MemoryRace
After a race, photos are published in bulk. Type your bib number, get yours.
- When
- Jan 2025 – Jun 2025, software-development project at Télécom SudParis
- Team
- 4 developers
- Code
- github.com/NnicolasN/MemoryRace — MIT
- Stack
- Java 17, Swing, JDBC, Maven, JUnit 5 · MariaDB in Docker · Python 3, Google Gemini API · GitHub Actions
How it works
- Organizers sign in, create races, upload photos; runners search with no account
- At upload, a Python script sends each image to Gemini and stores the bib numbers read
- A search is a plain SQL query: photos appear immediately
- Layered connection model: a remote server can replace the local database without touching the UI
- 30-photo micro-dataset annotated in CVAT as ground truth for detection