Early Detection of Lumpy Skin Disease in Cattle Using Deep Learning-A Comparative Analysis of Pretrained Models.

Area of Science:

• Veterinary Medicine
• Computer Science
• Artificial Intelligence

Background:

• Lumpy Skin Disease (LSD) severely impacts cattle health and agricultural economies.
• Early and accurate detection of LSD is crucial for effective outbreak management.
• Existing diagnostic methods may lack the speed and accuracy needed for widespread application.

Purpose of the Study:

• To develop and evaluate an automated system for Lumpy Skin Disease detection in cattle using deep learning.
• To compare the performance of various deep learning models for LSD identification.
• To assess the clinical applicability of AI-driven diagnostic tools in veterinary medicine.

Main Methods:

• Utilized publicly available datasets of healthy and LSD-affected cattle images, including images of other diseases.
• Employed image preprocessing, data augmentation, and dataset balancing techniques.
• Evaluated ten pretrained deep learning models (e.g., VGG16, MobileNetV2) using transfer learning and assessed performance metrics (accuracy, sensitivity, specificity).

Main Results:

• VGG16 and MobileNetV2 demonstrated high performance, with accuracies of 96.07% and 96.39%, respectively.
• Sensitivity reached 93.75% for VGG16 and 98.57% for MobileNetV2.
• Specificity was 97.14% for VGG16 and 94.59% for MobileNetV2, indicating robust detection capabilities.

Conclusions:

• Deep learning models, particularly VGG16 and MobileNetV2, show significant potential for automated LSD detection in cattle.
• High accuracy, sensitivity, and specificity are vital for the clinical utility of AI diagnostic tools.
• AI-driven systems can enhance early detection and control of Lumpy Skin Disease, benefiting animal health and the agricultural sector.