A human gesture recognition model for the American Sign Language (ASL) recognition scenario (word-level recognition). The model uses an S3D framework with MobileNet V3 backbone. Please refer to the MS-ASL-100* dataset specification to see the list of gestures that are recognized by this model.
The model accepts a stack of frames sampled with a constant framerate (15 FPS) and produces a prediction on the input clip.
|Top-1 accuracy (MS-ASL-100*)||0.847|
input, shape: [1x3x16x224x224]. An input image sequence in the format [BxCxTxHxW], where:
The model outputs a tensor with the shape [Bx100], each row is a logits vector of performed ASL gestures.
[*] Other names and brands may be claimed as the property of others.