Intellectual Property India Publishes Patent Application for 'System And Method For Detection Of Health Conditions And Disease Of Multi-Crops Using Deep Learning Convolutional Neural Networks And Machine Learning Models' Filed by Nitte Meenakshi Institute Of Technology, Nitte; Nalini N; P. N. Tengli; Hamsa G R; Geeta Budni; Harshith Kolluru; and Akarshak Sharma

MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641051254 A) filed by Nitte Meenakshi Institute Of Technology, Nitte; Nalini N; P. N. Tengli; Hamsa G R; Geeta Budni; Harshith Kolluru; and Akarshak Sharma, Bangalore, Karnataka, on April 22, for 'system and method for detection of health conditions and disease of multi-crops using deep learning convolutional neural networks and machine learning models.'

Inventor(s) include Nitte Meenakshi Institute Technology, Nitte; Nalini N; P. N. Tengli; Hamsa G R; Geeta Budni; Harshith Kolluru; and Akarshak Sharma.

The application for the patent was published on May 1, under issue no. 18/2026.

According to the abstract released by the Intellectual Property India: "The present invention relates to a system and method (100) for detecting health conditions and diseases of multiple crop species using a deep learning Convolutional Neural Network trained on the PlantVillage dataset. The system comprises a data input module (110) ingesting plant leaf images across 38 disease and healthy classes through Kaggle API connectivity with stratified 80:20 train-validation splitting; an image preprocessing engine (120) performing 224x224 resizing, pixel normalization, and on-the-fly augmentation via Keras ImageDataGenerator; a CNN classification engine (130) with two convolutional layers (32 and 64 filters) with ReLU activation and max pooling, a 256-neuron dense layer, and a 38-neuron softmax output layer; a model training module (140) using Adam optimizer with categorical cross-entropy loss; and a predictive inference engine (150) that preprocesses new leaf images, performs CNN forward pass, computes softmax probabilities, selects the top class via argmax, and maps it to a disease name with confidence score. The system achieves validation accuracy exceeding 90 percent across 38 classes spanning Apple, Grape, Tomato, Potato, Blueberry, Strawberry, and additional crop species, with single-image inference in 50-100 milliseconds on CPU, enabling practical real-time field deployment on mobile devices and edge computing platforms for early-stage disease diagnosis."

Disclaimer: Curated by HT Syndication.