Closed Nickil21 closed 3 years ago
You can use some clever math to figure out the axis (plane), direction (view/facing), and rotation https://www.youtube.com/watch?v=L95mA_h0CjA
And maybe use a simple hand shape classification model https://www.youtube.com/watch?v=pCKRWSNIaNQ Although it is not very accurate at this point.
Math + model inference code: https://github.com/sign-language-processing/playground/blob/master/src/app/modules/sign-writing/hands.service.ts
Hi @Nickil21, Please refer this blog for Gesture Recognition https://ai.googleblog.com/2019/08/on-device-real-time-hand-tracking-with.html
Hi @sgowroji , It said in the blog that the Gesture Recognition is already published
The existing pipeline supports counting gestures from multiple cultures, e.g. American, European, and Chinese, and various hand signs including “Thumb up”, closed fist, “OK”, “Rock”, and “Spiderman”.
But I can't seem to find it, can you help?
Many thanks
Is it possible to identify hand gesture types such as axis (vertical, horizontal), shape (arced/straight), direction (upward/downward/leftward/rightward), and handedness (left/right/both hands) using mediapipe for a video frame as the input? I know currently that it supports the handedness gesture type, but I wondered if something exists for the other types.