Artificial intelligence has learned to see through walls


Искусственный интеллект научился видеть сквозь стены Development can help to monitor patients with Parkinson’s disease.

Scientists have created a neural network that distinguishes between motion and posture of a person, even if it is behind an obstacle. About the new technology, the authors said at the conference on computer vision and pattern recognition in salt lake city. Briefly about the report can be read on the website EurekAlert.

According to the authors of the study, scientists from mit, the idea to create such technology appeared when they noticed that the movements a person can make conclusions about potential diseases, such as Parkinson’s disease or multiple sclerosis. While the naked eye to notice the first signs of the disease very difficult. Therefore, the researchers created a neural network that can distinguish the radio signals reflected from the person, and for him to recognize his movement or pose.

At the initial stage of creation of the neural network must be trained. For example, to make a neural network that learns on the cat photos, you must first show her lots of photos and point out where the cat really is, and where it is not. This is usually done manually. In the case of detection of postures and movements of the radio signal is very difficult to do. To solve this problem, the researchers used at the initial stage of a normal cell. Received with her image and became a “training manual” for the neural network. After artificial intelligence to “understand”, as related to the various radio signals with varying pose of the person, the researchers took the camera. It turned out that a neural network can “see” human movement even through walls, although these experiments while training was conducted.

The study’s authors believe that the development can help to monitor patients with Parkinson’s disease and multiple sclerosis. Artificial intelligence can recognize more detailed features of the movements of patients and help doctors to find treatment. Also new technology can become a basis for creation of security systems for elderly people — a neural network placed in the house could fix of falling and other injuries and to inform the relatives or the doctors.


Please enter your comment!
Please enter your name here