Intelligent safety control is a set of intelligent industrial safety and production control platform for manufacturing scenes. It integrates technologies such as computer vision, in-depth learning and industrial Internet of Things. It aims to solve the pain points such as lagging identification of potential safety hazards in workshops, high pressure of manual monitoring, untimely disposal of potential hazards, and non-closed loop. The system integrates multi-source video data, identifies the abnormal state of "people, objects and environment" in real time, and realizes second-level early warning and active intervention. In the core links of quality inspection, equipment maintenance and safety control, the platform has shown remarkable results: defect identification accuracy of 0.01mm, violation response of 200ms, and detection accuracy of more than 94%. The scheme supports privatization deployment and data closed-loop iteration, and combines the integrated delivery of software and hardware, which is widely used in chemical, power plant, electronic manufacturing and other industries. The project not only significantly reduces the accident rate and operating costs, but also injects core competitiveness into intelligent manufacturing and helps industrial upgrading.