Related Experiment Video
Updated: Jun 6, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
2.6K
Research on Behavior Recognition and Online Monitoring System for Liaoning Cashmere Goats Based on Deep Learning
Geng Chen1, Zhiyu Yuan1, Xinhui Luo1
1Animal Husbandry and Veterinary Research Institute, Jilin Academy of Agricultural Sciences, Shengtai Street, Changchun 130033, China.
Animals : an Open Access Journal From MDPI
|November 27, 2024
Summary
A new deep learning system accurately detects Liaoning Cashmere Goat behavior, improving health monitoring and welfare. This intelligent system enhances efficiency in intensive goat farming by automating behavior recognition.
Area of Science:
- Animal Science
- Computer Science
- Artificial Intelligence
Background:
- Liaoning Cashmere Goats are a vital national genetic resource in China, with intensive farming models developing.
- Efficient behavior recognition is crucial for improving health monitoring, welfare, and reducing labor costs in large-scale goat farming.
- Traditional behavior detection methods are inefficient and stressful for goats.
Purpose of the Study:
- To develop a deep learning-based system for online behavior recognition and detection in Liaoning Cashmere Goats.
- To compare the performance of Faster R-CNN and YOLO algorithms for goat behavior recognition.
- To optimize a deep learning model for accurate and real-time detection.
Main Methods:
- Compared Faster R-CNN and YOLOv8n for behavior recognition, with YOLOv8n serving as the baseline.
- Improved YOLOv8n through dataset expansion, algorithm lightweighting, attention mechanisms, and loss function optimization.
- Developed the Liaoning Cashmere Goat Online Behavior Detection System using the improved model.
Main Results:
- YOLOv8n achieved 95.31% average accuracy, converging within 50 epochs.
- The improved YOLOv8n model reached a detection accuracy of 98.11%, outperforming other state-of-the-art algorithms.
- The developed system demonstrated real-time detection with a low error rate, effectively replacing manual labor.
Conclusions:
- The deep learning-based system provides an effective solution for intelligent management of Liaoning Cashmere Goats.
- The optimized YOLOv8n model and the developed system significantly enhance the efficiency and accuracy of online behavior detection.
- This technology offers a promising approach to improving goat welfare and productivity in intensive farming settings.
