MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202611110895 A) filed by Greater Noida Institute Of Technology on September 16, 2026, for System And Method For Lightweight Deep Learning-Based Detection And Classification Of Wheat Leaf Diseases Using An Attention-Enhanced Mobilenetv2 Architecture.
Inventors include Isteyaque Alam; Pranshu Yadav; Pawan Kumar Hansda; Mohd Zafar; Imran Ansari; Arjun Singh; and Asif Khan.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: A wheat leaf disease detection and classification system (200) and method are disclosed, comprising an image acquisition module (210), a preprocessing module (220), and an augmentation module (226) that supply images to a pretrained lightweight convolutional neural network backbone (230), such as MobileNetV2. Shallow convolutional refinement layers (240) with progressively increasing filter depth, each followed by batch normalization and maxpooling, further process the extracted features. An efficient channel attention module (250) adaptively reweights the refined feature channels, and a global context module (260), such as a RepMLP module, captures long-range contextual dependencies to improve discrimination between visually similar disease symptoms. A classification head (270), comprising a fullyconnected layer (272), a dropout layer (274), and a SoftMax output layer (276), outputs a probability distribution over a plurality of predefined wheat disease classes (280), including yellow rust, brown rust, powdery mildew, leaf blight, smut, and healthy leaves. The combination of a lightweight backbone with attention and global-context refinement enables accurate disease classification while remaining computationally suitable for deployment on smartphones, drones, and other resource- constrained agricultural monitoring devices.
Disclaimer: Curated by HT Syndication.
