GSoC’20 RoboComp project: Human recognition (identification) using multi-modal perception system
11th June 2020
I am Shubh Maheshwari, currently in my final year at IIIT-H. I have been working for 2 years in the Computer Vision Center of our college. Currently, I am doing honors under Prof. C.V. Jawahar. I have also been teaching assistant for SMAI (Statistical Methods In AI) and Optimization Methods.
Previously I set up a tour guide robot for our lab. The agent was able to track, tell jokes and interact with users. For the past year, I have been working on skeleton-based action recognition & synthesis.
About the Project
In Robotics, it is crucial to identify humans and efficiently distinguish them. This gives the ability to perform further challenging tasks such as personalized interactions, social navigation and surveillance. Currently, Robocomp only uses face recognition for human identification. The aim of this project is to integrate other modalities, such as rgb image, skeleton-based pose, and silhouette of the body. Using these modalities, different human identification components (such as gait recognition & person re-identification) will be evaluated. Finally a multi-modal pipeline will be created to integrate these components into Robocomp.