Computer face recognition systems are already working quite well and even know how to distinguish emotions. But people transmit information not only through the face, but also through various gestures and postures. So scientists from the Carnegie Institution's Robotics Institute - Mellon decided to create a program that would easily "read" the language of human gestures and interpreted the information received.

In their development, scientists used 500 high-precision video cameras installed inside a huge dome with a height of 2 floors. Five hundred cameras (provided to researchers at Panoptic Studio) produce a huge amount of information, and even with a single frame for analysis of all visual information, considerable computing resources are required. In this case, the system should conduct simultaneous evaluation of the facial expression, the position of the head, trunk, legs and all fingers. As the project leader Yasser Sheikh said, "A person expresses his state and emotions with the help of facial expressions, movements and body postures no worse than with his own voice. But computers until recently remained "blind" to this language of our body. We managed to "teach" artificial intelligence to what we can do practically from the very beginning. Now we plan to improve the software in such a way that for analyzing the system there is only one camera.

At the same time, we want to improve the system so that it can decipher the body language of a whole group of people interacting with each other ".

Data on the new method of decoding the language of the human body, as well as its source code are now freely available. More detailed information on their development, scientists plan to present at the Computer Vision and Pattern Recognition Conference (CVPR) 2017, which will be held from July 21 to 26. Well, you can get acquainted with the novelty yourself already now with the help of the video below.