Juggle Buddy watches a juggling performance live, detects motion with OpenCV and YOLOv8, and compares the result against pretrained models for different juggling patterns.
I used the project to practice building a playful interface around a serious computer-vision loop: capture, classify, compare, and give feedback without making the user feel like they are debugging a model.
The test layer used PyTest and serialized model artifacts to keep iteration fast.