Sign Language Recognition: New Developments, Technologies and Challenges (Computer Science, Technology and Applications) - Tapa blanda

Sinopsis

This book surveys the fast-moving field of sign language recognition—from core vision models to inclusive, real-time translation—showing how today’s tools are crossing the gap between Deaf and hearing communities. It blends multimodal linguistics with deployable AI, emphasizing browser-first systems, mobile friendliness, and low-resource language settings. Early chapters introduce e-SALIN, a web platform that translates Tagalog speech into Cebuano, Ilocano, Waray, and Filipino Sign Language (FSL) images, grounding technical design in accessibility and gender-fair language practices. Readers then see the companion pipeline that recognizes continuous FSL and renders Tagalog text with a CNN–LSTM model achieving real-time performance and high accuracy—evidence that practical, local-language SLR is already feasible. Mid-book studies compare YOLOv5/YOLOv8 and Roboflow-trained detectors for static alphabets and common signs, reporting strong mAP/F1 trade-offs and millisecond-scale inference suitable for classrooms and public services. Results highlight YOLOv8 as a balanced choice (e.g., F1≈1.00 for alphabets; ≈0.98 for single-word sets), with guidance on dataset prep and deployment. The closing chapter elevates non-manual markers (NMMs)—facial expressions and head movements—from optional cues to first-class grammatical features, and formalizes a user-centric latency metric (“prediction time”).

"Sinopsis" puede pertenecer a otra edición de este libro.