Training Computers To Recognize People Through Body Parts

Training Computers To Recognize People Through Body Parts

By Shinji Tutoru

UNIVERSITY OF CALIFORNIA, CALIFORNIA, USA. A computer scientist is currently “training” computers to actually recognize people using three-dimensional flat photography. Once the training becomes successful, the computer will be able to identify a person not just with the facial features but also with the body features.

Unlike the typical face-recognition software which is being used for several years now, the new method will let the computer recognize people by detecting the human body parts such as the arms, torso and legs. That way, it will become easier to identify a person without needing a high-definition camera to distinguish the actual facial features from a distance. The process will also use lesser power than the traditional method.

This technology is being developed by Deva Ramanan, a computer scientist at University of California at Irvine. He also said that the process will become more like a divide-and-conquer approach. More than facial detection, it is actually meant for an accurate pedestrian-detection system. It will be used to track full-body movements even if the subject is far away.

Right now, Ramanan is focusing more on teaching the computer how to understand what a person is actually doing. The goal is to develop and improve on the computer’s reasoning ability until it gets closer to the human brain’s way of thinking in the future.

Invention Computers that recognize 3D humans in flat photography
Organization University of California, California, USA
Researcher Deva Ramanan
Field(s) Digital Photography, Face Recognition, Security, Pedestrian Detection System, Artificial Intelligence
Further Information POPSCI

Never miss an insight

Get insights delivered right to your inbox

More of Our Insights & Work
Popup Builder Mailchimp extension requires authentication.