ReGEAR: AR Gesture Recognition for Nuclear Facilities

The software ReGEAR performs real-time analysis of gestures and actions in video streams using a 3D Convolutional Deep Neural Network (3D CNN) model. It is written in Python using open-source image processing libraries and Tensorflow, a deep learning API by Google, for the 3D CNN model. It is easy to extend and recognize new gestures without changing the code. The training can be performed on Hoodoo cluster which is a Lambda Hyperplane deep learning distributed memory system of INL HPC system, and the software can run on a personal laptop with low computing power. The software is aimed to be integrated with augmented reality wearable devices for monitoring and assisting the field activities in the nuclear facilities where the field workers must follow the work procedures.

This software is under copyright. To purchase a license, please use the 'Contact Us' form on this page. We will respond as promptly as possible.