AI-Powered Photo Metadata Pipeline

Automatically classify and tag your wildlife photos with AI precision

✓ 10,000+ Species
✓ GPU Accelerated
✓ 100% Local & Private
✓ Docker Powered
✓ Free for Personal Use
New in v1.1 Watch Mode for continuous monitoring, RAW+JPEG pairing, and XMP keyword preservation See what's new →
🎯

Smart AI Detection

RT-DETR object detection combined with iNat21 species classification identifies 10,000+ species automatically.

Lightning Fast

Process thousands of photos in minutes. 10x faster with GPU acceleration - up to 10 photos per second.

📝

XMP Ready

Works seamlessly with Lightroom Classic, On1 PhotoRAW, Immich, and other photo management tools.

🔒

100% Local & Private

Your photos never leave your computer. All AI processing happens locally - no cloud uploads, no data sharing, complete privacy.

🚀

2-Minute Setup

Get started with a single Docker command. No complex installation, no dependencies to manage - just run and go.

🔄

Never Lose Progress

Checkpoint-based processing lets you pause and resume anytime. Perfect for large collections - pick up right where you left off.

👁️

Watch Mode New

Continuously monitor folders for new photos. Perfect for tethered shooting or automatic imports from your camera.

📷

RAW+JPEG Pairing New

Intelligent pairing of RAW and JPEG files. One shared XMP sidecar for both - works with NEF, CR2, CR3, ARW, DNG, and more.

See It In Action

Input Photo

IMG_1234.nef IMG_1234.jpg No keywords

Lumina Processing

AI Analysis: - RAW+JPEG Pairing - Object Detection (RT-DETR) - Species Classification (iNat21) - XMP Sidecar Generation

Output Keywords

IMG_1234.nef IMG_1234.jpg IMG_1234.xmp (shared) ├─ COCO Detection │ └─ bird └─ Inat21 └─ bird └─ European Starling

Get Started in 2 Minutes

# Create project directory
mkdir -p lumina-docker/{photos,cache} && cd lumina-docker

# Download docker-compose.yml
curl -o docker-compose.yml https://raw.githubusercontent.com/stevenvanassche/Lumina/master/docker/docker-compose.yml

# Create .env file
cat > .env << EOF
PHOTOS_PATH=./photos
CACHE_PATH=./cache
EOF

# Add your photos to photos/ directory, then run:
docker compose --profile cpu up    # CPU mode (~1 photo/sec)
docker compose --profile gpu up    # GPU mode (~10 photos/sec)

# NEW in v1.1: Watch mode - continuously monitor for new photos
LUMINA_WATCH_MODE=on docker compose --profile cpu up

First Run: Models download automatically (~2.5GB, 5-15 minutes)
View Full Documentation →

Perfect For

🦅

Wildlife Photographers

Automatically tag thousands of bird and animal species in your photo collection.

📸

Photo Enthusiasts

Organize your collection with AI precision. Works with your existing workflow.

🔬

Researchers

Batch process biodiversity data for ecological studies and citizen science.

Proven Performance

1.14
photos/sec (CPU)
~15 minutes for 1,000 photos
10.74
photos/sec (GPU)
~1.5 minutes for 1,000 photos
10x
faster with GPU
NVIDIA RTX 5090 tested

Supported Formats

Standard

JPEG PNG

RAW Formats New

NEF CR2 CR3 ARW DNG RAF ORF RW2

Works With Your Tools

Adobe Lightroom Classic Full hierarchical keyword support
On1 PhotoRAW Complete XMP sidecar integration
Immich External library XMP import
Capture One, DigiKam, ACDSee Standard XMP sidecar support

View Integration Guides →

Open & Free

Free for Personal Use

  • Personal photo collections
  • Educational purposes
  • Research projects

Commercial Licensing

  • Business & enterprise use
  • Contact us to discuss your requirements
Get in Touch

View Full License →

Ready to Transform Your Photo Workflow?

Start processing your photos with AI precision today