Computer Vision News - February 2022

13 AI Spotlight News England Has an £850 Million Problem with Litter. This Startup Is Using AI to Fix It British tech company Littercam has developed a technology that uses AI and computer vision to detect littering from cars and match it with the offender’s license plate. The software is already being trialed and it is able to detect very small items being littered , like cigarette butts; BTW, did you know that it takes a cigarette butt about 10 years to decompose? But that’s another story. What this story tells is that authorities arenowusingAI inall kinds ofweather conditions to have a penalty notice issued to the car’s registered owner, in the hope of preventing littering in the future. Watch the Video Artificial Intelligence Can Now Make Jokes - Jon the Robot Robots are apparently able now to tell jokes and the novelty is that they adapt their lines to the feedback they receive from their audience. Created by Naomi Fitter , an assistant professor at Oregon State University, Jon the Robot performs live while at the same time it learns how to respond to its audience. Jon does not understand why we laugh at its jokes, and it is certainly not ready yet to craft its own repertoire of funnies. Still, you can say that it “feels” the listeners and adapt to their reactions (or lack of) in almost real time. “ You can probably tell from looking at me that I am from the Valley! ” Watch the Video Synthetic Data: A Solution for Improving Driver Safety Beginning in 2022, all new cars entering the EU must be equipped with advanced safety systems: distraction recognition and alert systems on trucks and buses to warn about vulnerable pedestrians or cyclists in close proximity. Meeting the new safety regulations will not be an easy task for car manufacturers, who will be faced with dedicating vast amounts of resources to build and deploy cars to collect diverse datasets to train AI models. One of their allies is synthetic datasets – computer- generated simulations that ensure an ample supply of diverse and anonymous training data. Read More