Wearable technology is capable of tracking various measures of human health and is getting better all the time. That could mean real-time feedback on posture and body mechanics, new research suggests. A Cornell University research team has demonstrated this capability in a novel wrist camera system, which it hopes to apply to future smartwatches.
The system, called BodyTrak, comes from the same lab behind the face-tracking wearable we saw earlier this year, which was able to reproduce facial expressions on a digitized body via sonar. This time, the team utilized a tiny micron-scale RGB camera and a custom artificial intelligence to build a model of the entire body.
Worn on the wrist, the camera passes basic images of body parts in motion to a deep neural network trained to turn the segments into virtual representations of the body. Working in real-time, the network filled in the gaps left by the camera images to build 3D models of the body in 14 different poses.
The researchers demonstrated the potential of their BodyTrack system with the help of nine participants who wore the cameras while performing everyday activities such as walking, sitting or exercising. With a single camera worn on the wrist and aimed at the body, the system was able to build a complete model with an average localization error of 6.9 cm (2.7 inches).
“Our study shows that we don’t need our body frame to be fully within the camera’s field of view for body sensing,” said Hyunchul Lim, lead author of the study. “If we can capture just a part of our body, a lot of information can be inferred to reconstruct the entire body model.”
This technique could have potential in next-generation motion capture systems that traditionally involve expensive equipment to track the complex mechanics of moving bodies. The team imagines it could also find use in smartwatches, which it can track body position during physical activity in real-time. They point out that this is still a distant goal, as current smartwatches don’t have the necessary cameras and hardware for full-body sensing. Still, they’re excited about the potential.
“Since smartwatches already have a camera, technologies like BodyTrak can understand the user’s posture and give real-time feedback,” said Cheng Zhang, senior author of the paper. “It’s convenient and affordable, and doesn’t limit the user’s area of movement.”
