Automated Lumpy Skin Diseases Detection Using Machine Learning

Lumpy Skin Diseases (LSD) has history of almost century now, the first recorded LSD outbreak was in Northern Rhodesia of Zambia (Africa) in 1929, later during 1989 in the Middle East. Recently South Asia comes under expansion LSD as of in 2019 and 2022 India faces large breakout of LSD. LSD is expanding from Africa to Middle East to South Asia and now can be seen across Eurasia. Algorithms were available to process images of lumpy portions of skin of livestock which results in prediction at early stage, it saves livestock life and financial losses for dairy and allied farmers. Precision of various algorithms has a limitation of valid datasets, image classification, and connected layers. To address this limitation, we developed a Random Forest to address large dataset. The model was trained on a dataset comprising 819 training and 205 validation images across eight classes, utilizing data augmentation techniques to enhance generalization. Through iterative optimization, including dropout regularization and increased model depth, we achieved more classification accuracy on the validation set. Our results indicate that deep learning, specifically Random Forest outperforms traditional SVM-based approaches in image classification tasks.