Bovine Tuberculosis Detection in Meat Products Based on Deep Learning Models
Özdemir, Mustafa Burkay, Ekin Ekıncı, Zeynep B. Garip, Furkan Göz · DergiPark (Istanbul University) · 2025
Bovine tuberculosis (bTB) is an infectious disease that threatens human health and possibly leads to death. It is known that tuberculosis can be transmitted to a healthy person through the air. However, it can also be transmitted through the consumption of animal products, such as meat and milk, which contain the bacteria that cause tuberculosis. Detecting tuberculosis in animal products is challenging and requires specialized expertise. There are many studies in which the detection of tuberculosis is performed using Deep Learning (DL) approaches. However, there is a lack of studies on the detection of bTB in animal products. In this study, we address this problem and propose a Convolutional Neural Network (CNN) DL approach for detecting bTB in meat products. We conducted experiments on a new dataset containing images of bTB infected and healthy meats. We evaluate the performance of different CNN architectures such as ResNet-50, DarkNet-53, MobileNet-v2, GoogleNet, and EfficientNet-b0 in detecting tuberculosis in meat products with respect to several metrics. We have been able to achieve validation accuracies of 100%, 99.67%,100%, 98.04%, and 100%, for ResNet-50, DarkNet-53, MobileNet-v2, GoogleNet and EfficientNet-b0, respectively.