24 / 2025-03-30 15:09:54

An Intelligent Vision Framework for Respiration Rate Estimation in Group-Housed Pigs Infected with Mycoplasma hyopneumoniae

Computer vision,Respiratory rate,Group-housed Pigs,Mycoplasma hyopneumoniae

3. 智慧牧场与动物福利 > （2）智慧养殖的大数据、机器学习和物联网应用

  Respiratory rate (RR) is a critical physiological indicator for assessing the welfare and health of pigs. While numerous studies have focused on monitoring the respiratory rate of pigs, most existing research has largely overlooked the issue of mutual occlusion among group-housed pigs. Additionally, studies on respiratory rates in pigs under diseased conditions remain relatively scarce. Therefore, this study concentrated on pigs infected with Mycoplasma hyopneumoniae (Mhp) and proposed a method for estimating the RR of group-housed pigs based on computer vision. An oriented object detection model was employed to identify individual pigs and automatically select regions of interest (ROI). RR was estimated by analyzing periodic fluctuations in the R, G, and B channel intensities within each ROI. To mitigate interference from neighboring pigs, an occlusion-handling mechanism was integrated into the workflow. The results demonstrated that this method effectively enhanced RR estimation in pigs under occlusion conditions, with a mean absolute error (MAE) of 1.38 breaths per minute (bpm), root mean square error (RMSE) of 2.16 bpm, mean absolute percentage error (MAPE) of 5.92%, and a correlation coefficient (R) of 0.99. The study further explored the potential of the model for long-term monitoring of respiratory rate in pigs infected with Mhp, revealing the correlation between RR variations and the progression of Mhp infection. This approach presents a novel method for the early detection and warning of Mhp infection.