Abstract: Once the teeth of loading equipment in open-pit mines fall off or break,they are highly likely to enter subsequent crushing processes along with the material,potentially causing catastrophic damage to the crushing system and incurring immeasurable direct and indirect economic losses to the enterprise. Although a few literature reports and patents on online monitoring and alarming for missing teeth exist,the real-time performance of such systems remains inadequate for on-site requirements due to limitations of traditional machine learning algorithms and the computing power of mobile devices. As a result,few technologically mature products have actually been applied in engineering practice. To address this major safety issue that has long plagued open-pit mines worldwide,this study employs the multi-object detection YOLOv5 algorithm based on machine vision,combined with a hierarchical feature description strategy from low to high levels,to develop an integrated monitoring system that performs real-time online monitoring and intelligent alarming of the working status of dipper teeth,capable of issuing timely warnings against tooth loss accidents. The system has been put into practical application on various types of loading equipment in China,and its operational performance has verified its reliability,real-time capability,and accuracy. It can effectively reduce safety risks in mining production,prevent significant economic losses caused by missing teeth,and provide solid technical support for the construction and improvement of digital,intelligent,unmanned,and scientific management platforms in open-pit mines.