Cattle Farms Foot-And-Mouth Disease Detection Using Convolutional Neural Network

Abstract

Foot and Mouth Disease (FMD) is a highly contagious and economically significant livestock disease, particularly in countries like Ethiopia where agriculture and livestock farming play a crucial role in the economy. From 2007 to 2021, the predicted pooled prevalence of FMD in cattle in Ethiopia was 21.39% (16.53-26.56%). This suggests that FMD affects more than 1 in 5 cattle in Ethiopia. Ethiopian cattle farms face several challenges in detecting and controlling FMD outbreaks. Limited resources, including trained personnel and diagnostic facilities, make it difficult to identify and respond to outbreaks in a timely manner. This research aims to bridge this gap by developing a Convolutional Neural Network (CNN) system specifically tailored to the detection of FMD in cattle farms. The dataset used for training included images of both FMD infected and healthy cattle. CNN architecture models such as VGG16, Inception V3, and Densenet 201 were utilized, and their accuracies were compared. The research achieved impressive results, with Densenet 201 exhibiting the highest performance with training, validation and testing accuracy of 99.55%, 99.00% and 98.87% respectively. This study has confirmed that Densenet 201 has shown the most significant improvement over other models in terms of evaluation metrics and performance.