Plate Finder

Use a pretrained model to display license plates extracted from images.

Demo: https://better.platefinder.space

Development notes: DEVLOG.md

Architecture

• Vite + React + TypeScript frontend
• FastAPI backend using a pretrained YOLO-based model in a Docker container

Development

This repo includes a devcontainer that has the github CLI and act, a local Github actions runner.

• Github authentication: To use act, you need to authenticate with Github. Run:

  gh auth login -s repo,gist,read:org,write:packages,read:packages,delete:packages

  The package permissions are needed for act to write to the Github package registry.

  If you want to run just a single action, you can use the --job flag:

  act --job build-and-push

• Host mounting: the host's Docker socket (assuming MacOS and Linux) is mounted into the container workspace.

You'll want to run both the frontend and api to test things locally:

cd api/src && uv run python3.12 -m uvicorn main:app --port 8000

cd web && pnpm dev