Related Experiment Video
Updated: Mar 15, 2026
06:45
Automated Joint Space Detection Improves Bone Segmentation Accuracy
Published on: November 28, 2025
253
Cattle lameness detection using depth image and deep learning
San Chain Tun1, Pyke Tin2, Masaru Aikawa3
1Interdisciplinary Graduate School of Agriculture and Engineering, University of Miyazaki, Miyazaki, 889-2192, Japan.
Scientific Reports
|March 14, 2026
Summary
This study introduces a deep learning framework for cattle lameness detection using depth images. The system achieves high accuracy in identifying and tracking lame animals, improving welfare monitoring.
Area of Science:
- Animal Science
- Computer Vision
- Machine Learning
Background:
- Lameness in cattle presents significant animal welfare and economic challenges.
- Current monitoring methods are often manual and subjective.
- Automated systems are needed for continuous and objective assessment.
Purpose of the Study:
- To develop and evaluate an end-to-end deep learning framework for 24/7 cattle lameness monitoring.
- To compare different instance segmentation models and tracking algorithms.
- To optimize a spatio-temporal model for accurate lameness classification.
Main Methods:
- The framework utilizes instance segmentation (YOLOv11m-seg), a custom tracking algorithm (PTAV3), and a spatio-temporal classification model (EfficientNet-B7 + LSTM).
- Top-down depth images of cattle were used for detection, tracking, and classification.
- Multiple model configurations and pre-processing techniques were evaluated.
Main Results:
- YOLOv11m-seg achieved high detection accuracy (Mask AP@50: 99.26%) at 75.49 FPS.
- The PTAV3 tracking algorithm reached 99.94% overall accuracy.
- The best classification model (EfficientNet-B7 + LSTM) achieved 95.95% accuracy and 96.06% F1-score.
Conclusions:
- The developed deep learning framework offers a robust, automated, and objective solution for cattle lameness scoring.
- This system demonstrates significant potential for real-time animal welfare monitoring in agricultural settings.
- The integrated approach enhances efficiency and accuracy in managing cattle health.