Advances in Gesture Recognition and Sign Language Technology

The field of gesture recognition and sign language technology is rapidly advancing, with a focus on developing more accurate, efficient, and user-friendly systems. Researchers are exploring new architectures and techniques, such as lightweight transformer models and editable speech-to-sign language transformers, to improve the performance and accessibility of these systems. Notable advancements include the development of real-time sign language recognition systems, harmonization of complementary pose modalities for coherent sign language generation, and markerless handheld augmented reality frameworks. These innovations have the potential to significantly improve communication and interaction for individuals with hearing or motor impairments. Noteworthy papers include: SLRNet, which demonstrates the feasibility of inclusive, hardware-independent gesture recognition with a validation accuracy of 86.7%. WaveFormer, a lightweight transformer-based architecture that achieves 95% classification accuracy on the EPN612 dataset with a 6.75 ms inference latency. SignAligner, which significantly improves the accuracy and expressiveness of generated sign videos through a novel method for realistic sign language generation. GHAR, a markerless HAR framework that offers improved usability, manipulability, and comprehensibility for architectural building models. Design of an Editable Speech-to-Sign-Language Transformer System, which enables direct user inspection and modification of sign segments, thus enhancing naturalness, expressiveness, and user agency.

Sources

SLRNet: A Real-Time LSTM-Based Sign Language Recognition System

WaveFormer: A Lightweight Transformer Model for sEMG-based Gesture Recognition

SignAligner: Harmonizing Complementary Pose Modalities for Coherent Sign Language Generation

GHAR: GeoPose-based Handheld Augmented Reality for Architectural Positioning, Manipulation and Visual Exploration

Design an Editable Speech-to-Sign-Language Transformer System: A Human-Centered AI Approach

Accessible Gesture-Driven Augmented Reality Interaction System

Built with on top of