MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202621054337 A) filed by Symbiosis International Deemed University on April 28, 2026, for Bio-Inspired Genetic Algorithm Optimized Convolutional Neural Network System For Automated Weed Detection In Agricultural Farmland.
Inventors include Dr. Snehlata Wankhade; Gargi Balamwar; Karan Misal; and Anshuma Padole.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: ABSTRACT BIO-INSPIRED GENETIC ALGORITHM OPTIMIZED CONVOLUTIONAL NEURAL NETWORK SYSTEM FOR AUTOMATED WEED DETECTION IN AGRICULTURAL FARMLAND The present invention provides a bio-inspired Genetic Algorithm optimized Convolutional Neural Network system (100) for automated weed detection in agricultural farmland. The system comprises an image acquisition module (110) for capturing field images, a preprocessing module (120) for image normalization and augmentation, a CNN classification module (130) for feature extraction and weed-crop classification, a Genetic Algorithm optimization module (140) for automated hyperparameter tuning through evolutionary operations including selection, crossover, and mutation, and an output interface module (150) for displaying results. The GA optimization module (140) automatically determines optimal values for learning rate, dropout rate, and filter configurations, eliminating manual trial-and-error tuning. The system achieves classification accuracy exceeding 96% on benchmark datasets while maintaining computational efficiency. The invention enables precision agriculture applications with reduced herbicide usage and improved crop productivity. [
Disclaimer: Curated by HT Syndication.
