A new spike-based image model enables accurate hand gesture recognition by capturing dynamic motion patterns using neural-inspired data processing.

The recorded HD-sEMG are decomposed into channel-wise CSTs, transformed into cwCST-image according to the channel distribution, and decoded to different gestures by CNN-based model. The CSTs in each ...

If this classification is successful, subsequent steps in the hand gesture recognition process only compare the input gesture with stored samples of the same hand type.