PicFinder AI 🔍

Smart Local Image Search Engine powered by OpenAI CLIP

Semantic Search · 100% Privacy · Offline · Multilingual Support

PicFinder AI is a desktop application that allows you to search through your local photo albums using natural language. Instead of remembering filenames, simply search for "A cat sleeping on the sofa" or "Red sports car".

Powered by the CLIP model and ChromaDB, it runs 100% offline, ensuring your photos and data never leave your device.

• Semantic Search: Search by content description, not just keywords.
• Privacy First: Completely offline. No cloud uploads.
• Smart Indexing: Incremental scanning that only processes new images to save time.
• Multilingual Support: Supports queries in English, Chinese, and Japanese.
• Cross-Platform: Built with Electron & Python.

git clone [https://github.com/Breathinggg/PicFinder-AI.git](https://github.com/Breathinggg/PicFinder-AI.git)
cd PicFinder-AI

Run the installer script to automatically set up the Python virtual environment and install Node.js dependencies.

install.bat

Start both the backend and frontend with a single command.

run_app.bat

⚠️ Important Note on First Run:

When you launch the app for the first time, the backend will automatically download the AI model (~1.5GB) from Hugging Face.
Please wait patiently and keep the terminal window open until the download completes. Once downloaded, the app will work offline forever.

• Frontend: Electron, Vue 3, Vite
• Backend: Python (FastAPI, Uvicorn)
• AI Core: sentence-transformers (clip-ViT-L-14)
• Database: ChromaDB (Vector Store)

Contributions are welcome! Please feel free to submit a Pull Request.

Distributed under the MIT License.