Deep Learning Mechanisms for Intelligent Lumpy Disease Detection for Enhancing Rural Livelihoods, Sustainable Farming, and Farmer Support System

Abstract

Lumpy skin disease (LSD) poses a significant threat to cattle farming, causing economic losses due to reduced productivity and hide damage. As of 2022, more than 67,000 cattle had died across the country. States like Rajasthan and Gujarat were heavily affected, with over 25,000 cases reported in Rajasthan alone and more than 37,000 cases in Gujarat. This viral disease, primarily transmitted by biting flies, necessitates early detection for effective management. In this project, we propose a novel approach for early detection of LSD in cattle using deep learning techniques applied to image analysis. Our methodology involves gathering a diverse dataset of cattle images, including both healthy and infected animals of various breeds, ages, and environmental conditions. The proposed healthcare system helps the farmers to check the early lumpy disease in cattle. Also, the AI based approach using multiple deep learning algorithms, including CNN, ResNet, and EfficientNet for disease detection can make the disease detection. Second stage we are providing the recommendation system for the treatment of cattle and providing the roadmap how the treatment will be and which hospital we can prefer based on the location. We are creating the system, where we employ advanced techniques for lumpy skin disease (LSD) management can involve a combination of traditional veterinary approaches and modern technologies. Here are some advanced techniques that can be employed. This proposed system will help farmer society who are suffering from the financial losses due to the diseases in cattle .The project provides the proper disease detection and recommendation system for the treatment by adopting the one health approach to lumpy skin disease management, involving collaboration between veterinary professionals, public health experts, and farmers. This interdisciplinary approach recognizes the interconnectedness of animal, human, and environmental health and aims to address complex health challenges holistically. The experimental results reveal that the Enhanced CNN algorithm achieved a training accuracy of 87.31%. In comparison, ResNet demonstrated a higher accuracy of 92.66%, while EfficientNet V2B0 outperformed both with an impressive training accuracy of 93.31%.