This repository hosts deployable community projects for the reComputer AI Lab.
Projects are grouped by hardware platform, while container definitions are kept
under docker/.
| Project | Hardware | Description |
|---|---|---|
| Retail Shelf Detection | Raspberry Pi 5 + Hailo-8 | Product detection, inventory tracking, pickup events, and low-stock alerts from video |
| Pothole Detection | Raspberry Pi 5 + Hailo-8 | Road pothole detection from video with ROI filtering, temporal tracking, and gap-fill display |
| Abandoned Luggage Detection | Raspberry Pi 5 + Hailo-8 | YOLO11m-based abandoned luggage detection with owner association, static persistence, and alarm hold |
Published containers:
ghcr.io/seeed-projects/recomputer-ai-lab-project/retail_shelf_detection:latest
ghcr.io/seeed-projects/recomputer-ai-lab-project/pothole_detection:latest
ghcr.io/seeed-projects/recomputer-ai-lab-project/abandoned_luggage:latest
| Project | Hardware | Description |
|---|---|---|
| Pull-up Counter | reComputer RK3576 | YOLO11 Pose pull-up counting with RKNN NPU inference and a browser dashboard |
Hailo-8 Dockerfiles are stored under docker/hailo8/. Each Dockerfile uses its
project directory as the build context; project-specific build and run commands
are documented in that project's README.
GitHub Actions builds linux/arm64 images and publishes them to GHCR when the
corresponding project or Dockerfile changes on main.