Object recognition based on neural networks is becoming more perfect every year. They simplify and automate many processes, make our life more comfortable and safer. But where exactly are they used? We are talking about unusual applications of computer vision that are available now.
Cat smart home
In China, Baidu Brain engineer Wan Xi made a multi-level shelter for stray cats, equipped with a camera at the entrance. With the help of it, as well as artificial intelligence, it is possible to recognize "guests" - the door to the shelter opens only if it recognizes a cat in the animal at the entrance. Now computer vision is able to recognize 174 different cats, their health status, and even whether they are spayed. As an avid cat lover, Wan Xi hopes that his invention will simplify the hard life of abandoned animals by providing them with shelter and warmth.
Scanning coins
Researchers at the Fraunhofer Institute have developed software that scans old coins and recognizes them in seconds. The system can be used to identify archaeological finds.
Promotional video:
The invention was named OSCAR, short for "Optical System for Coin Analysis and Recognition" (Optical System for Analysis and Coin Recognition). It not only recognizes the print of a coin, but also detects unique details: scratches, scuffs, dents, contours and edges.
From left to right: scans of color, surfaces, and final result.
Recognition of buyers, traffic offenders and toilet paper thieves
Let's go back to China, where computer vision and object recognition are already used in everyday life. For example, in public latrines, visitors are recognized by a system that dispenses toilet paper. At the same time, she strictly monitors that no one takes too much.
Another use case is automatic payment for purchases. Cameras record buyers at the entrance and exit, and artificial intelligence finds their electronic profile in a special application. When a person goes to the exit with purchases, they are scanned by an automatic validator, and the account is sent to a personal profile.
Also, computer vision was taught to identify traffic offenders crossing the road in the wrong place. The system is comparable to the one that “catches” drivers who have violated the rules and issues a fine.
Artem Bagdasarov