processed_at (timestamp)
  ```
 - **JSONL Export:** Alternative line-delimited JSON format for streaming analytics
  ```json
  {"video_id": 12345, "file_path": "/data/input/video.mp4",
   "model_version": "v1.2.0", "sample_count": 8,
   "confidence_scores": [0.82, 0.79, 0.85, 0.76, 0.81, 0.83, 0.78, 0.80],
   "video_confidence": 0.85, "routing": "MATCH", "processed_at": "2026-08-03T10:30:00Z"}
  ```
 - **Output Directory:** `/data/output/{model_version}/` organized by model version
 - **File Naming:** `{model_version}_{batch_id}.parquet` and `{model_version}_{batch_id}.jsonl`
 - **Batch ID:** Sequential counter or timestamp-based (e.g., `v1.2.0_20260803_103000`)
 - **Compression:** Parquet with Snappy compression
 - **Export Trigger:** Export after all videos in batch are processed (or periodically)
 ### 6.4 Scratch Cleanup
 **File:** `src/scratch_manager.py` (cleanup section)
 Features:
 - **Post-Processing Cleanup:** Delete all frames and crops after results are persisted
 - **Cleanup Order:** Results → Logs → Export → Cleanup (critical ordering)
 - **Cleanup Verification:** Verify DB update before deleting scratch files
 - **Cleanup on Error:** If persistence fails, do NOT delete scratch (allow retry)
 - **Cleanup Logging:** Log cleanup completion with video_id and files removed
 ### 6.5 Configuration Updates
 **File:** `config.yaml` (updates)
 New fields:
 ```yaml
 results:
  persist_before_cleanup: true
  transaction_safe: true
  state_guard: true  # only update PROCESSING → COMPLETED
 export:
  format: parquet  # parquet, jsonl, or both
  output_path: /data/output
  compression: snappy
  batch_size: 100  # export after N videos
  include_frame_confidences: true
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] Video status transitions from PROCESSING to COMPLETED after successful processing
 - [ ] Processing log entry is inserted for every processed video
 - [ ] Frame-level confidence scores are stored (all p_i values)
 - [ ] Video-level confidence score is stored correctly
 - [ ] Routing decision is persisted accurately
 - [ ] Model version is recorded in both videos and processing_logs tables
 - [ ] Parquet export produces valid, readable files with correct schema
 - [ ] JSONL export produces valid JSON lines with correct fields
 - [ ] Export files are organized by model version in /data/output/
 - [ ] Scratch space is cleaned up after persistence (verified by directory check)
 - [ ] Re-processing a video updates existing records (no duplicates)
 - [ ] Transaction rollback on error prevents partial state updates
 ### Non-Functional
 - [ ] Result persistence completes in < 5 seconds per video
 - [ ] Parquet export for 100 videos completes in < 10 seconds
 - [ ] JSONL export for 100 videos completes in < 10 seconds
 - [ ] Export files are valid (verified by parquet.read_table and jsonl parsing)
 - [ ] Scratch cleanup frees > 95% of allocated space per video
 - [ ] No data loss if worker crashes between processing and persistence (PROCESSING state preserved)
 ### Technical Constraints
 - [ ] All DB updates use parameterized queries (no SQL injection)
 - [ ] Transaction isolation level is READ COMMITTED or higher
 - [ ] Parquet files use Snappy compression (not gzip, for speed)
 - [ ] JSONL files are UTF-8 encoded
 - [ ] Confidence scores are stored with float64 precision
 - [ ] Timestamps are in UTC (ISO 8601 format)
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation — DB schema), STORY-05 (Classification — provides results)
 - **Depends on:** None (runs after classification)
 - **Enables:** STORY-07 (Review UI — provides routing decisions), STORY-08 (Active Learning — provides training data)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | NAS write latency slows export | Batch writes; use local tmpfs for export staging |
 | Parquet library compatibility with CUDA container | Test pyarrow in Worker container early; pin version |
 | Transaction rollback leaves video in PROCESSING state | DB lock timeout (5 min) prevents permanent lock |
 | Scratch cleanup before persistence causes data loss | Enforce ordering: persist → verify → cleanup |
 ## Estimated Effort
 - **Sprint:** 5-6 (second half)
 - **Story Points:** 13
 - **Dependencies:** STORY-01, STORY-05
@@ -1,216 +0,0 @@
 # STORY-07: Review Interface
 ## Epic
 **E2: Routing & Review** — As an annotator, I can view low-confidence videos and label them.
 ## Related Requirements
 | ID | Requirement |
 |----|------|
 | FR-05 | Manual review interface for low-confidence videos: Displays video + contributing frames/crops + model confidence; Supports binary labeling |
 | FR-07 | Metadata logging & audit trail: Stores annotated labels for active learning |
 | NFR-06 | Observability: Tracks review queue depth |
 | TC-06 | Network Security: Internal LAN only; no auth required |
 ## Description
 Implement a lightweight web-based review interface for annotating low-confidence videos. Display the video player, top-k contributing frames, model confidence scores, and allow annotators to toggle the ground truth label. Support CSV/JSON export of annotated data for active learning.
 ## Scope
 ### In Scope
 - Lightweight web UI (Flask/FastAPI) serving on internal LAN
 - Query DB for videos with routing_decision = REVIEW
 - Video player with playback controls
 - Display top-k contributing frames (highest confidence frames)
 - Display model confidence scores per frame
 - Binary label toggle (True/False — target class present or not)
 - Label persistence to DB (review_queue table)
 - CSV/JSON export of annotated data with ground truth
 - Accessible via internal IP:Port (no auth, no SSL)
 ### Out of Scope
 - Frame sampling (covered in STORY-03)
 - Face detection (covered in STORY-04)
 - Classification (covered in STORY-05)
 - Confidence aggregation (covered in STORY-05)
 - Active learning pipeline / model retraining (covered in STORY-08)
 - Monitoring dashboards (covered in STORY-09)
 ## Deliverables
 ### 7.1 Review Backend
 **File:** `src/review_api.py`
 API Endpoints:
 - `GET /api/review/queue` — List videos in review queue
  - Query: `SELECT * FROM review_queue WHERE annotated = false ORDER BY created_at DESC`
  - Pagination: 20 items per page
  - Response: `{videos: [...], total: N, page: P, per_page: 20}`
 - `GET /api/review/{video_id}` — Get video details for annotation
  - Response: `{video_id, file_path, confidence_score, routing_decision, model_version, frame_count, contributing_frames: [{timestamp, crop_path, confidence}], video_duration}`
 - `POST /api/review/{video_id}/label` — Submit annotation
  - Body: `{ground_truth: true/false, notes: string (optional)}`
  - Updates: `review_queue.annotated = true`, `review_queue.ground_truth = value`, `review_queue.annotated_at = NOW()`
  - Response: `{status: 'annotated', video_id, ground_truth}`
 - `GET /api/review/export` — Export annotated data
  - Query params: `format=csv|json`, `annotated=true/false`, `date_from`, `date_to`
  - Response: File download with annotated data
 - `GET /api/review/stats` — Review queue statistics
  - Response: `{total_in_queue: N, annotated_today: N, avg_confidence: F, confidence_distribution: {...}}`
 ### 7.2 Review Frontend
 **File:** `ui/review/`
 Pages:
 - **Queue Page (`/`):** List of videos awaiting review
  - Table columns: Video ID, File Path, Confidence Score, Model Version, Date Added, Actions (View)
  - Sortable by confidence, date, file path
  - Filter by confidence range, model version
  - Pagination (20 items per page)
 - **Annotation Page (`/review/{video_id}`):** Video annotation interface
  - Video player with playback controls (HTML5 `<video>` element)
  - Top-k contributing frames displayed as thumbnails (k=5 default)
  - Confidence scores displayed per frame
  - Label toggle button (True/False) with confirmation
  - Optional notes field
  - Submit button (saves to DB via API)
  - Navigation: Previous/Next video in queue
 ### 7.3 Data Export
 **File:** `src/review_export.py`
 Features:
 - **CSV Export:**
  ```csv
  video_id,file_path,confidence_score,routing_decision,model_version,ground_truth,annotated_at,contributing_frames
  12345,/data/input/video.mp4,0.62,REVIEW,v1.2.0,true,2026-08-03T10:30:00Z,"[{'timestamp': 30.0, 'crop_path': '/scratch/12345/crops/30000.jpg', 'confidence': 0.82}, ...]"
  ```
 - **JSON Export:**
  ```json
  [
    {
      "video_id": 12345,
      "file_path": "/data/input/video.mp4",
      "confidence_score": 0.62,
      "routing_decision": "REVIEW",
      "model_version": "v1.2.0",
      "ground_truth": true,
      "annotated_at": "2026-08-03T10:30:00Z",
      "contributing_frames": [
        {"timestamp": 30.0, "crop_path": "/scratch/12345/crops/30000.jpg", "confidence": 0.82}
      ]
    }
  ]
  ```
 - **Export Options:**
  - Filter by annotation status (annotated/unannotated)
  - Filter by date range
  - Filter by model version
  - Filter by ground truth label
 - **Output Location:** `/data/output/reviews/`
 ### 7.4 UI Container
 **File:** `ui/Dockerfile`
 Base image: `python:3.10-slim`
 Installed packages:
 - Flask 3.0+ or FastAPI 0.100+ (lightweight web framework)
 - Jinja2 3.1+ (template engine)
 - PyMySQL (database connection for API)
 - gunicorn (WSGI server)
 ### 7.5 Docker Compose Update
 **File:** `docker-compose.yml` (update)
 Add UI service:
 ```yaml
 ui:
  build:
    context: ./ui
    dockerfile: Dockerfile
  ports:
    - "5000:5000"
  volumes:
    - ./ui:/app/ui
  environment:
    - DB_HOST=mariadb
    - DB_PORT=3306
    - DB_NAME=videodetect
    - DB_USER=videodetect
    - DB_PASSWORD=${DB_PASSWORD}
  depends_on:
    - mariadb
  networks:
    - videodetect-network
 ```
 ### 7.6 Database Update
 **File:** `db/schema.sql` (review_queue table — from STORY-01)
 The `review_queue` table was defined in STORY-01. This story populates and queries it.
 ### 7.7 Configuration Updates
 **File:** `config.yaml` (updates)
 New fields:
 ```yaml
 review_ui:
  host: "0.0.0.0"
  port: 5000
  per_page: 20
  top_k_frames: 5
  export_path: /data/output/reviews
  auth_enabled: false  # per TC-06
  ssl_enabled: false   # per TC-06
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] Review queue displays all videos with routing_decision = REVIEW and annotated = false
 - [ ] Video player loads and plays the video correctly
 - [ ] Top-k contributing frames are displayed as thumbnails with confidence scores
 - [ ] Annotator can toggle label (True/False) and submit
 - [ ] Submitted label is persisted to DB (review_queue table)
 - [ ] Annotated videos are removed from the default queue view
 - [ ] CSV export produces valid CSV with all required fields
 - [ ] JSON export produces valid JSON with all required fields
 - [ ] Export includes ground truth labels and contributing frame data
 - [ ] UI is accessible via http://<server-ip>:5000 (no auth, no SSL)
 - [ ] Pagination works correctly (20 items per page)
 - [ ] Sort and filter operations work on the queue page
 ### Non-Functional
 - [ ] Queue page loads in < 2 seconds (with 1000+ videos in queue)
 - [ ] Video player loads in < 3 seconds
 - [ ] Label submission completes in < 1 second
 - [ ] Export of 1000 annotated videos completes in < 10 seconds
 - [ ] UI uses < 100MB RAM at idle
 - [ ] No authentication or SSL configured (per TC-06)
 ### Technical Constraints
 - [ ] UI runs in Docker container (per TC-05)
 - [ ] No reverse proxy configured
 - [ ] No SSL certificates configured
 - [ ] No authentication mechanism configured
 - [ ] All API responses are JSON
 - [ ] Database queries use parameterized statements
 - [ ] Export files are UTF-8 encoded
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation — DB schema), STORY-05 (Classification — provides routing decisions)
 - **Depends on:** None (can be built in parallel with STORY-04, STORY-05)
 - **Enables:** STORY-08 (Active Learning — provides labeled training data)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | Video playback in browser requires compatible format | Serve videos in web-compatible format (H.264 MP4); transcode if needed |
 | Crop paths may not be accessible from UI container | Store crop paths in DB; serve via API endpoint |
 | No auth means anyone on LAN can access | Acceptable per TC-06; document in security notes |
 | Large review queue slows page loads | Implement server-side pagination; lazy load thumbnails |
 ## Estimated Effort
 - **Sprint:** 7
 - **Story Points:** 21
 - **Dependencies:** STORY-01, STORY-05
@@ -1,245 +0,0 @@
 # STORY-08: Active Learning Pipeline
 ## Epic
 **E3: Active Learning** — As an ML engineer, I can fine-tune the classifier head with reviewed data.
 ## Related Requirements
 | ID | Requirement |
 |----|------|
 | FR-06 | Active learning pipeline (label ingestion → fine-tuning → deployment): Batch retraining only. No online learning. Versioned model swaps. |
 | NFR-04 | Determinism & Reproducibility: Config-seeded randomness, versioned models |
 | NFR-05 | Fault Tolerance: Auto-retry on transient failures; skip & log on fatal errors |
 | TC-01 | Hardware: 2× Tesla P40 24GB (compute capability 5.2, PCIe 3.0, no Tensor Cores) |
 | TC-02 | CUDA/Torch Compatibility: CUDA ≤ 11.8, PyTorch ≤ 2.1.0, FP32 inference only |
 ## Description
 Implement the active learning pipeline that ingests annotated review data, fine-tunes the classifier head, validates the candidate model, and deploys it if it meets quality gates. This enables incremental model improvement using human-labeled data from the review queue.
 ## Scope
 ### In Scope
 - Label ingestion from review queue (CSV/JSON export or direct DB query)
 - Dataset versioning and preparation for training
 - Head-only fine-tuning on face crops (freeze backbone, train classification head)
 - Training with AdamW, configurable learning rate, early stopping
 - Candidate model validation on held-out validation set
 - F1 score and ECE (Expected Calibration Error) calculation
 - Quality gate validation (ΔF1 > 0.02, ECE < 0.08)
 - Model registry updates (ACTIVE, CANDIDATE, ARCHIVED states)
 - Hot reload of new TensorRT engine
 - Rollback on regression
 ### Out of Scope
 - Frame sampling (covered in STORY-03)
 - Face detection (covered in STORY-04)
 - Classification inference (covered in STORY-05)
 - Review UI (covered in STORY-07)
 - Monitoring dashboards (covered in STORY-09)
 ## Deliverables
 ### 8.1 Label Ingestion
 **File:** `src/active_learning/label_ingestor.py`
 Features:
 - **Data Source:** Query `review_queue` table for annotated labels
  ```sql
  SELECT video_id, ground_truth, contributing_frames
  FROM review_queue
  WHERE annotated = true AND ground_truth IS NOT NULL
  ```
 - **Crop Extraction:** Extract face crops from stored paths or re-extract from video
 - **Dataset Versioning:** Create versioned dataset directory structure
  ```
  /data/training/v2.0.0/
      crops/
          class_0/  (negative samples)
          class_1/  (positive samples)
      labels.csv
      metadata.json
  ```
 - **Train/Val Split:** 80/20 split (stratified by class)
 - **Augmentation:** Apply standard augmentations at training time (not pre-computed)
  - Random horizontal flip
  - Random color jitter
  - Random affine transform (±10 degrees rotation, ±10% scale)
 ### 8.2 Training Pipeline
 **File:** `src/active_learning/trainer.py`
 Features:
 - **Model Loading:** Load pre-trained MobileNetV3 backbone (frozen)
  ```python
  model = load_pretrained_mobilenetv3()
  for param in model.features.parameters():
      param.requires_grad = False
  # Replace classification head
  model.classifier = nn.Sequential(
      nn.Linear(1280, 256),
      nn.ReLU(),
      nn.Dropout(0.3),
      nn.Linear(256, 2)  # binary classification
  )
  ```
 - **Head-Only Fine-Tuning:** Only train the classification head
  - Backbone weights are frozen (no gradient updates)
  - Head weights are trainable
  - This is faster and requires less data than full fine-tuning
 - **Optimizer:** AdamW with configurable parameters
  - Learning rate: 1e-3 (default), configurable
  - Weight decay: 1e-2
  - Betas: (0.9, 0.999)
 - **Loss Function:** Binary Cross-Entropy with class weights (if imbalanced)
 - **Training Configuration:**
  - Epochs: 10-30 (configurable)
  - Batch size: 32
  - Early stopping: patience=5 epochs (no validation improvement)
  - Learning rate scheduler: ReduceLROnPlateau (factor=0.5, patience=3)
 - **Checkpointing:** Save checkpoint to `/models/candidate/` after each epoch
  ```
  /models/candidate/
      v2.0.0_epoch_01.pt
      v2.0.0_epoch_02.pt
      ...
      v2.0.0_best.pt  (best validation F1)
  ```
 ### 8.3 Validation Pipeline
 **File:** `src/active_learning/validator.py`
 Features:
 - **Validation Set:** Held-out 20% of annotated data (never seen during training)
 - **Metrics Calculation:**
  - **F1 Score:** Macro F1 on validation set
  - **ECE (Expected Calibration Error):**
    ```python
    def compute_ece(predictions, labels, n_bins=15):
        bin_boundaries = np.linspace(0, 1, n_bins + 1)
        ece = 0.0
        for i in range(n_bins):
            mask = (predictions >= bin_boundaries[i]) & (predictions < bin_boundaries[i+1])
            if mask.sum() > 0:
                bin_confidence = predictions[mask].mean()
                bin_accuracy = labels[mask].mean()
                ece += (mask.sum() / len(predictions)) * abs(bin_confidence - bin_accuracy)
        return ece
    ```
  - **Accuracy, Precision, Recall:** Standard classification metrics
 - **Quality Gates:**
  - ΔF1 > 0.02 (improvement over current model)
  - ECE < 0.08 (calibration acceptable)
  - Both gates must pass for deployment
 ### 8.4 Model Registry & Deployment
 **File:** `src/active_learning/registry.py`
 Features:
 - **Model Registry:** Update DB `models` table
  ```sql
  -- Promote candidate to active
  UPDATE models SET status = 'ACTIVE' WHERE version = 'v2.0.0';
  UPDATE models SET status = 'ARCHIVED' WHERE status = 'ACTIVE' AND version != 'v2.0.0';
  ```
 - **TensorRT Engine Build:** Convert candidate model to TensorRT engine
  ```bash
  trtexec --onnx=models/candidate/v2.0.0.onnx \
          --saveEngine=models/candidate/v2.0.0.trt \
          --fp32 --maxBatch=32 --workspace=2048
  ```
 - **Hot Reload:** Signal worker process to reload new engine
  - Option A: Restart worker container (`docker-compose restart worker`)
  - Option B: In-process reload (graceful, no downtime)
  - Default: In-process reload via signal handling
 - **Rollback:** If new model causes issues, rollback to archived model
  ```sql
  UPDATE models SET status = 'ACTIVE' WHERE version = 'v1.2.0';
  UPDATE models SET status = 'ARCHIVED' WHERE version = 'v2.0.0';
  ```
 ### 8.5 Training Configuration
 **File:** `config.yaml` (updates)
 New fields:
 ```yaml
 active_learning:
  enabled: true
  min_annotated_samples: 100  # minimum labeled data to trigger training
  training:
    epochs: 20
    batch_size: 32
    learning_rate: 1e-3
    weight_decay: 1e-2
    early_stopping_patience: 5
    lr_scheduler: ReduceLROnPlateau
    lr_factor: 0.5
    lr_patience: 3
  validation:
    val_split: 0.2
    min_f1_improvement: 0.02
    max_ece: 0.08
  deployment:
    auto_deploy: true  # deploy if quality gates pass
    hot_reload: true
    rollback_enabled: true
  augmentation:
    horizontal_flip: true
    color_jitter: true
    affine: true
    affine_degrees: 10
    affine_scale: 0.1
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] Annotated labels are ingested from review_queue table correctly
 - [ ] Dataset is versioned with proper directory structure
 - [ ] Train/val split is stratified by class (80/20)
 - [ ] Data augmentation is applied at training time (not pre-computed)
 - [ ] Backbone weights are frozen during fine-tuning (verified by checking requires_grad)
 - [ ] Classification head is trainable and receives gradient updates
 - [ ] AdamW optimizer is used with correct parameters
 - [ ] Early stopping works (training stops if no improvement for patience epochs)
 - [ ] Candidate model is saved to /models/candidate/ with correct versioning
 - [ ] F1 score is calculated correctly on validation set
 - [ ] ECE is calculated correctly on validation set
 - [ ] Quality gates are enforced (ΔF1 > 0.02 AND ECE < 0.08)
 - [ ] Model registry is updated (ACTIVE, CANDIDATE, ARCHIVED states)
 - [ ] TensorRT engine is built from candidate model
 - [ ] Hot reload deploys new model without downtime
 - [ ] Rollback restores previous model if deployment fails
 ### Non-Functional
 - [ ] Training completes in < 4 hours on Tesla P40 (typical dataset: 1000 samples)
 - [ ] Validation completes in < 30 minutes
 - [ ] TensorRT engine build completes in < 10 minutes
 - [ ] Hot reload completes in < 30 seconds
 - [ ] Training uses < 18GB GPU memory
 - [ ] Training is deterministic (same data + same config → same results)
 ### Technical Constraints
 - [ ] CUDA 11.8 compatible (verified via torch.version.cuda)
 - [ ] PyTorch ≤ 2.1.0 (verified via torch.__version__)
 - [ ] FP32 training only (no mixed precision)
 - [ ] No Tensor Cores used (CC 5.2 constraint)
 - [ ] Model versioning follows semantic versioning (MAJOR.MINOR.PATCH)
 - [ ] All training hyperparameters are configurable via config.yaml
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation), STORY-05 (Classification — provides base model), STORY-07 (Review UI — provides labeled data)
 - **Depends on:** None (runs independently, triggered by labeled data threshold)
 - **Enables:** STORY-05 (provides new model for inference)
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | Insufficient labeled data for meaningful fine-tuning | Set min_annotated_samples threshold (e.g., 100); wait until reached |
 | Head-only fine-tuning may not be enough for domain shift | Offer full fine-tuning as option; document limitations |
 | Model regression in production | Strict quality gates; keep previous model in ARCHIVED state for quick rollback |
 | Long training times on Tesla P40 | Head-only training is faster; batch size tuning; early stopping |
 | ECE calculation sensitivity | Use proper binning; report ECI with confidence intervals |
 ## Estimated Effort
 - **Sprint:** 7-8
 - **Story Points:** 34
 - **Dependencies:** STORY-01, STORY-05, STORY-07
@@ -1,360 +0,0 @@
 # STORY-09: Observability, Monitoring & Hardening
 ## Epic
 **E4: Operations** — As a DevOps engineer, I can schedule, monitor, and resume batch jobs.
 ## Related Requirements
 | ID | Requirement |
 |----|------|
 | NFR-05 | Fault Tolerance: Auto-retry on transient failures; skip & log on fatal errors |
 | NFR-06 | Observability: Prometheus/Grafana metrics + structured logging; Tracks FPS, queue depth, confidence distribution, drift alerts |
 | NFR-01 | Throughput: ≥ 30 videos/hour/GPU |
 | NFR-03 | GPU Memory Safety: ≤ 18GB per GPU sustained |
 | NFR-04 | Determinism & Reproducibility: Config-seeded randomness, versioned models |
 | NFR-07 | Data Volume Handling: Efficient indexing for ~30TB dataset |
 ## Description
 Implement comprehensive monitoring, metrics collection, fault tolerance, and system hardening. Expose Prometheus metrics for throughput, GPU utilization, queue depth, and confidence distribution. Build Grafana dashboards for real-time system health. Implement crash recovery, idempotency, retry logic, and drift detection.
 ## Scope
 ### In Scope
 - Prometheus metrics exposure (videos processed, GPU utilization, queue depth, confidence distribution)
 - Grafana dashboard configuration (throughput, error rates, confidence drift)
 - Crash recovery with checkpointing
 - Idempotent processing guarantees
 - Retry logic for transient errors (up to 3 attempts)
 - Weekly drift detection job
 - Alerting on confidence distribution shifts and review queue growth
 - Worker health checks and auto-restart
 ### Out of Scope
 - Frame sampling (covered in STORY-04)
 - Face detection (covered in STORY-05)
 - Classification (covered in STORY-06)
 - Review UI (covered in STORY-08)
 - Active learning pipeline (covered in STORY-09)
 ## Deliverables
 ### 9.1 Prometheus Metrics
 **File:** `src/metrics.py`
 Metrics to expose:
 - **Counter Metrics:**
  - `videos_processed_total` (label: routing_decision=MATCH|REVIEW|SKIP|UNSCANNABLE|ERROR)
  - `videos_processed_by_model_total` (label: model_version)
  - `frames_extracted_total`
  - `faces_detected_total`
  - `inference_errors_total` (label: error_type)
  - `retry_attempts_total` (label: step=probe|extract|detect|classify)
 - **Gauge Metrics:**
  - `gpu_utilization_percent` (label: gpu_id)
  - `gpu_memory_used_bytes` (label: gpu_id)
  - `gpu_memory_free_bytes` (label: gpu_id)
  - `queue_depth_pending`
  - `queue_depth_processing`
  - `queue_depth_review`
  - `scratch_usage_bytes`
  - `scratch_usage_percent`
 - **Histogram Metrics:**
  - `video_processing_duration_seconds` (label: routing_decision)
  - `confidence_score_distribution` (buckets: 0.0, 0.1, 0.2, ..., 0.9, 1.0)
  - `frame_count_per_video`
  - `face_count_per_video`
 - **Summary Metrics:**
  - `throughput_videos_per_hour` (calculated from counter)
  - `average_confidence` (calculated from histogram)
 Implementation:
 ```python
 from prometheus_client import start_http_server, Counter, Gauge, Histogram
 # Start metrics HTTP server
 start_http_server(9090)  # /metrics endpoint
 # Define metrics
 videos_processed = Counter(
    'videos_processed_total',
    'Total videos processed',
    ['routing_decision']
 )
 gpu_memory = Gauge(
    'gpu_memory_used_bytes',
    'GPU memory usage',
    ['gpu_id']
 )
 confidence_hist = Histogram(
    'confidence_score_distribution',
    'Video confidence scores',
    buckets=[0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0]
 )
 ```
 ### 9.2 Grafana Dashboards
 **File:** `monitoring/grafana/dashboards/`
 Dashboard 1: **System Overview**
 - Panels:
  - Throughput (videos/hour) — line chart, 1h window
  - Queue depth (pending/processing/review) — stacked bar
  - GPU utilization (both GPUs) — line chart
  - GPU memory usage (both GPUs) — line chart
  - Error rate (errors/100 videos) — bar chart
  - Confidence distribution — histogram
 Dashboard 2: **Processing Details**
 - Panels:
  - Processing duration per video — scatter plot
  - Frame count per video — histogram
  - Face count per video — histogram
  - Confidence score by model version — box plot
  - Routing decision distribution — pie chart
 Dashboard 3: **Active Learning**
 - Panels:
  - Review queue size over time — line chart
  - Annotation rate (labels/day) — bar chart
  - Model version timeline — timeline panel
  - F1 score by model version — line chart
  - ECE by model version — line chart
 **File:** `monitoring/grafana/dashboards/system_overview.json`
 **File:** `monitoring/grafana/dashboards/processing_details.json`
 **File:** `monitoring/grafana/dashboards/active_learning.json`
 ### 9.3 Crash Recovery
 **File:** `src/crash_recovery.py`
 Features:
 - **Checkpointing:** Periodically save processing state
  ```python
  def save_checkpoint(video_id, state, progress):
      checkpoint = {
          'video_id': video_id,
          'state': state,  # 'processing', 'extracting', 'detecting', 'classifying'
          'progress': progress,  # dict of step -> completed
          'timestamp': datetime.utcnow().isoformat()
      }
      with open(f'/scratch/checkpoints/{video_id}.json', 'w') as f:
          json.dump(checkpoint, f)
  ```
 - **Recovery on Startup:** Scan for PROCESSING videos and re-queue them
  ```sql
  UPDATE videos SET status = 'PENDING', updated_at = NOW()
  WHERE status = 'PROCESSING' AND updated_at < NOW() - INTERVAL 5 MINUTE
  ```
 - **Lock Timeout:** 5 minutes (videos stuck in PROCESSING beyond this are re-queued)
 - **Re-queue Logic:** Only re-queue if worker is down (detected via health check)
 - **Idempotency:** Re-processing a video produces same results (no duplicates)
 ### 9.4 Retry Logic
 **File:** `src/retry.py`
 Features:
 - **Retry Decorator:**
  ```python
  def retry(max_attempts=3, delay=1.0, backoff=2.0, exceptions=(Exception,)):
      def decorator(func):
          def wrapper(*args, **kwargs):
              last_exception = None
              for attempt in range(max_attempts):
                  try:
                      return func(*args, **kwargs)
                  except exceptions as e:
                      last_exception = e
                      if attempt < max_attempts - 1:
                          wait = delay * (backoff ** attempt)
                          time.sleep(wait)
              raise last_exception
          return wrapper
      return decorator
  ```
 - **Applicable Steps:** ffprobe, frame extraction, face detection, classification, DB writes
 - **Transient Errors:** Network timeout, GPU OOM, file lock contention
 - **Non-Retryable Errors:** Codec unsupported, corrupt file, invalid path → skip and log
 - **Retry Logging:** Log each retry attempt with error type and delay
 ### 9.5 Drift Detection
 **File:** `src/drift_detector.py`
 Features:
 - **Weekly Job:** Compare current confidence distribution to baseline
  ```python
  def detect_drift(current_confidences, baseline_confidences, threshold=0.10):
      # Compare p(C > 0.5) shift
      current_high = sum(1 for c in current_confidences if c > 0.5) / len(current_confidences)
      baseline_high = sum(1 for c in baseline_confidences if c > 0.5) / len(baseline_confidences)
      shift = abs(current_high - baseline_high)
      if shift > threshold:
          alert(f"Confidence drift detected: {shift:.2%} shift in p(C > 0.5)")
          return True
      return False
  ```
 - **Baseline:** Stored in DB or config (computed from last training cycle)
 - **Alert Conditions:**
  - `p(C > 0.5)` shifts > 10% from baseline
  - Review queue grows unbounded (> 1000 items for > 24 hours)
  - Throughput drops below 20 videos/hour/GPU for > 1 hour
  - Error rate exceeds 5% for any 1-hour window
 - **Alert Channels:** Email, Slack webhook, or log entry (configurable)
 ### 9.6 Worker Health Checks
 **File:** `src/health_check.py`
 Features:
 - **Health Endpoint:** `/health` returns worker status
  ```json
  {
      "status": "healthy",
      "gpu_available": true,
      "gpu_memory_used_gb": 12.5,
      "queue_depth": 42,
      "uptime_seconds": 86400,
      "videos_processed_today": 156,
      "last_error": null
  }
  ```
 - **Auto-Restart:** Docker restart policy for worker container
  ```yaml
  restart: unless-stopped
  healthcheck:
    test: ["CMD", "curl", "-f", "http://localhost:9090/health"]
    interval: 30s
    timeout: 10s
    retries: 3
  ```
 - **GPU Health:** Monitor GPU temperature and error counts
 - **Disk Health:** Monitor scratch space usage and NAS connectivity
 ### 9.7 Docker Compose Update
 **File:** `docker-compose.yml` (update)
 Add monitoring services:
 ```yaml
 prometheus:
  image: prom/prometheus:v2.48.0
  volumes:
    - ./monitoring/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml
    - prometheus_data:/prometheus
  ports:
    - "9090:9090"
  networks:
    - videodetect-network
 grafana:
  image: grafana/grafana:10.2.0
  volumes:
    - grafana_data:/var/lib/grafana
    - ./monitoring/grafana/dashboards:/etc/grafana/provisioning/dashboards
  environment:
    - GF_SECURITY_ADMIN_PASSWORD=${GRAFANA_PASSWORD}
  ports:
    - "3000:3000"
  networks:
    - videodetect-network
 volumes:
  prometheus_data:
  grafana_data:
 ```
 ### 9.8 Configuration Updates
 **File:** `config.yaml` (updates)
 New fields:
 ```yaml
 monitoring:
  prometheus:
    enabled: true
    port: 9090
    metrics_path: /metrics
  grafana:
    enabled: true
    port: 3000
  alerts:
    confidence_drift_threshold: 0.10
    review_queue_max_size: 1000
    review_queue_max_age_hours: 24
    throughput_min_videos_per_hour: 20
    throughput_min_duration_hours: 1
    error_rate_threshold: 0.05
    error_rate_window_hours: 1
  drift_detection:
    enabled: true
    schedule: weekly  # cron: 0 2 * * 0 (Sundays at 2 AM)
    baseline_source: db  # db or config
  crash_recovery:
    lock_timeout_minutes: 5
    auto_requeue: true
  retry:
    max_attempts: 3
    initial_delay: 1.0
    backoff_factor: 2.0
    retryable_errors:
      - timeout
      - gpu_oom
      - file_lock
    non_retryable_errors:
      - codec_unsupported
      - file_corrupt
      - invalid_path
 ```
 ## Acceptance Criteria
 ### Functional
 - [ ] Prometheus metrics are exposed at `/metrics` endpoint and queryable
 - [ ] All required metrics are present (counters, gauges, histograms, summaries)
 - [ ] Grafana dashboards load and display correct data
 - [ ] System Overview dashboard shows throughput, queue depth, GPU metrics
 - [ ] Processing Details dashboard shows duration, frame count, confidence distribution
 - [ ] Active Learning dashboard shows review queue, annotation rate, model versions
 - [ ] Crash recovery re-queues stuck PROCESSING videos on worker restart
 - [ ] Idempotent processing: re-processing produces same results (no duplicates)
 - [ ] Retry logic retries transient errors up to 3 times with exponential backoff
 - [ ] Non-retryable errors are skipped and logged (no infinite retry)
 - [ ] Drift detection runs weekly and alerts on > 10% confidence shift
 - [ ] Review queue growth alert triggers when queue > 1000 for > 24 hours
 - [ ] Worker health check returns correct status
 - [ ] Auto-restart triggers on health check failure
 ### Non-Functional
 - [ ] Metrics collection overhead < 2% of CPU
 - [ ] Grafana dashboard loads in < 3 seconds
 - [ ] Crash recovery completes in < 30 seconds
 - [ ] Drift detection completes in < 5 minutes
 - [ ] Alert delivery completes in < 60 seconds
 - [ ] Prometheus data retention: 30 days (configurable)
 - [ ] Grafana data source refresh: 30 seconds
 ### Technical Constraints
 - [ ] Prometheus metrics follow naming conventions (unit suffixes, proper labels)
 - [ ] Grafana dashboards are JSON-exportable and version-controlled
 - [ ] Crash recovery is idempotent (running twice produces same result)
 - [ ] Retry logic does not retry non-transient errors
 - [ ] Drift detection baseline is stored and versioned
 - [ ] Health check endpoint responds in < 1 second
 - [ ] All alerts are logged with timestamp and context
 ## Dependencies
 - **Prerequisites:** STORY-01 (Foundation), STORY-05 (Classification — provides metrics data), STORY-08 (Active Learning — provides model metrics)
 - **Depends on:** None (can be built in parallel with other stories)
 - **Enables:** Production deployment and long-term operation
 ## Risks & Mitigations
 | Risk | Mitigation |
 |------|--|
 | Prometheus metrics cardinality explosion | Limit label cardinality; use histograms instead of individual values |
 | Grafana dashboard load time with large datasets | Use Prometheus aggregations; pre-compute panels |
 | Crash recovery misses in-flight writes | Use DB transactions; lock timeout prevents permanent locks |
 | Drift detection baseline becomes stale | Update baseline with each training cycle |
 | Alert fatigue from too many alerts | Tune thresholds; implement alert grouping |
 ## Estimated Effort
 - **Sprint:** 9+
 - **Story Points:** 34
 - **Dependencies:** STORY-01, STORY-05, STORY-08
@@ -0,0 +1,27 @@
 FROM ubuntu:22.04
 # Install Perl and system dependencies
 RUN apt update && apt install -y \
    dnsutils \
    libdancer2-perl \
    libdancer2-plugin-database-perl \
    libdancer-plugin-database-core-perl \
    libdbd-mysql-perl \
    libhttp-lite-perl \
    libjson-perl \
    libspreadsheet-parsexlsx-perl \
    libsql-splitstatement-perl \
    libyaml-perl \
    perl \
    libdbi-perl \
    build-essential \
    ffmpeg
 EXPOSE 3000
 RUN mkdir -p /app
 WORKDIR /app
 COPY . /app
 # Simple placeholder command
 CMD ["perl", "app.pl"]
@@ -0,0 +1,157 @@
 #!/usr/bin/perl
 use strict;
 use Dancer2;
 use Dancer2::Plugin::Database;
 use Data::Dumper;
 use SQL::SplitStatement;
 use HTTP::Lite;
 $SIG{'INT'} = sub { exit; };
 hook before => sub {
    my $origin = request_header('Origin') || 'http://localhost:8890';
    my $method = request_header('Access-Control-Request-Method') || '';
    my $headers = request_header('Access-Control-Request-Headers') || '';
    response_header 'Access-Control-Allow-Origin' => $origin;
    response_header 'Access-Control-Allow-Methods' => $method;
    response_header 'Access-Control-Allow-Headers' => $headers;
 };
 options qr{/api/v1/.+} => sub {
    return {};
 };
 get '/api/v1/hello' => sub {
    my $user = session('user') || 'world';
    return {'hello' => $user };
 };
 get '/api/v1/videos' => sub {
    my $sth = database->prepare("SELECT id, file_path FROM videos ORDER BY id");
    $sth->execute();
    my $ref = $sth->fetchall_hashref('id');
    $sth->finish();
    return $ref;
 };
 get '/api/v1/video' => sub {
    my $path = query_parameters->get("path");
    my $sth = database->prepare("SELECT * FROM videos WHERE file_path=?");
    $sth->execute($path);
    my $ref = $sth->fetchrow_hashref();
    $sth->finish();
    unless(ref $ref) {
        send_error("No video", 404);
    }
    return $ref;
 };
 get '/api/v1/video/:video' => sub {
    my $video = route_parameters->get("video");
    my $sth = database->prepare("SELECT * FROM videos WHERE id=?");
    $sth->execute($video);
    my $ref = $sth->fetchrow_hashref();
    $sth->finish();
    unless(ref $ref) {
        send_error("No video", 404);
    }
    return $ref;
 };
 post '/api/v1/video' => sub {
    my $file_path = body_parameters->get("file_path");
    my $file_hash = body_parameters->get("file_hash") || undef;
    my $resolution_w = body_parameters->get("resolution_w");
    my $resolution_h = body_parameters->get("resolution_h");
    my $codec = body_parameters->get("codec");
    my $sth = database->prepare("INSERT INTO videos (file_path, file_hash, resolution_w, resolution_h, codec) VALUES (?, ?, ?, ?, ?)");
    $sth->execute($file_path, $file_hash, $resolution_w, $resolution_h, $codec);
    my $video_id = database->last_insert_id(undef, undef, 'videos', undef);
    $sth->finish();
    return { id => $video_id };
 };
 get '/api/v1/tasks' => sub {
    my $status = query_parameters->get("status") || 'PENDING';
    my $sth = database->prepare("SELECT id, video_id, status FROM tasks WHERE status=? ORDER BY id");
    $sth->execute($status);
    my $ref = $sth->fetchall_hashref('id');
    $sth->finish();
    return $ref;
 };
 get '/api/v1/task/:task' => sub {
    my $task = route_parameters->get("task");
    my $sth = database->prepare("SELECT * FROM tasks WHERE id=?");
    $sth->execute($task);
    my $ref = $sth->fetchrow_hashref();
    $sth->finish();
    unless(ref $ref) {
        send_error("No task", 404);
    }
    return $ref;
 };
 get '/api/v1/nexttask/:type' => sub {
    my $type = route_parameters->get("type");
    my $sth = database->prepare("SELECT * FROM tasks WHERE task_type=? AND status='PENDING' ORDER BY id LIMIT 1");
    $sth->execute($type);
    my $ref = $sth->fetchrow_hashref();
    $sth->finish();
    unless(ref $ref) {
        send_error("No task", 404);
    }
    # Mark the task as in progress and assign it to a worker
    my $assign_key = "worker_" . int(rand(1000)); # Example assign key, you can customize this
    my $update_sth = database->prepare("UPDATE tasks SET status='IN_PROGRESS', assign_key=?, assigned_at=NOW() WHERE id=? AND status='PENDING'");
    $update_sth->execute($assign_key, $ref->{id});
    if($update_sth->rows == 0) {
        send_error("Failed to claim task", 409);
    }
    $update_sth->finish();
    return { task => $ref, assign_key => $assign_key };
 };
 post '/api/v1/task/:task/complete' => sub {
    my $task = route_parameters->get("task");
    my $assign_key = body_parameters->get("assign_key");
    my $results = body_parameters->get("results");
    # Verify the task is assigned to the worker
    my $sth = database->prepare("SELECT * FROM tasks WHERE id=? AND assign_key=? AND status='IN_PROGRESS'");
    $sth->execute($task, $assign_key);
    my $ref = $sth->fetchrow_hashref();
    $sth->finish();
    unless(ref $ref) {
        send_error("Task not assigned to this worker or not in progress", 403);
    }
    # Update the task as completed
    my $update_sth = database->prepare("UPDATE tasks SET status='COMPLETED', results=?, updated_at=NOW() WHERE id=?");
    $update_sth->execute($results, $task);
    if($update_sth->rows == 0) {
        send_error("Failed to complete task", 500);
    }
    $update_sth->finish();
    return { message => "Task completed successfully" };
 };
 start();
@@ -0,0 +1,39 @@
 serializer: JSON
 logger: console
 engines:
    serializer:
        JSON:
            allow_nonref: 1
            allow_blessed: 1
            pretty: 1
 plugins:
    Database:
        driver: 'mysql'
        database: 'videodetect'
        host: 'mariadb'
        port: 3306
        username: 'videodetect'
        password: 'changeme_videodetect'
        connection_check_threshold: 10
        dbi_params:
            RaiseError: 1
            AutoCommit: 1
        on_connect_do: ["SET NAMES 'utf8'", "SET CHARACTER SET 'utf8'" ]
        log_queries: 1
 server:
  port: 5000
  workers: 4
  bind: 0.0.0.0
 logging:
  level: info
  logdir: /var/log/videodetect/api
  logfile: videodetect-api.log
  format: "[%d] %l [%P] %m"
 paths:
  scratch: /scratch
  input: /data/input
  output: /data/output
  models: /models
@@ -18,119 +18,35 @@ USE videodetect;
 CREATE TABLE IF NOT EXISTS videos (
    id BIGINT AUTO_INCREMENT PRIMARY KEY,
    file_path VARCHAR(2048) NOT NULL,
-    file_hash CHAR(64) NOT NULL COMMENT 'SHA-256 hash of file',
+    file_size BIGINT NOT NULL COMMENT 'Size of the file in bytes',
    file_hash CHAR(64) COMMENT 'SHA-256 hash of file',
    resolution_w INT DEFAULT NULL,
    resolution_h INT DEFAULT NULL,
    codec VARCHAR(50) DEFAULT NULL,
    duration FLOAT DEFAULT NULL COMMENT 'Duration in seconds',
    status ENUM('NEW', 'PENDING', 'PROCESSING', 'COMPLETED', 'UNSCANNABLE', 'ERROR')
        NOT NULL DEFAULT 'NEW' COMMENT 'Processing status',
    last_scan_time DATETIME DEFAULT NULL,
    last_processed_time DATETIME DEFAULT NULL,
    confidence_score FLOAT DEFAULT NULL COMMENT 'Video-level confidence score',
    routing_decision ENUM('MATCH', 'REVIEW', 'SKIP') DEFAULT NULL COMMENT 'Routing decision',
    model_version VARCHAR(50) DEFAULT NULL COMMENT 'Model version used for processing',
    frame_count INT DEFAULT NULL COMMENT 'Number of frames sampled',
    error_message TEXT DEFAULT NULL,
    created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
    updated_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
-    UNIQUE KEY uk_file_hash (file_hash),
+    UNIQUE KEY uk_path (file_path),
    INDEX idx_videos_status (status),
    INDEX idx_videos_last_scan (last_scan_time),
    INDEX idx_videos_status_last_scan (status, last_scan_time),
    INDEX idx_videos_file_path (file_path(255))
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
-- -----------------------------------------------------
+
-- Table: processing_logs
+CREATE TABLE IF NOT EXISTS tasks (
 -- Audit trail for each video processing job
 -- -----------------------------------------------------
 CREATE TABLE IF NOT EXISTS processing_logs (
    id BIGINT AUTO_INCREMENT PRIMARY KEY,
    task_type VARCHAR(16) NOT NULL,
    video_id BIGINT NOT NULL,
-    model_version VARCHAR(50) NOT NULL,
+    status ENUM('PENDING', 'IN_PROGRESS', 'COMPLETED', 'FAILED') NOT NULL DEFAULT 'PENDING',
    frame_count INT NOT NULL DEFAULT 0,
    confidence_score FLOAT DEFAULT NULL,
    confidence_scores JSON DEFAULT NULL COMMENT 'Frame-level confidence scores',
    routing_decision ENUM('MATCH', 'REVIEW', 'SKIP') NOT NULL,
    processed_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
    error_message TEXT DEFAULT NULL,
    FOREIGN KEY (video_id) REFERENCES videos(id) ON DELETE CASCADE,
    INDEX idx_processing_logs_video (video_id),
    INDEX idx_processing_logs_processed_at (processed_at),
    INDEX idx_processing_logs_routing (routing_decision)
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
 -- -----------------------------------------------------
 -- Table: models
 -- Model registry tracking all deployed and archived models
 -- -----------------------------------------------------
 CREATE TABLE IF NOT EXISTS models (
    version VARCHAR(50) PRIMARY KEY,
    status ENUM('ACTIVE', 'CANDIDATE', 'ARCHIVED') NOT NULL DEFAULT 'CANDIDATE',
    path VARCHAR(2048) NOT NULL,
    calibration_temp FLOAT DEFAULT NULL COMMENT 'Temperature scaling parameter',
    f1_score FLOAT DEFAULT NULL COMMENT 'F1 score on validation set',
    ece_score FLOAT DEFAULT NULL COMMENT 'Expected Calibration Error',
    deployed_at DATETIME DEFAULT NULL,
    created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
    INDEX idx_models_status (status)
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
 -- -----------------------------------------------------
 -- Table: review_queue
 -- Videos awaiting manual annotation
 -- -----------------------------------------------------
 CREATE TABLE IF NOT EXISTS review_queue (
    id BIGINT AUTO_INCREMENT PRIMARY KEY,
    video_id BIGINT NOT NULL,
    confidence_score FLOAT NOT NULL,
    routing_decision ENUM('REVIEW') NOT NULL DEFAULT 'REVIEW',
    annotated BOOLEAN NOT NULL DEFAULT FALSE,
    ground_truth BOOLEAN DEFAULT NULL COMMENT 'True label from annotator',
    annotated_at DATETIME DEFAULT NULL,
    notes TEXT DEFAULT NULL,
    created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
    updated_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
    assign_key VARCHAR(64) DEFAULT NULL COMMENT 'Key to identify the worker assigned to this task',
    assigned_at DATETIME DEFAULT NULL COMMENT 'Timestamp when the task was assigned to a worker',
    results JSON DEFAULT NULL COMMENT 'Task-specific results or metadata',
    FOREIGN KEY (video_id) REFERENCES videos(id) ON DELETE CASCADE,
-    INDEX idx_review_queue_annotated (annotated),
+    INDEX idx_tasks_video (video_id),
-    INDEX idx_review_queue_created (created_at)
+    INDEX idx_tasks_status (status),
    INDEX idx_tasks_type_status (task_type, status),
    UNIQUE KEY uk_tasks_video_type (video_id, task_type)
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
 -- -----------------------------------------------------
 -- Table: scan_history
 -- Tracks directory scan operations for incremental sync
 -- -----------------------------------------------------
 CREATE TABLE IF NOT EXISTS scan_history (
    id BIGINT AUTO_INCREMENT PRIMARY KEY,
    scan_start DATETIME NOT NULL,
    scan_end DATETIME DEFAULT NULL,
    files_discovered INT DEFAULT 0,
    files_new INT DEFAULT 0,
    files_modified INT DEFAULT 0,
    files_removed INT DEFAULT 0,
    files_unscannable INT DEFAULT 0,
    duration_seconds FLOAT DEFAULT NULL,
    status ENUM('RUNNING', 'COMPLETED', 'FAILED') NOT NULL DEFAULT 'RUNNING',
    error_message TEXT DEFAULT NULL
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
 -- -----------------------------------------------------
 -- Table: scanner_lock
 -- Single-instance guard for the directory scanner. Ensures only one scan
 -- runs at a time across all replicas. A lease (locked_at + lease_seconds)
 -- lets a live scanner keep the lock via heartbeats, and lets a crashed
 -- scanner's lock be taken over once it goes stale.
 -- -----------------------------------------------------
 CREATE TABLE IF NOT EXISTS scanner_lock (
    lock_name VARCHAR(64) PRIMARY KEY,
    owner VARCHAR(128) NOT NULL COMMENT 'Instance id (host-pid-uuid) holding the lock',
    locked_at DATETIME NOT NULL COMMENT 'Last time the lock was acquired or heartbeated',
    lease_seconds INT NOT NULL DEFAULT 21600 COMMENT 'Lock is stale if older than this (6 hours)'
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
 -- -----------------------------------------------------
 -- Insert default model entry
 -- -----------------------------------------------------
 INSERT IGNORE INTO models (version, status, path, calibration_temp, created_at)
 VALUES ('v0.0.0-placeholder', 'ACTIVE', '/models/placeholder', 1.0, NOW());
@@ -5,8 +5,8 @@ services:
    restart: unless-stopped
    environment:
      MYSQL_ROOT_PASSWORD: ${DB_ROOT_PASSWORD:-rootpass}
-      MYSQL_DATABASE: videodetect
+      MYSQL_DATABASE: ${DB_NAME:-videodetect}
-      MYSQL_USER: videodetect
+      MYSQL_USER: ${DB_USER:-videodetect}
      MYSQL_PASSWORD: ${DB_PASSWORD:-videodetect123}
    ports:
      - "3306:3306"
@@ -25,50 +25,6 @@ services:
        limits:
          memory: 2G
  worker:
    build:
      context: .
      dockerfile: worker/Dockerfile
    container_name: videodetect-worker
    restart: unless-stopped
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=compute,utility
      - DB_HOST=mariadb
      - DB_PORT=3306
      - DB_NAME=videodetect
      - DB_USER=videodetect
      - DB_PASSWORD=${DB_PASSWORD:-videodetect123}
      - CONFIG_PATH=/app/config.yaml
    volumes:
      - /dev/null:/dev/null  # tmpfs mounted at /scratch in container
      - nas_input:/data/input:ro
      - ${NAS_OUTPUT_PATH:-./output}:/data/output
      - ${MODELS_PATH:-./models}:/models
      - ${TRAINING_PATH:-./training}:/data/training
    tmpfs:
      - /scratch:noexec,nosuid,size=100G
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: ["gpu"]
        limits:
          memory: 24G
    networks:
      - videodetect-network
    depends_on:
      mariadb:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "python3", "-c", "import torch; print(torch.cuda.is_available())"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 60s
  ui:
    build:
      context: ./ui
@@ -98,50 +54,32 @@ services:
        limits:
          memory: 512M
-  prometheus:
+  api:
-    image: prom/prometheus:v2.48.0
+    build:
-    container_name: videodetect-prometheus
+      context: ./api
-    restart: unless-stopped
+      dockerfile: Dockerfile
-    ports:
+    container_name: videodetect-api
-      - "9090:9090"
+    restart: no
    volumes:
      - ./monitoring/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
      - prometheus_data:/prometheus
    networks:
      - videodetect-network
    deploy:
      resources:
        limits:
          memory: 1G
  grafana:
    image: grafana/grafana:10.2.0
    container_name: videodetect-grafana
    restart: unless-stopped
    ports:
      - "3000:3000"
    environment:
-      - GF_SECURITY_ADMIN_PASSWORD=${GRAFANA_PASSWORD:-admin}
+      - DB_HOST=mariadb
-      - GF_USERS_ALLOW_SIGN_UP=false
+      - DB_PORT=3306
      - DB_NAME=videodetect
      - DB_USER=videodetect
      - DB_PASSWORD=${DB_PASSWORD:-videodetect123}
      - DANCER_ENVIRONMENT=production
    volumes:
-      - grafana_data:/var/lib/grafana
+      - ./api:/app
      - ./monitoring/grafana/provisioning:/etc/grafana/provisioning:ro
    networks:
      - videodetect-network
    depends_on:
-      - prometheus
+      mariadb:
-    deploy:
+        condition: service_healthy
      resources:
        limits:
          memory: 512M
 volumes:
  mariadb_data:
    driver: local
  prometheus_data:
    driver: local
  grafana_data:
    driver: local
  nas_input:
    driver: local
    driver_opts:
@@ -0,0 +1,23 @@
 FROM ubuntu:22.04
 # Install Perl, system deps, and ffmpeg (includes ffprobe)
 RUN apt update && apt install -y \
    libdbd-mysql-perl \
    libhttp-lite-perl \
    libjson-perl \
    libyaml-perl \
    perl \
    libdbi-perl \
    build-essential \
    ffmpeg \
    libstring-shellquote-perl
 EXPOSE 3000
 RUN mkdir -p /app
 WORKDIR /app
 COPY . /app
 # Simple placeholder command
 CMD ["perl", "scanner.pl"]
@@ -0,0 +1,34 @@
 # VideoDetect — Scanner service (standalone compose file)
 # Run: docker compose -f scanner-compose.yml up -d
 services:
  scanner:
    build:
      context: ./
      dockerfile: Dockerfile
    container_name: videodetect-scanner
    restart: no
    environment:
      - DB_HOST=${DB_HOST}
      - DB_PORT=3306
      - DB_NAME=${DB_NAME}
      - DB_USER=${DB_USER}
      - DB_PASSWORD=${DB_PASSWORD}
      - API_HOST=${API_HOST}
    volumes:
      - nas_input:/data:ro
      - $PWD/scanner.pl:/app/scanner.pl:ro
    networks:
      - videodetect_videodetect-network
 volumes:
  nas_input:
    driver: local
    driver_opts:
      type: nfs
      o: addr=10.0.0.2,ro,nfsvers=4,hard,intr
      device: ":/mnt/Bulk/Homes/ryan/Prawns"
 networks:
  videodetect_videodetect-network:
    external: true
@@ -0,0 +1,161 @@
 #!/usr/bin/perl
 use strict;
 use warnings;
 use DBI;
 use Digest::SHA qw(sha256_hex);
 use JSON;
 use String::ShellQuote qw/shell_quote/;
 my $db_host = $ENV{'DB_HOST'} || 'mariadb';
 my $db_name = $ENV{'DB_NAME'} || 'videodetect';
 my $db_user = $ENV{'DB_USER'} || 'videodetect';
 my $db_pass = $ENV{'DB_PASSWORD'} || 'videodetect123';  
 $SIG{INT} = sub { die "Interrupted\n"; };
 my $dbh = DBI->connect("DBI:mysql:database=$db_name;host=$db_host", $db_user, $db_pass);
 my @video_extensions = qw(mp4 mkv avi mov flv wmv mpg mpeg webm);
 my @dir_queue = ('/data');
 # Get NOW() from the database
 my $sth = $dbh->prepare("SELECT NOW() AS now");
 $sth->execute();
 my $row = $sth->fetchrow_hashref();
 my $now = $row->{now};
 $sth->finish();
 while(my $dir = shift @dir_queue) {
    opendir(my $dh, $dir) or die "Cannot open directory $dir: $!";
    print "$dir\n";
    while (my $file = readdir($dh)) {
        next if ($file eq '.' || $file eq '..');
        my $full_path = "$dir/$file";
        if (-d $full_path) {
            unshift @dir_queue, $full_path;
        } elsif (-f $full_path) {
            process_file($full_path);
        }
    }
    closedir($dh);
 }
 # Any entry where last_scan_time < $now is considered deleted or moved, so we can mark them as such
 my $sth_delete = $dbh->prepare("DELETE FROM videos WHERE last_scan_time < ? OR last_scan_time IS NULL");
 $sth_delete->execute($now);
 sub process_file {
    my ($file_path) = @_;
    # Determine if the file is a video based on mime info or extension. For simplicity, let's check the extension.
    if (my ($ext)=$file_path =~ /^.+\.(\S+?)$/) {
        unless (grep { lc($ext) eq $_ } @video_extensions) {
            return;  # Not a video file
        }
    }
    print "$file_path\n";
    my $sth = $dbh->prepare("SELECT id,file_size FROM videos WHERE file_path=?");
    $sth->execute($file_path);
    if(my $row = $sth->fetchrow_hashref()) {
        my $file_size = -s $file_path;
        if ($file_size != $row->{file_size}) {
            warn "File size mismatch for $file_path. Updating record.";
            my $update_sth = $dbh->prepare("UPDATE videos SET file_size=?, last_scan_time=NOW() WHERE id=?");
            $update_sth->execute($file_size, $row->{id});
            $update_sth->finish();
            create_aiscan_task($row->{id});
        }
        return;  # Already exists and size matches
    }
    my $info = get_video_info($file_path);
    unless ($info) {
        warn "Failed to get video info for $file_path. Skipping.";
        return;
    }
    $sth = $dbh->prepare("INSERT INTO videos (file_path, file_size, file_hash, resolution_w, resolution_h, codec, duration, last_scan_time) VALUES (?, ?, ?, ?, ?, ?, ?, NOW())");
    $sth->execute(
        $file_path,
        -s $file_path,
        $info->{file_hash},
        $info->{resolution_w},
        $info->{resolution_h},
        $info->{codec},
        $info->{duration}
    );
    $sth->finish();
    my $video_id = $dbh->last_insert_id(undef, undef, 'videos', undef);
    create_aiscan_task($video_id);
 }
 sub create_aiscan_task {
    my ($video_id) = @_;
    $dbh->do("DELETE FROM tasks WHERE video_id=$video_id AND task_type='AISCAN'");
    my $sth = $dbh->prepare("INSERT INTO tasks (video_id, task_type, status) VALUES (?, 'AISCAN', 'PENDING')");
    $sth->execute($video_id);
    $sth->finish();
 }
 sub get_video_info {
    my ($file_path) = @_;
    # --- Compute SHA-256 hash of the file ---
    return undef unless -f $file_path && -r $file_path;
    open(my $fh, '<:raw', $file_path) or do { warn "Cannot open $file_path: $!"; return undef; };
    my $hash = sha256_hex($fh);
    close($fh);
    # --- Run ffprobe to extract metadata ---
    my $probe_cmd = 'ffprobe -v quiet -print_format json -show_format -show_streams ' . shell_quote($file_path);
    my $output = `$probe_cmd`;
    return undef unless defined $output && length($output);
    my $json = JSON->new->utf8->canonical(1);
    my $data = $json->decode($output);
    # --- Extract codec from video stream (prefer first video stream found) ---
    my $codec = undef;
    if (exists $data->{streams} && ref($data->{streams}) eq 'ARRAY') {
        for my $stream (@{$data->{streams}}) {
            if ($stream->{codec_type} eq 'video') {
                $codec = $stream->{codec_name};
                last;
            }
        }
    }
    # --- Extract resolution and duration from format/streams ---
    my ($resolution_w, $resolution_h, $duration);
    # Duration from format first, then stream
    if (exists $data->{format} && exists $data->{format}{duration}) {
        $duration = $data->{format}{duration} + 0;   # force numeric
    } elsif (exists $data->{streams}[0] && exists $data->{streams}[0]{duration}) {
        $duration = $data->{streams}[0]{duration} + 0;
    }
    # Resolution from video stream first, then format side data
    if (exists $data->{streams} && ref($data->{streams}) eq 'ARRAY') {
        for my $stream (@{$data->{streams}}) {
            next unless $stream->{codec_type} eq 'video';
            if ($stream->{width} && $stream->{height}) {
                $resolution_w = $stream->{width};
                $resolution_h = $stream->{height};
                last;
            }
        }
    }
    return {
        file_hash      => $hash,
        resolution_w   => $resolution_w // 0,
        resolution_h   => $resolution_h // 0,
        codec          => $codec // 'unknown',
        duration       => $duration,
    };
 }
Author	SHA1	Message	Date
Ryan Shpeherd	e4e5d75336	API working	2026-09-09 11:19:00 -04:00
Ryan	ea46d418bd	Move to new db setup	2026-09-08 20:31:06 -04:00