Abstract:
Cow identification is a critical aspect of livestock management, contributing to efficient
record-keeping, insurance, and tracking systems. Traditional methods are labour-intensive
and prone to errors. In this thesis, we propose an automated system for cow identification
using nose print images. We evaluated several models, including Support Vector Machine
(SVM), K-Nearest Neighbor (KNN), VGG16, VGG19, and a custom Convolutional Neural
Network (CNN). Our custom CNN model demonstrated superior performance with an
accuracy of 99.31% and classification scores (precision, recall, and F1-score) all at 0.99.
This high level of accuracy indicates the potential of the proposed system to replace manual
identification methods, thereby enhancing the reliability and efficiency of livestock
management. Our findings suggest that the custom CNN model is particularly well-suited
for this application, offering a robust solution for cow identification that can be integrated
into existing livestock management systems in Bangladesh. This research highlights the
significance of leveraging advanced machine learning techniques for improving
agricultural practices and livestock management.
