No. 02 · 2021 · Personal
Automatic dart scoring system
Computer vision + ML
How it was built
Started with two webcams and OpenCV blob detection, which wasn't accurate enough in 2D.
Moved to 4 elevated cameras with a light ring, generated training data and trained a YOLOv8 network.
Settled on 3 cameras for performance, then printed the frame and built the GUI and game logic.
Skills
- Computer vision (OpenCV, YOLO, 3D reconstruction)
- Python
- 3D printing & parametric CAD (SolidWorks, OpenSCAD)
- Git

