New dataset and models advance sign language handshape recognition

Researchers have developed a new dataset and baseline models for fine-grained isolated handshape recognition in sign language, utilizing the HamNoSys notation system. The dataset comprises 144,000 RGB images from 15 participants across 160 handshape classes. Evaluations using models like ResNet-18 and ViT-B/16 demonstrated reproducible performance in subject-dependent tests, but a significant drop occurred when generalizing to unseen participants, highlighting the challenges in creating accessible sign-language technologies. AI

IMPACT This research provides a valuable resource for developing more accessible sign-language technologies by improving computational transcription and translation.

RANK_REASON The item is an academic paper detailing a new dataset and baseline models for a specific AI task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• ASL Fingerspelling Dataset A
• HamNoSys
• HamNoSys 4 Handshapes Chart
• LSWH100
• ResNet-18
• ViT-B/16
• XGBoost

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →