Measurement of Pig Body Temperature Based on Ear Segmentation and Multifactor Infrared Temperature Compensation

Infrared thermography offers a potential solution for noncontact measurement of porcine body temperature, which is critical for disease diagnosis and health monitoring. However, existing methods face challenges in accurately capturing abnormal temperatures in specific body parts of sick pigs. They have difficulty distinguishing specific pig body parts from the surrounding background and lack multifactor error compensation, resulting in suboptimal accuracy. To address these issues, this article proposes the model for pig ear segmentation (S-EAR) and establishes a multifactor infrared temperature compensation model by considering the effects of multiple environmental and intrinsic factors on infrared temperature measurement to improve the accuracy of temperature measurement. By combining the pig ear pixel set with the temperature-compensated image, the proposed method successfully extracts the temperature of the pig ear. Finally, the pig’s body temperature value was obtained from the ear temperature to body temperature mapping model. The experimental results show that the method has a pig S-EAR accuracy of 0.921, an inference time of only 0.019 s per image, and an average absolute error of $0.12~^{\circ }$ C for temperature measurement, which is in line with the acceptable error range for disease diagnosis ( $\pm 0.30~^{\circ }$ C). This method provides an effective measurement protocol for monitoring pig body temperature.