MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202621093403 A) filed by Rk University on July 31, 2026, for Computer Vision And Deep Learning-Based Intelligent Yoga Pose Detection And Training Assistance System.
Inventors include Amit Lathigara; Nirav Bhatt; Paresh Tanna; Chetan Shingadiya; and Anju Kakkad.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: Abstract COMPUTER VISION AND DEEP LEARNING-BASED INTELLIGENT YOGA POSE DETECTION AND TRAINING ASSISTANCE SYSTEM A system for computer-vision-based yoga training assistance includes a Camera and Display Terminal, a Deep Learning Pose Network, a Occlusion Confidence Unit, a Reference Pose Matcher, and a Overlay Guidance Generator. The system receives video frames, skeletal keypoints, occlusion indicators, reference-pose representations, and training-stage information. The acquired observations are checked for quality, source association, and timing before a model-ready representation is generated. The Occlusion Confidence Unit produces an analytical result and a confidence indication. The Reference Pose Matcher applies a stored response criterion and determines whether to release an output, request reacquisition, or require review. The Overlay Guidance Generator provides a recognized pose, an alignment overlay, and a next- step training prompt. The system particularly performs maintaining confidence-aware pose tracking during partial occlusion and generating spatial overlay guidance. An event record stores input identifiers, a processing version, the analytical result, and the delivered response to maintain traceability between acquisition, processing, and output. Fig. 1 Dated July 30, 2026 / Mungalpara Jigneshbhai Chhaganbhai IN/PA-2640 Agent for the Applicant
Disclaimer: Curated by HT Syndication.
