Researchers at Indian Institute of Information Technology Allahabad (IIIT-A) have developed CVGG-16, a new and efficient AI-Model to detect crop diseases. The technology allows farmers to detect crop diseases directly in their fields without any oversight from experts.
This solution combines the use of Artificial Intelligence (AI), Internet of Things (IoT), Deep Learning, and Federated Learning to inspect plant health based on the images of leaves and other environmental conditions. This CVGG-16 AI Model has been developed by Pramod Kumar Singh, a research scholar at IIIT-A, working under Associate Professor Dr. Manish Kumar.
The model has been published in the international journal Internet of Things by Elsevier. The CVGG-16 model was trained using real world images from farms that included dusty, low-light, and weather-affected conditions. The CVGG-16 does not solely relies on photos, rather works on a multi-concept data fusion network. The technology integrates data from soil moisture, temperature, humidity and weather patterns that enables it to deliver a highly accurate analysis of the concerned plant.
For testing the viability of the model, we collected the sample data of the farms around Prayagraj, and we can assure that the model not only saves the precious time of the farmers (when it comes to identifying the health of the crop) but also provides precise information regarding any diseases very much in the preliminary stage, which helps in containing the same in an effective manner
Also Read: India Deploys Space Tech for Agricultural Forecasting, Drought Monitoring & Insurance
CVGG-16 for Smarter Agriculture
The CVGG-16 model integrates Federated learning, ensuring data privacy for farmers, and it is assisted by a central server that uses using a new algorithm named Extreme Client Aggregation. The CVGG-16 is a locally trained model and the test results have shown an accuracy rate of 97.25%. The model showcased 96.75% accuracy in identifying maize diseases and 93.55% in potato disease detection. The model has been designed to integrate use of datasets from different geographical regions , while keeping data decentralization as a primary objective.
Dr. Manish Kumar has stated that, having access to picture of entire farmland, it is easy to spot culprit even in deepest part of the crop. He also hailed CVGG-16 for its scalability and adaptability, stating that it can be applied to any crop in any region across India. The model works on a multi-layered structure allowing it to adjust to varied agricultural conditions.
Dr. Manish Kumar stated that the team of researchers is working to deliver mobile applications and local language support to make the technology more accessible, especially for small and marginal farmers. He emphasized that CVGG-16 could become a game-changer for Indian agriculture by giving farmers real-time, intelligent disease diagnostics, leading to healthier crops and reducing their reliance on external experts.