BrainChip and Tata Consultancy Services unveil gesture recognition technology at NeurIPS 2019
BrainChip, in partnership with Tata Consultancy Services (TCS), will showcase its hand gesture recognition technology using Dynamic Vision Sensor (DVS) spiking input on the Akida Neuromorphic Technology Platform at Neural Information Processing Systems conference (NeurIPS 2019) in Vancouver this month, the company announced.
The technology is outlined in a research paper named “Human Gesture Recognition using Spiking Input on Akida Neuromorphic Platform,” written by Sounak Dey, Arijit Mukherjee from TCS Research and Gilles Bézard, Douglas McLelland from BrainChip. The demonstration will include a presentation of the research.
The Akida Neuromorphic System-on-Chip (SoC) Technology will be used to recognize and classify hand gestures of people attending the presentation. A Dynamic Vision Sensor camera will detect hand gestures and hand positions and will leverage the Akida platform for unsupervised live learning and classification. BrainChip claims the spiking neural network (SNN) and the Akida neuromorphic chip need less power and data compared to a traditional neural network (DNN).
Akida can be integrated as a licensable IP technology into ASIC devices and will be ready to use as an SoC , with both deployments available for surveillance, advanced driver assistance systems (ADAS), autonomous vehicles (AV), vision guided robotics, drones, augmented and virtual reality (AR/VR), acoustic analysis, and Industrial Internet of Things applications. It executes neural processing and memory accesses on the edge which saves computing resources needed for the system host CPU and optimizes functions such as training, learning, and inferencing.
“While the recognition of a simple hand gesture might at first seem simple, it is in fact quite a revolutionary advance in the state of today’s human and robot interactive environments,” said Roger Levinson, COO of BrainChip. “We are pleased to have the opportunity to showcase the extent of this revolutionary advance at NeurIPS and are eager to show how an Akida-based platform can be utilized to ingest imagery, train the system to recognize what it has seen and learn in a way that is vastly more efficient and accurate than what other solutions have been able to achieve thus far.”
In November, BrainChip was granted a U.S. patent for its dynamic neural networks, a valuable feature of its AI processing chip Akida.