A wrist-worn system for 3D hand pose estimation has been developed by researchers at the Tokyo Institute of Technology (Tokyo Tech) working in partnership with colleagues at Carnegie Mellon University , the University of St Andrews and the University of New South Wales. The device consists of a camera that captures back-of-hand images and is assisted by a neural network called DorsalNet that can interpret complex gestures accurately.
In the development of augmented reality (AR ) and virtual reality (VR) devices, which are already beginning to be very common in the medical, sports and entertainment industries, being able to monitor hand movements is of crucial importance. To date, the use of bulky data gloves has been involved in these applications, which appear to impede natural motion or controllers with a restricted sensing range.
Now, at Tokyo Tech, a research team led by Hideki Koike has developed a camera-based wrist-worn 3D hand pose recognition device that could be on par with a smartwatch in the future. Their system can dramatically allow hand movements in mobile settings to be recorded.
“This work is the first real-time 3D hand pose estimator based on vision using visual features from the region of the dorsal hand,” the researchers say. The device consists of a camera assisted by a neural network called DorsalNet that, by detecting changes in the back of the hand, can accurately estimate 3D hand poses.
The researchers confirmed that their method exceeds previous work with an average of 20 percent higher accuracy in identifying complex movements and achieves an accuracy of 75 percent in detecting eleven different forms of grasp.
The work could advance the development of controllers that facilitate bare-hand communication. In preliminary studies , the researchers showed that it would be possible to use their method to monitor smart devices, such as altering the time on a smartwatch by simply adjusting the angle of the finger. They also demonstrated that the device could be used as a virtual mouse or keyboard, such as by rotating the wrist to control the pointer location and using a basic 8-key keyboard for input typing.
They point out that more device changes would be needed for real-world use, such as using a camera with a higher frame rate to capture fast wrist movement and to be able to cope with more diverse lighting conditions.
Source of Story: The Tokyo Institute of Technology provides materials. Note: For style and length, material can be edited.
No Comments
Leave a comment Cancel