A deep learning approach to detect lumpy skin disease in cows

Abstract

Diseases in cows are an influential point for human concern. There are some diseases in animals identified in the early phases that can be diagnosed and cured in the early phases of the disease itself. The effect of lumpy skin disease can cause large capital losses in the farm animal industry if it is not taken care of properly. The main reason for this disease is the lumpy skin virus, and this virus is a part of the Poxviridae family. The major symptom of lumpy skin disease is the Neethling strain, and other symptoms are a few mild forms of circumscribed skin nodules. These symptoms also include mucous membranes of internal organs like respiratory organs and reproductive organs. By the infection of such disease, animals like cattle get their skin permanently damaged. Some of the detrimental outcomes of this disease in cows are reduction in milk projection, infertility, poor growth, abortion and sometimes death. In this research work, an architecture using machine learning techniques to detect the disease is proposed. This architecture employs the pre-trained models like VGG-16, VGG-19 and Inception-v3 for feature extraction and then followed by multiple classifiers. The work is tested on our manually collected dataset, and the extracted features were further classified using the classifiers like kNN, SVM, NB, ANN and LR. Using this methodology, the state-of-the-art solution obtaining a classification accuracy of 92.5% over the test dataset.