Abstract: The present invention relates to a system and method for real-time sign language recognition. The invention provides a motion-driven framework configured to process continuous video streams capturing a user performing sign gestures. Positional coordinates corresponding to anatomical landmarks are extracted from each video frame, and a motion energy value is computed to identify frames containing significant gesture activity. Selected frames are encoded into gesture tokens representing spatial and temporal features of the user. The gesture tokens are sequentially processed using a temporal modeling module to generate a temporal representation of the gesture. Class probabilities for predefined sign gesture categories are computed, and a recognition result is generated based on a predefined confidence threshold. Figure 1 is the illustrative figure.
Description:FORM-2 IS ATTACHED , Claims:ATTACHED IN FORM-2
| # | Name | Date |
|---|---|---|
| 3 | 202641023159-FIGURE OF ABSTRACT [26-02-2026(online)].pdf | 2026-02-26 |
| 4 | 202641023159-EVIDENCE FOR REGISTRATION UNDER SSI(FORM-28) [26-02-2026(online)].pdf | 2026-02-26 |
| 5 | 202641023159-EDUCATIONAL INSTITUTION(S) [26-02-2026(online)].pdf | 2026-02-26 |
| 6 | 202641023159-DRAWINGS [26-02-2026(online)].pdf | 2026-02-26 |
| 7 | 202641023159-DECLARATION OF INVENTORSHIP (FORM 5) [26-02-2026(online)].pdf | 2026-02-26 |
| 8 | 202641023159-COMPLETE SPECIFICATION [26-02-2026(online)].pdf | 2026-02-26 |
| 9 | 202641023159-Proof of Right [02-03-2026(online)].pdf | 2026-03-02 |
| 10 | 202641023159-FORM-9 [20-04-2026(online)].pdf | 2026-04-20 |
| 11 | 202641023159-PATENT_APPLICATION_PUBLICATION.pdf | 2026-05-02 |
| 12 | 202641023159-FORM-26 [07-05-2026(online)].pdf | 2026-05-07 |