Application of infrared thermography and machine learning techniques in cattle health assessments: A review

Abstract

To maintain animal welfare and reduce the economic loss associated with pastures, it is essential to monitor the health status of livestock. Accurate health assessments are the heart of the future of high-quality and high-yielding livestock production. As a non-invasive detection method, infrared thermography (IRT) technology can be used to indicate changes in thermal biological characteristics in animal metabolism, and it is a useful tool to detect issues in animal health. In addition, the development of emerging technologies represented by machine learning (ML) technology provides an opportunity to achieve non-contact, high-precision, automated cattle health assessments. Therefore, in this review, we examine the more than 100 related research papers to summarise and analyse the application of IRT and ML technology in cattle health management. Firstly, IRT and ML technology are introduced, including the concepts of acquiring of thermal images and obtaining parameter information. Secondly, the development status and research progress of IRT technology in cattle health assessment is summarised, focusing not only on bovine disease detection, including mastitis, lameness, respiratory diseases, etc., but also involving the indicators for assessing the health status, including physiological characteristics, stress, temperament and the oestrus of cattle. Thirdly, we focus on the tasks and application potential of ML and the deep learning (DL) algorithms in thermal infrared imaging data analysis before finally discussing the challenges associated with IRT and ML technology in the field of cattle health assessments are and examining the relevant suggestions put forward in this paper.