MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641083368 A) filed by Rajalakshmi Engineering College on July 07, 2026, for Edge-Ai And Federated Learning System For Real-Time Pest And Disease Detection In Agriculture.
Inventor includes Dr. P. Kumar, Ms. Dhivyashree.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: This technology reveals a hardware-enhanced Edge-Al and Federated Learning network for instant pest and disease recognition in agriculture. Edge units with camera modules take pictures of crops and run them via an embedded pre-processing pipeline. A small neural model, which is made efficient by quantization and pruning, is used to perform inference directly on the device This allows for the pest/disease identification to be done immediately with very little latency. In order to be able to keep the model updated all the lime and at the same time not send the images, each device calculates the local gradients and then sends them through an adaptive communication stack that supports LoRaWAN, Wi-Fi, 4G, or 5G. The Federated Learning Coordinator gets local updates from several devices, merges them, and sends the updated global model back. The invention is characterized by technical effects such as decreased memory access time, lower bandwidth consumption, increased inference speed, and better device energy efficiency. The system is designed to be implemented in different agricultural settings with unstable connectivity and limited computational resources.
Disclaimer: Curated by HT Syndication.
