Built for UTA's SCAI RoboMaster team, this project turns tournament recordings into a competition-oriented object-detection dataset. The lightweight nano model was selected to keep the pipeline compatible with real-time inference on robot hardware.
Engineering contributions
- Extracted representative frames from RoboMaster tournament footage
- Annotated enemy robots and pressure plates as separate detection classes
- Trained a YOLOv8n object-detection model
- Prepared the pipeline for expanded footage and edge-case data
System path
01Match footage
02Frame extraction
03Annotation
04YOLO training
05Edge inference