Software Used

• python - programming language used, tested on v3.8
• miniconda - for managing virtual environment

Libraries Used

• opencv - pip install opencv-python
• imutils - pip install imutils
• pillow - pip install Pillow
• tensorflow
  • pip install tensorflow - for CPU and GPU
  • pip install tensorflow-gpu - for GPU
  • pip install tensorflow-cpu - for CPU
  • keras
• numpy - pip install numpy
• scikit-learn - pip install scikit-learn
• matplotlib - pip install matplotlib

Modules

• Image Segmentation - just for leaning image segmentation
• Data Generation - for generating the gestures dataset
• Data Training - for training the CNN model
• Data Prediction - for predicting the gestures
• Test GPU - if you are using GPU use this for test if you have done CUDNN setup properly.

Image Segmentation

• This module is just for learning purpose.
• You can see here how segmentation code works.
• Use this module to play around and understand image segmentation.

Data Generation

• Contains the code for dataset generation.
• You can add new gestures in this notebook and the generate the data.
• Produce 1000 train data, and 100 test data images.
• This can be done by setting the no_of_images and start_image_num variables.
• After adding new gesture modify the gestures list for both data training and data generation module.

Data Training

• Contains the CNN model.
• Modify this model to crete your own new model and train it.
• Use GPU for faster training.
• If you have a Nvidia GPU, follow this https://www.tensorflow.org/install/gpu to make tensorflow work with your GPU.

Data Prediction

• Contains the code for predicting gesture.
• Loads the CNN model and make the prediction.

License

MIT