Show HN: Offline AI Photo Search (local VLM and semantic search)

1 point

6 months ago

I built a small offline photo search tool that uses a local vision-language model (NexaAI Qwen3-VL-4B) to describe images and sentence-transformers to generate embeddings for semantic search.

Everything runs 100% on-device (no cloud, no API keys).

GitHub repo: https://github.com/pankaj4152/smart-photo-finder

You can try it with: 1. python app.py 2. Choose "Process images" 3. Choose "Search images"

Would love feedback on architecture, performance, and improvements.