Lumpy Skin Disease Detection in Cattle: A Machine Learning and Deep Learning Approach

Lumpy Skin Disease (LSD) is a viral infection in cows that can easily be transmitted from one animal to another leading to great losses in the potential milk production capacity as well as to the quality of the meat, accompanied by abortions and cases of sterility.Fast and accurate detection is crucial for implementing and organizing control measures to prevent spread, but classical diagnostic methods are time-consuming, labor-intensive, and do not usually cover inaccessible rural areas.This work proposes an automated identification of the LSD situation in cattle by using the dataset of images through artificial intelligence (AI) as a solution.Cow Lumpy Disease Dataset, the dataset created by Kaggle, comprising the images of healthy and infected cows, was utilized to train and test the deep learning models.The transfer learning method, which is to say, the technique to transfer the learned knowledge of one domain to the domain that we are interested in, was implemented in the architectures of MobileNetV2, InceptionV3, and Xception to differentiate the cow images with high accuracy.Besides these, such regression models as soft voting and stacking also were involved in the process to boost the performance of the prediction.The results of the tests have demonstrated that ensemble models were much more successful than individual models and the combination ensemble could almost reach the rate of 97%.The results show that the application of deep learning and ensemble strategies is capable of creating a very accurate, reliable, and very fast diagnostic tool for LSD.This mechanism, driven by AI, can offer assistance to veterinarians and farmers in the early stages of the disease, thus helping them to take steps quickly and manage the disease well to minimize the economic losses when the situation of lacking access to the veterinary services is real.