MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202621110809 A) filed by Dr. Khushbu Arvindbhai Patel; Dr. Nileshkumar L Patel; Mr. Vishal Rajendrabhai Patel; Dr & Lr Hirenkumar Rameshbhai Patel; Dr. Ankitkumar Ratilal Patel; Dr. Mahammad Idrish I. Sandhi; Dr. Mamta P. Khanchandani; and Dr. Upasana S. Mehta on September 15, 2026, for Privacy-Preserving On-Device Visual Recognition And Waste-Categorization System With Confidence-Guided Recycling Recommendation And Local Eco-Impact Recording.
Inventors include Dr. Khushbu Arvindbhai Patel; Dr. Nileshkumar L Patel; Mr. Vishal Rajendrabhai Patel; Dr & Lr Hirenkumar Rameshbhai Patel; Dr. Ankitkumar Ratilal Patel; Dr. Mahammad Idrish I. Sandhi; Dr. Mamta P. Khanchandani; and Dr. Upasana S. Mehta.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: A privacy-preserving on-device system and method for visual categorization of waste materials is disclosed. An image received from a camera or uploaded file is preprocessed and classified locally by a trained visual classification model to generate candidate labels and confidence values. A waste-domain mapping engine applies ordered keyword or label rules to transform a general-purpose visual output into a waste-specific category and associated recycling, reuse, upcycling or recovery action. A confidence decision component permits automatic categorization only when a confidence threshold and valid domain mapping are satisfied; otherwise a manual category-selection interface is provided. Upon user confirmation, a local bucket and eco-impact engine records the categorized waste and updates a quantifiable eco- impact score. The core recognition pipeline may operate entirely on a client device without requiring transmission of the captured image to a remote classification server. Total Number of Claims: 10 Total Number of Sheets of Drawing: 4
Disclaimer: Curated by HT Syndication.
