A custom CNN-RNN model was developed using Tensorflow and Python for the purpose of gesture recognition. It incorporates the Keras API, MobileNetV2 CNN, Bidirectional LSTM RNN, Feed Forward Network, Video Frame Generator, Frame Sampling, Data Loader, JSON Label Splitter, Data Visualization, and Evaluation Metrics. The model aims to detect and classify body movements from video based on annotated labels. It utilizes body movement data captured by OpenPose and requires arranging the data according to the specified format. The project includes a requirements.txt file for downloading the necessary libraries and requires modifying folder paths in the relevant files. To run the model, execute models.py.
Note: proprietary dataset used.