Hand_Gesture_Recognition

Contains reproducible code for Hand Gesture recognizer.

*Dependencies:

1. Tensorflow-gpu=1.14 => conda install -c anaconda tensorflow-gpu=1.14
2. Keras=2.2.4 => conda install -c conda-forge keras=2.2.4
3. Open cv => conda install opencv-python
4. Python=3.6
5. PIL

*Code Directory contains the file as follows:

1. Conv_net.py : Contains the code to train the model on the processed data.
2. getting_data.py : Using open cv, and numpy capture the frames of the webcam and generate the dataset.
3. process_data.py : Helps in processing the images to specified shape, and stores them as numpy array in pickle file.
4. testing_model.py : Can be used for realtime prediction as well as inference on static images of specified format.

*Results Directory contains the accuracy and loss plots of the trained model, achieving over 99% accuracy on both train and test sets.

*Pretrained model file as well as pickle files are uploaded now!