MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202631052684 A) filed by National Institute Of Technology Silchar on April 24, 2026, for System And Method For Sign Language Recognition Using Deep Learning.
Inventors include Mr. Pranjal Gogoi; Ms. Bhumika Karsh; Mr. Sandipan Dutta; Dr. Ram Kumar Karsh; Prof. Rabul Hussain Laskar; and Prof. Manas Kamal Bhuyan.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: The invention relates to a system and method for recognizing sign language gestures corresponding to predefined medical conditions. An optical image capture unit (102) acquires video frames within a spatial boundary, while an image conditioning unit (104) converts the frames into normalized multi-channel spatial-temporal structures. A spatial feature extraction unit (106) employs three-dimensional convolutional processing, spatial down-sampling, channel adjustment, frame resizing, and feature extraction stages to derive motion representations while preserving frame ordering. A recurrent sequence temporal encoding unit (112), comprising stacked bidirectional memory units, captures forward and backward contextual dependencies across gesture sequences. A temporal alignment processing unit (108) derives normalized temporal importance scores and performs weighted aggregation to emphasize salient gesture transition frames. An attention evaluation unit generates attention-weighted sequence context representations. A classification processing unit (110) processes the feature signal through neural network stages to generate class probabilities corresponding to gesture categories including allergy, asthma, cough, diabetes, fever, injury, pain, tuberculosis, and vomiting.
Disclaimer: Curated by HT Syndication.
