Related Experiment Video
Updated: Jan 18, 2026
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
5.3K
PigStressNet: A Real-Time Lightweight Vision System for On-Farm Heat Stress Monitoring via Attention-Guided Feature
Shuai Cao1,2, Fang Li1,2, Xiaonan Luo1,2
1School of Computer Science and Information Security, Guilin University of Electronic Technology, Guilin 541004, China.
Sensors (Basel, Switzerland)
|September 13, 2025
Summary
This study introduces PigStressNet, a novel AI system for detecting pig heat stress using subtle cues like skin redness. It offers efficient, real-time monitoring for improved animal welfare and farm productivity.
Area of Science:
- Animal Science
- Computer Vision
- Machine Learning
Background:
- Heat stress negatively impacts pig welfare and farm productivity.
- Current detection methods struggle with subtle physiological cues and real-time efficiency in complex farm settings.
Purpose of the Study:
- To develop an accurate and efficient AI-based system, PigStressNet, for recognizing pig heat stress.
- To address limitations in detecting subtle heat stress indicators like skin erythema and altered posture.
Main Methods:
- Developed PigStressNet, a lightweight object detector incorporating a Normalization-based Attention Module (NAM) and Rectangular Self-Calibration Module (RCM).
- Utilized an MBConv-optimized detection head (MBHead) and MPDIoU loss function for computational efficiency and improved bounding box accuracy.
- Created a novel, fine-grained dataset with 710 images across 5 classes (standing, eating, sitting, lying, stress), fusing posture and physiological traits.
Main Results:
- PigStressNet achieved a state-of-the-art mean Average Precision (mAP) of 0.979 for heat stress detection.
- The system demonstrated superior efficiency with 15.9% lower computation (5.3 GFLOPs) and 11.7% fewer parameters than the YOLOv12-n baseline.
- Real-time inference capability was achieved on embedded devices.
Conclusions:
- PigStressNet provides an accurate, efficient, and real-time solution for detecting pig heat stress.
- The system's ability to identify subtle cues like skin erythema and posture changes enhances its practical application in intelligent livestock management.
- This technology offers a significant advancement in monitoring and mitigating the adverse effects of heat stress on pigs.
