Date/Time
Date(s) - 02/20/24
6:00 pm
Categories
Vansh Amin :: Unlocking Enhanced Perception for Autonomous Vehicles using Generative Adversarial Networks
Vansh is an Electrical Engineering student at Texas A&M University on a full-ride scholarship as a Brockman Scholar. He is currently the president of Aggies Create, a engineering consulting and product development group at Texas A&M and a researcher in the CAST (Connected Autonomous Safe Technologies) lab. His passions revolve around vehicles and how to make them more safe and accessible with next generation technology.
Presentation Title: Unlocking Enhanced Perception for Autonomous Vehicles using Generative Adversarial Networks
Presentation Description: With the advent and increasing popularity of Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicles (AVs), unconventional passive visual sensors such as hyperspectral imaging have the ability to provide superior object detection abilities due to the rich spectral information they record. These capabilities can change paradigms for the AV industry, and result in safer systems to deploy in the real world. This ongoing work aims to improve hyperspectral imaging through the use of a custom Generative Adversarial Network (GAN) in order to correct for scene illumination, contrast, and shadowing issues. This enhancement would result in higher performing segmentation and reduce the possibility of missed detections further down the AV perception stack.