MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202621063765 A) filed by Vishwakarma Institute Of Technology, Pune on May 20, 2026, for An Ai-Enabled Writing Pad & Stylus For Students With Special Learning Disabilities..
Inventors include Gaikwad Vijay Dattatray; Desai Neel Sachin; Deshkar Sukhada Jitendra; and Deshmukh Kshitij Amit.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: The invention relates to an AI-enabled smart writing pad for supporting students with Specific Learning Disabilities (SLDs), including dyslexia and dysgraphia, by combining stylus-based handwriting capture, on-device machine learning, and mode-dependent educational feedback. The system comprises a touch-sensitive display coupled to an embedded processing unit (for example a single-board computer class device) that receives handwritten input in real time and performs convolutional neural network (CNN) based multi-class alphanumeric classification, with optional word-level segmentation, per-character inference, and vocabulary-assisted spelling repair in a Recognition-oriented mode intended to produce readable output for teachers and parents. In parallel, a Training-oriented mode uses the same recognition backbone but applies a distinct policy that does not silently auto-correct student writing and instead provides explicit, instruction-aligned feedback together with structured literacy exercises and local persistence of session and progress data for adaptive remediation. For robust evaluation under classroom-like conditions, a representative embodiment is assessed on the combined EMN1ST ByClass test set (digits and letters), comprising 126,323 samples across 62 classes. Quantitative evaluation using one-versus-rest confusion-matrix analysis reports an overall accuracy of 0.8757, with micro precision/recall/F1 of 0.8757/0.8757/0.8757, macro precision/recall/F 1 of 0.7571/0.7191/0.7147, and weighted precision/recall/Fl of 0.8483/0.8757/0.8524, indicating strong aggregate recognition together with class- imbalance-sensitive behaviour relevant to confusion-prone symbol pairs common in dyslexic writing. The invention further incorporates dyslexia-aware orientation analysis for reversal-prone characters and supports longitudinal analytics for monitoring improvement trends. Accordingly, the disclosed system provides alow-cost, on- device assistive platform for handwriting interpretation, honest cognitive training, and data-driven educational intervention for learners with SLDs.
Disclaimer: Curated by HT Syndication.
