Intellectual Property India Publishes Patent Application for Intelligent Image Processing System And Method For Real-Time Object Detection And Context- Aware Scene Analysis

MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641108271 A) filed by Ms. S. Nithiya; Ms. K. Sornalakshmi; Mr. S. Murugaanandam; Ms. V. Preethi; Ms. T. Kavitha; Dr. Inderpreet Kaur; Mr. W. Stanley Karunakaran; Ms. A. Kalaivani; Ms. R. Deepa; and Mr. Abijith G. R. on September 09, 2026, for Intelligent Image Processing System And Method For Real-Time Object Detection And Context- Aware Scene Analysis.

Inventors include Ms. S. Nithiya; Ms. K. Sornalakshmi; Mr. S. Murugaanandam; Ms. V. Preethi; Ms. T. Kavitha; Dr. Inderpreet Kaur; Mr. W. Stanley Karunakaran; Ms. A. Kalaivani; Ms. R. Deepa; and Mr. Abijith G. R..

The application for the patent was published on September 18, 2026, under issue no. 38/2026.

Abstract: An intelligent image processing system and computer-implemented method are disclosed for real-time object detection and context-aware scene analysis. Image frames are processed through an adaptive preprocessing engine and a multi-resolution feature engine to obtain first-pass object detections. For a candidate object or image region, a composite inference condition is determined from at least two of detection uncertainty, visual novelty, local image quality, temporal inconsistency, contextual inconsistency, and scene change. When an adaptive criterion is satisfied, only an affected region is selectively re-inferred using an alternate resolution, preprocessing operation, feature depth, detector configuration, or inference precision, and the alternate result is reconciled with the first-pass result. Detected objects are associated across frames and used to incrementally maintain a contextual scene graph and an event memory. A resource-aware inference scheduler dynamically allocates processing effort according to latency, computing state, scene complexity, and region priority. The system thereby concentrates computational effort on ambiguous or significant portions of a scene while preserving temporal and contextual continuity for generation of object metadata, scene interpretations, alerts, or control outputs. Accompanied Drawing [FIGS. 1-2].

Disclaimer: Curated by HT Syndication.