Abstract: Under the background of China's dairy farming industry transitioning to large-scale and intensive production models, early screening and warning of mastitis still faced challenges such as long detection cycles and difficulties in timely intervention. To address these issues, a mastitis early-warning system for dairy cow was designed by integrating AIoT, infrared thermal imaging technology, and machine learning. Machine learning algorithms were adopted to conduct fusion analysis of multidimensional data, automatically identify early abnormal thermal distribution patterns and behavioral characteristics associated with mastitis, and realize risk grading and early warning for mastitis at the subclinical stage. The early warning system achieved closed-loop management spanning automated data collection, encrypted transmission, intelligent analysis, and decision feedback. It could help identify potentially affected cows, assist farms in timely intervention, and reduce farming-related economic losses. The system was expected to promote the transformation of dairy farming management toward digital, intelligent, and preventive models, and supported the high-quality development of the dairy industry.

Key words: artificial intelligence of things, machine learning, infrared thermal imaging technology, dairy cow mastitis