MIT-developed software analyzes Wi-Fi signals to recognize human silhouettes through walls
MIT’s Computer Science and Artificial Intelligence Lab has developed software that analyzes the variations in Wi-Fi signals to recognize human silhouettes through walls, according to a report by Gizmodo.
In a study published by MIT, the researchers discuss the RF-Capture device, which transmits wireless signals and then analyzes the reflections of those signals to construct a human form.
The technology serves as an extension of another solution the researchers have been developing for a few years. The team previously used similar radio frequency technology back in 2013 to detect motion on the other side of a wall.
Two years later, the researchers have been able to advance RF-Capture’s capabilities to determine slight discrepancies in body shapes.
The software is able to distinguish between 15 different people through a wall with 90 percent accuracy, as well as determine a person’s breathing patterns and heart rate.
The device works by emitting wireless signals, which are sent through the wall and bounce off various parts of the individual’s body as he or she moves. During this interaction, the RF-Capture takes snapshots of the reflected signals.
The software uses an algorithm to identify body parts and construct the images into a silhouette of the moving figure.
The researchers said that RF-Capture could be used in various application cases, such as tracking the movements of an elderly person living alone as a safety precaution, or in a smart home to control certain appliances.
Meanwhile, the team expects the software’s accuracy to significantly improve over time.
The researchers also acknowledge that certain regulations would need to be put in place in order to alleviate any privacy concerns regarding the use of the software, as it could potentially be used to spy on others.