FaceCrime - AI Facial Recognition Offender Registry Search

Full-stack facial recognition portal that compares user images against a large offender database using Jina CLIP embeddings and a fast vector search via Postgres/pgvector.

The frontend is a fast React/Vite/TypeScript application using Shadcn/ui for the UI. Users submit an image, which is processed and sent to the FastAPI backend. The Python/FastAPI backend uses Jina CLIP (via requests) to generate a 768-dimensional embedding from the image. It then performs a similarity search against a PostgreSQL database with the pgvector extension to find the most similar face from the mugshot dataset. It calculates a `matchPercent` and returns public offender attributes (offense, height, etc.) to the frontend.