Computer Vision News - August 2021

240 Artificial Intelligence Computer Vision News has found great new stories, written somewhere else by somebody else. We share them with you, adding a short comment. Enjoy! A I S P O T L I G H T N E W S CERN Sparks Podcasts Explore Artificial Intelligence This brand-new podcast by CERN (the European Organization for Nuclear Research ) brings together some of the delegates from the Sparks! Serendipity Forum in September. These are thought leaders among which computer scientists, neuroscientists, philosophers, physicists and psychologists. Why is it interesting? Because they are going to ask the same questions we are asking today about AI - and they will also try to answer. The first episode, Brainy AI , will collide the neurons of machine learning expert Stuart Russell and Tomaso Poggio, one of the founders of computational neuroscience. Read More Tesla Places Big Bet on Vision- Only Self-Drivin g More on autonomous vehicles. This is from IEEE Spectrum , a blog about the sensors, software, and systems that make cars smarter, more entertaining, and autonomous. Just like many of us, these gentlemen were at the CVPR workshop on Autonomous Driving , at which Andrej Karpathy gave a keynote and explained Tesla ’s decision to drop radars. For those who didn’t follow, he said that after training on more than 1.5 petabytes of video augmented with both radar data and human labeling, the vision-only system now significantly outperforms their previous tactic. The blog also quotes Marc Pollefeys ! Autonomous Vehicles: AI Must Accelerate This very nice article is authored by a gentleman who is CEO of a company developing high-performance, low-power and low-cost Artificial Intelligence accelerators. Thus, it is not by mere chance that he recommends innovations in edge computing for autonomous vehicles to realize their full potential. Apparently, he has studied the problem from very close. You will read why it is highly impractical to send even a fraction of the 1 GB of data per second (this is roughly what self-driving cars generate) for analysis to a centralized server because of the processing bandwidth and latency . Read More