RowdyHacks 2023 Project
This project uses the MediaPipe Gesture Recognizer Task found here: https://developers.google.com/mediapipe/solutions/vision/gesture_recognizer
We were able to follow a tutorial given by Ivan Goncharov that helped us understand how to add our own gestures, found here: https://youtu.be/a99p_fAr6e4
The Github repo that includes the model and keypoints classifications we used is found here: https://github.com/kinivi/hand-gesture-recognition-mediapipe
To run the program, just run the app.py
When adding new gestures to the dataset, press 'K' while the video feed is open. Then use 0-9 to add keypoints to the appropriate label. Open -- 0 Close -- 1 Pointer -- 2 OK -- 3 Peace -- 4 A chord -- 5 C chord -- 6 E chord -- 7 D chord -- 8 G Chord -- 9
The keypoints get added to the keypoint.csv under the label number.
Then rerun the keypoint_classification jupyter notebook so the data is trained with the new keypoints from the keypoint.csv.
Run app.py again