Hands-On-Learning-in-Computer-Vision

Built an assembly line CV system to track hand-tool interactions and verify screw tightening status Hey everyone! Wanted to share a quick demoof an industrial computer vision pipeline I’ve been working on for assembly line quality control. The setup is designed to monitor manual assembly processes, specifically tracking worker hands, tools, and individual fastener states to ensure zero-defect assembly. What it’s doing in the clip: - Hand & Tool Tracking: Accurately bounding the worker's hands and the power tool during operation. - State Classification: Dynamically tracking bolts/screws and switching their states from "Loose" to "Tightened" in real-time as the assembly steps are completed. It’s been a fun challenge dealing with heavy occlusion (hands getting in the way of bolts) and fast-moving industrial environments. Would love to hear your thoughts, feedback, or any ideas on how to optimize this further for edge deployment! Check out the clip above. video: Link github repo: Link