Easy to use and easy to understand. A simple Custom Hand Gesture Recognition using a Random Forest Classifier.
What functionality does each file in this repo have?
- collectDataSet.py
- To collect the images for each gesture that we are going to classify.
- createDataSet.py
- From the images that we collected, we extracted the hand landmark information and created 2 lists. One with landmark information of hand and the other with corresponding gesture class. And created a pickle file (data.pickle) using 2 lists.
- modelTraining.py
- We trained a Random Forest Classification model using the landmark data inside a pickled file(data.pickl e). And saved the best accurate model (model.p) using pickle.
- modelTesting.py
- We access the camera, read the frames, and predict the gesture. This model can predict a maximum of 2 hand gestures at a time. And we display the frame with a visualization of landmark points, and a bounding box with a gesture name on it.