A Rapid Prototyping System for Hand Gesture Recognition
Tim Straubinger and Nam Hee Gordon Kim
Abstract
Abstract
Prototyping gesture recognition systems in a fast, reliable, and cheap way can make HCI applications easier to develop and more accessible. We present a simple system that combines a 3D hand pose estimator and a speed-invariant gesture recognizer. Our system allows for rapid prototyping of vision-based hand gesture recognition software without the burden of massive data collection.
Raw Data: Clutch Pose
Raw Data: Clutch Pose
clutch23.mp4clutch22.mp4Raw Data: Dynamic Gestures
Raw Data: Dynamic Gestures
ok02.mp4OK
thumbs_up01.mp4Thumbs up
paper02.mp4Paper
scissors00.mp4Scissors
call_me01.mp4Call me
lets_drink00.mp4Let's drink
Bias Correction for Clutch Pose with Residuals
Bias Correction for Clutch Pose with Residuals
554x-residuals-with-subs.webm