ApexCull
RAW culling that groups every shot by car.
A race weekend produces thousands of RAW frames. ApexCull scores each for sharpness, detects and tracks vehicles with YOLO, and clusters every shot of the same car across the batch using Re-ID embeddings — so a shoot collapses to keepers, grouped by car, in minutes. I designed and built it end to end: the Next.js app, the FastAPI + Celery vision pipeline, and the infrastructure.
- STACK
- Next.js · FastAPI · Celery · Redis · Postgres
- VISION
- YOLO11 · OSNet Re-ID · DBSCAN · OpenCV
- INFRA
- Docker · Terraform · Cloudflare R2
- ROLE
- Solo — product, ML pipeline, infra
- YEAR
- 2025
- STATUS
- Live