Biometric-enabled Personalized Augmentative and Alternative Communications
About
This study focuses on the roadmapping of biometric technologies onto personalized Augmentative and Alternative Communication (AAC), a branch of assistive technologies for people with communication disabilities. This technology roadmapping revolves around the proposed notions of an AAC biometric register and biometric-enabled reconfigurable AAC channels. The biometric register is referred to as a tool for acquiring and processing physiological and behavioural traits that are essential for augmentative and alternative communication. It links biometric traits, such as gestures, to intermediate traits, such as synthesized speech, for customizable communication channels. The proposed methodology is used to assess the gaps between the social and practical demands, such as assisting people with communication disabilities in the contemporary semi-automated border control, and the emerging advances in AI, such as advanced video and speech processing. We provide two case studies of the AAC that rely on hand gesture recognition and sign language word recognition, and conclude that the current accuracy of those AI technologies does not meet the practical requirements. The proposed roadmapping provides recommendations for further improvement to close these gaps.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Sign Language Recognition | SLR500 | -- | 19 | |
| Word-level sign language recognition | WLASL 300 | -- | 11 | |
| Word-level sign language recognition | WLASL 2000 | Top-1 Acc69 | 8 | |
| Gesture Recognition | DHG-14/28 | -- | 6 | |
| Sign Language Recognition | WLASL100 | -- | 5 | |
| Sign Language Recognition | WLASL 1000 | -- | 3 | |
| Gesture Recognition | Static HAnd PosE (SHAPE) Dataset | -- | 1 | |
| Gesture Recognition | NTU-RGBD | -- | 1 | |
| Gesture Recognition | Kinetics-Skeleton | -- | 1 | |
| Sign Language Recognition | ASLLVD Dataset | -- | 1 |