MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202611107304 A) filed by Chandigarh University on September 07, 2026, for Automated Multi-Cnn System For Multi-Crop Plant Disease Classification.
Inventors include Kamalpreet Kaur; Laksheta; Anupreet Kaur; Vinod Kumar; and Parveen Bajaj.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: The present invention provides an automated multi-CNN training and selection framework for multi-crop plant disease classification from foliar images. The framework comprises a pre-processing module configured to resize, augment, and normalize images; a multi-architecture training module configured to train Baseline CNN, VGG16, ResNet50, and DenseNet121 using transfer learning; an automated model selection module configured to evaluate validation accuracy, loss, and per- class performance; and a deployment module configured for real-time offline disease diagnosis. The framework processes 37,107 images representing 18 disease classes across cassava, potato, and rice crops and automatically selects ResNet50 as the optimal classifier. ResNet50 achieves 99.88% validation accuracy with only 25 misclassifications, representing a 91% error reduction over the baseline. The framework maintains greater than 99% accuracy under severe class imbalance and varying field conditions, enabling robust deployment in resource-limited agricultural environments.
Disclaimer: Curated by HT Syndication.
