Skip to content

demo_commands.md

Latest commit

 

History

History
243 lines (179 loc) · 6.43 KB

demo_commands.md

File metadata and controls

243 lines (179 loc) · 6.43 KB

VideoAnnotator Demo Commands & Usage

📖 Navigation: Getting Started | Accessing Results | Pipeline Specs | Installation Guide | Main Documentation

This guide demonstrates how to use VideoAnnotator with its API server and CLI interface for video processing.

🚀 Quick Start Commands

If you are running via Docker Compose (recommended), you can run commands inside the container like this:

# CPU service
docker compose exec videoannotator va --help

# GPU service (if you launched it)
docker compose exec videoannotator-gpu va --help

Start API Server

# Start the VideoAnnotator API server
uv run videoannotator server --port 18011

# Server will be available at http://localhost:18011
# Interactive documentation at http://localhost:18011/docs

Process Single Video via CLI

# Submit a video processing job through CLI
uv run videoannotator job submit video.mp4 --pipelines scene,person,face

# Check job status
uv run videoannotator job status <job_id>

# Get detailed results
uv run videoannotator job results <job_id>

# Download annotations (ZIP)
uv run videoannotator job download-annotations <job_id> --output ./results/

Process Video via HTTP API

# Set your API key (from server startup output)
export API_KEY="va_api_your_key_here"

# Submit job via HTTP POST
curl -X POST "http://localhost:18011/api/v1/jobs/" \
  -H "Authorization: Bearer $API_KEY" \
  -F "video=@video.mp4" \
  -F "selected_pipelines=scene,person,face"

# Check status
curl -H "Authorization: Bearer $API_KEY" \
  "http://localhost:18011/api/v1/jobs/{job_id}"

# Download artifacts (ZIP)
curl -H "Authorization: Bearer $API_KEY" \
  "http://localhost:18011/api/v1/jobs/{job_id}/artifacts" \
  --output artifacts.zip

🛠️ CLI Management Commands

System Information

# Show system status and database info
uv run videoannotator info

# List available pipelines
uv run videoannotator pipelines --detailed

# Validate configuration files
uv run videoannotator config --validate configs/default.yaml

Job Management

# List all jobs
uv run videoannotator job list

# List completed jobs only
uv run videoannotator job list --status completed

# Get job results with details
uv run videoannotator job results <job_id>

Job queueing

VideoAnnotator processes jobs in a simple FIFO queue:

  • Jobs are dequeued in increasing created_at order (oldest pending job runs first).
  • API responses may include queue_position for pending jobs.
    • queue_position is 1-based among pending jobs.
    • queue_position is null (or absent) for non-pending jobs.
    • queue_position can change over time as other jobs complete.

🚀 Modern API-First Architecture

Key Features:

  • Integrated Background Processing - No separate worker processes needed
  • Real-time Job Status - Live job tracking and progress updates
  • Complete Pipeline Integration - All pipelines working through API
  • Modern CLI Interface - Comprehensive command-line tools
  • Production Ready - Designed for research and production workflows

📋 Complete API Reference

Available Pipelines

  • scene_detection - Scene boundary detection with CLIP classification
  • person_tracking - YOLO11 + ByteTrack multi-person pose tracking
  • face_analysis - OpenFace 3.0 + LAION facial behavior analysis
  • audio_processing - Whisper speech recognition + pyannote diarization

Pipeline Combinations

# Run all pipelines
uv run videoannotator job submit video.mp4 --pipelines scene,person,face,audio

# Scene + person analysis
uv run videoannotator job submit video.mp4 --pipelines scene,person

# Face analysis only
uv run videoannotator job submit video.mp4 --pipelines face

🔄 Working with Job Results

Get Job Results

# Get summary of job results
uv run videoannotator job results <job_id>

# API endpoint for results
curl -H "Authorization: Bearer $API_KEY" \
  "http://localhost:18011/api/v1/jobs/{job_id}/results"

# Download specific pipeline result file
curl -H "Authorization: Bearer $API_KEY" \
  "http://localhost:18011/api/v1/jobs/{job_id}/results/files/scene_detection" -O

Configuration Options

# Use custom configuration
uv run videoannotator job submit video.mp4 --config configs/high_performance.yaml

# Validate config before use
uv run videoannotator config --validate configs/high_performance.yaml

# View default configuration
uv run videoannotator config --show-default

System Management

# Show version and system info
uv run videoannotator version
uv run videoannotator info

# Backup database
uv run videoannotator backup backup_$(date +%Y%m%d).db

# Server management
uv run videoannotator server --host 0.0.0.0 --port 18011

📊 Expected Output Format

Job Submission Response

{
  "id": "job_abc123",
  "status": "pending",
  "queue_position": 1,
  "video_path": "/path/to/video.mp4",
  "selected_pipelines": ["scene", "person", "face"],
  "created_at": "2025-08-26T10:30:00Z"
}

Job Status Response

{
  "id": "job_abc123",
  "status": "pending",
  "queue_position": 2,
  "created_at": "2025-08-26T10:30:00Z",
  "selected_pipelines": ["scene", "person", "face"]
}

Job Results Response

{
  "job_id": "job_abc123",
  "status": "completed",
  "pipeline_results": {
    "scene_detection": {
      "status": "completed",
      "processing_time": 15.2,
      "annotation_count": 8,
      "output_file": "/path/to/output/video_scene_detection.json"
    },
    "person_tracking": {
      "status": "completed",
      "processing_time": 45.7,
      "annotation_count": 156,
      "output_file": "/path/to/output/video_person_tracking.json"
    }
  },
  "output_dir": "/path/to/output/"
}

🎯 Next Steps

  1. Start the server: uv run videoannotator server
  2. Submit a job: uv run videoannotator job submit your_video.mp4
  3. Monitor progress: uv run videoannotator job status <job_id>
  4. Get results: uv run videoannotator job results <job_id>
  5. Explore API: Visit http://localhost:18011/docs for interactive documentation

For more advanced usage, see: