Dynamic Processing at the Intersection of Gaming and Deep Learning for AV Applications
Deep convolutional neural networks are great in extracting mappings of where high level features are found within static images, which in turn are used for object recognition and detection. But for getting autonomous vehicles on the road soon, we would like to study how the objects move across frames in a video. Enhanced object detection requires fusing motion cues (spatiotemporal streams) with static images (RGB based appearance streams). Maintaining state information has proven to be critical to aggregate motion information over time to assist in future predictions. Data obtained from rich virtual world found in games and other simulated environments need to be utilized to bootstrap our research efforts.
Gaurav Kumar Singh is a Machine and Deep Learning Researcher at Research and Advanced Engineering at Ford Motor Company, located in Dearborn, Michigan. He has over 6 years of research experience ranging from Control Systems to Machine Learning and Data Science. His side gigs involve consulting friends in ways to utilize machine learning techniques in their startups. He has served as project reviewer and mentor for Machine Learning and Self Driving Car Nanodegree at Udacity as well. Gaurav graduated with a Masters’ degree in Electrical and Computer Engineering from University of Michigan, Ann Arbor in December 2015. He received his Bachelors of Technology (B.Tech) degree from National Institute of Technology, Trichy, India in 2014.