Extraction of Gesture Features

This is the repository for all the code created during the GSoC 2023 at Red Hen Lab for the project topic 'Extraction of Gesture Features'.
The main aims of this project were:

• Goal is to develop a relevant pipeline using deep learning methods and modules to identify such gestures.
• Further Identifying the type of gesture, spoken words, speaker, and speech type in that specific frame.
• Later find an analysis between different parameters to find the meaningful patterns and insights.
• The baseline output entails either a video overlaying all parameters or a comprehensive data frame containing frame-specific information.

For further information about the work, information on How to use and to check out my work documentation:

• check my blog @ https://medium.com/@Dhruv_Tyagi/gsoc-2023-red-hen-lab-feaa572f2cdc

Structure of The Code

The src directory contains all the code. The code for all the steps is under their associated folder such as the code for models under the model folder. The workflow folder contains the main code for overlaying of all the parameters. All the trained models are under Models folder.