Smart Vision System for Sheep Face Detection using Deep Neural Networks

Abstract

This thesis investigates the application of deep learning techniques to sheep face detection, focusing on two prominent object detection models: SSD MobileNet 320x320 and CenterNet Hourglass 512x512. Our study aims to enhance livestock management by providing an efficient method for identifying individual sheep within a flock. By leverag ing advanced object detection algorithms, we demonstrate significant improvements in the accuracy and efficiency of sheep identification. The results indicate that while CenterNet Hourglass 512x512 achieves higher accuracy, SSD MobileNet 320x320 offers faster pro cessing times. This research contributes to the field of agricultural AI, offering practical solutions for enhanced livestock monitoring and management.