A smart image deduplication tool that uses AI to detect similar images based on content, not just file hashes.
- AI-Powered Detection: Uses OpenAI's CLIP model to understand image content
- Smart Similarity: Finds duplicates based on visual similarity, not exact matches
- Fast Search: FAISS-based similarity search with HNSW indexing
- Easy to Use: Simple Makefile commands for common operations
- Local Storage: All data stored locally for privacy
# Install dependencies
make install
# Upload a single image
make upload IMAGE=path/to/image.jpg
# Process all images in a folder
make bulk-custom FOLDER=path/to/images
# Check status
make status
- Python 3.7+
- ~150MB for CLIP model (downloaded automatically)
- Clean up photo libraries
- Remove duplicate screenshots
- Organize image collections
- Find similar images across folders
├── main.py # Core deduplication logic
├── Makefile # Easy commands
├── models/ # CLIP model wrapper
├── store/ # Images and metadata
└── utils/ # Helper functions
- Similarity Threshold: 0.9 (adjustable)
- Supported Formats: PNG, JPG, JPEG, WEBP, BMP, TIFF
- Storage: Local
store/
directory
MIT License