Deep Learning-Based Detection of Lumpy Skin Disease in Livestock using CNNs

Authors

Walid Abdullah, Faculty of Computers and Informatics, Zagazig University, Zagazig, Sharqia, 44519, Egypt
Sudeep Tanwar, Department of Computer Science and Engineering, Institute of Technology, Nirma University, Ahmedabad, India
Mohamed Abouhawwash, Department of Industrial and Systems Engineering, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia

Abstract

Lumpy Skin Disease (LSD) is a viral disease affecting cattle, resulting in a large economic loss in the livestock industry. Early and accurate detection is essential to prevent outbreaks and minimize its impact. In this study, we explore the use of deep learning models for automated LSD classification from cattle images. We evaluate five state-of-the-art convolutional neural networks (CNNs) including MobileNetV2, ResNet50V2, DenseNet121, InceptionV3, and Xception using transfer learning. The dataset consists of images categorized as either healthy or infected, with data augmentation applied to improve model generalization. Performance is assessed based on accuracy, precision, recall, and F1-score. Among the tested models, DenseNet121 achieved the highest accuracy of 90.36%, outperforming the other architectures. The results demonstrate the potential of deep learning for rapid and reliable LSD detection, which could aid veterinarians and farmers in early disease identification.

How to Cite

Abdullah, Walid; Tanwar, Sudeep; and Abouhawwash, Mohamed (2025) "Deep Learning-Based Detection of Lumpy Skin Disease in Livestock using CNNs," Sustainable Machine Intelligence Journal: Vol. 11: Iss. 1, Article 6.
DOI: https://doi.org/10.61356/SMIJ.2025.11515
Available at: https://smij.sciencesforce.com/journal/vol11/iss1/6