Fingerspelling

Rethinking Generalization in American Sign Language Prediction for Edge Devices with Extremely Low Memory Footprint
A model of just 185KB is proposed for ASL prediction on ARM Cortex-M7 microcontroller, which works at 20 fps. Interpolation augmenation is shown to be effective for improving classification generalization.
PDF Cite Video Arxiv Abstract
Rethinking Generalization in American Sign Language Prediction for Edge Devices with Extremely Low Memory Footprint