Researchers at the Massachusetts Institute of Technology (MIT) have developed a technology (via Gizmodo) that uses WiFi signals to “see” through walls, discerning different people, hand signals, and movements.
The device, which is named “RF-Capture,” transmits signals through the wall, which are then reflected back to a sensor. This creates an “image” of the person, which is clear enough to distinguish the identity of the individual, hand movements in the air, and how the person is moving.
Processing the imagery requires a lot of computing power as the signals capture background noise which distorts the image. To reduce this, researchers start by capturing a series of images of the scene. “At a high level, we suppress noise by combining information across time and fitting the data into a model,” Fadel Adi, one of the researchers, told Gizmodo.
All of the data is then collected and analysed. “The algorithms that we developed fit all of these snapshots into a coarse human model with major body parts — such as head, chest, arms, and feet,” said Adi. “That is, we combine these snapshots in a manner that maximizes the ability of the reconstructed silhouette in representing the human body.”
The team can discern a number of characteristics about a person from the imagery, including height and shoulder width which can be used to create an overall profile. This data can then be “trained” and used by researchers to spot patterns in body shapes and sizes.
“[W]e use the captured human silhouettes from our reconstruction algorithm [to] train a classifier on these silhouettes which allows us to distinguish between people,” Adi told Gizmodo. During tests, the system had an accuracy rate of over 90%.
The technology is only in the lab phase at present, but the team envisages uses such as scanning the house of elderly people and alerting an ambulance if they fall over.