Recognizing Cow Muzzle Patterns using the Convolution Neural Network (CNN) Algorithm

In today's digital era, any task or problem can be solved with minimal effort, especially in livestock identification such as cattle. Numerous systems and algorithms have been developed to recognize cattle, ranging from body shape, fur patterns, to specific parts of the cattle. This research aims to develop a cattle muzzle identification system using convolutional neural networks method with Alexnet architecture and to identify the factors that can decrease the accuracy of prediction results. The results of this research can help cattle farmers manage their livestock data more effectively, as traditional identification methods can cause discomfort and stress to the cattle. This research also serves as a reference for future researchers in developing cattle recognition research. Additionally, this research can be used to support insurance programs such as Cattle Farming Insurance (AUTS) to protect farmers from losses due to cattle theft and death. Cattle recognition through their muzzles using the CNN method can produce relatively high results. By slightly modifying the AlexNet architecture, this system can recognize cattle with an accuracy of 85%..