MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641108270 A) filed by Prof. Yashaswini A R; Sudhindra Kumar; and Dr. Ranjith K C on September 09, 2026, for System And Method For Telemetry-Driven Hardware-Aware Edge Video Processing With Adaptive Ai Inference.
Inventors include Prof. Yashaswini A R; Sudhindra Kumar; and Dr. Ranjith K C.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: [065] The present invention provides a telemetry-driven hardware-aware edge video-processing system and method in which image preprocessing is dynamically coordinated with real-time operating conditions of an edge computing platform. A hardware telemetry monitoring module obtains processor, accelerator, memory, thermal, power, battery, storage, bandwidth or network information. An adaptive decision engine determines a hardware operating state and dynamically selects one or more Image Signal Processor (ISP) configuration parameters before Artificial Intelligence (AI) inference. The ISP preprocesses acquired image or video frames according to the selected configuration and supplies the preprocessed data to an AI inference engine. A feedback control module evaluates processing performance including latency, resource utilization, temperature, power consumption or inference-performance information and returns adaptive feedback to the adaptive decision engine for updating the ISP configuration in a subsequent processing cycle. The closed-loop arrangement permits preprocessing demand to be reduced under constrained hardware conditions and increased when resource headroom is available, thereby improving resource utilization, latency, energy efficiency and thermal stability while maintaining real-time AI inference. Accompanied Drawing [FIGS. 1 to 5].
Disclaimer: Curated by HT Syndication.
