[Python] How can I detect a circular motion in object tracking using YOLO keypoints?

What I want to achieve: Identify when a foot starts moving in a circular trajectory. Track the foot's path to ensure it completes a loop. Validate the motion (e.g., foot passes above and then below a ball).

My Questions: How can I mathematically detect a circular movement using YOLO keypoints? Should I use a trajectory-based approach (e.g., tracking displacement over time)? Are there any libraries (like OpenCV or NumPy) that can help identify circular paths in keypoint tracking? Any insights, code snippets, or algorithm suggestions would be greatly appreciated!

• I tracked the (x, y) positions of the left and right ankles across multiple frames.
• I calculated the displacement of the foot by checking changes in x and y coordinates.
• I attempted to detect upward and downward motion using:

if foot_y < previous_foot_y:
moving_up = True
elif foot_y > previous_foot_y:
moving_down = True



Issues I'm facing:

• Simple x/y tracking doesn’t capture circular motion accurately.
• The foot sometimes moves diagonally but isn't necessarily circular.
• How do I mathematically determine if the foot completes a full circular trajectory? Any guidance on detecting circular motion using YOLO keypoints would be really helpful!

Continue reading...