Even if you’re alone in a room

[rakhirani458]

In a broader sense, computer vision is becoming the sum of sensor data that can be represented as visual patterns. With the advancement of AI, it is becoming possible to create a highly accurate visual portrait based on the body’s reflected radio frequency signals, the pressure on the supporting surface, the vibrations created by a person when walking, and their thermal radiation. All of these, individually and together, represent a unique signature that can be used to “see” a person, even if they successfully hide their face, voice, fingerprints, and genome from prying eyes.

It may soon no longer be necessary to install cameras everywhere to piece together a good enough picture of what's happening indoors and outdoors. Here's a quick look at the innovations emerging from recent research, most of which use advanced AI.

Reflective Wi - Fi Recognition . , the door is closed, and the curtains are drawn, you can be identified by the way your body reflects Wi-Fi signals. Researchers at MIT have developed a india mobile database that combines a Wi-Fi emitter, sensors, and AI algorithms. It models the actions of a person on the other side of an opaque barrier. Like sonar, the technology, called RF-Pose, identifies 2D shapes of people and other objects based on the patterns of Wi-Fi signals they reflect. When correlated and cross-trained with AI applications that recognize gait, gestures, and motion, these shapes can identify people 83% of the time.

Gait recognition based on ground pressure. The University of Manchester has created an AI-based gait recognition system called SfootBD with almost 100% accuracy. It uses a passive sensor to analyze weight distribution, speed, and walking style. Ground pressure signals are correlated with the walking style captured by a high-resolution camera. To train the AI, the researchers created a database of gait signals from more than 120 people, measuring floor pressure. The data was collected in public places (airport checkpoints, workplaces) and at home. The algorithm was tested on a control group of gait imitators, as a result of which it was able to recognize attempts to copy someone else's gait.