Abstract: 5T railway vehicle operation safety detection systems are crucial technical facilities in Chinese railway for ensuring the safe operation of rolling stock. The trackside equipment at the 5T detection points(abbreviated as 5T trackside equipment) directly conducts real-time detection on the safety status of moving rolling stock. Installed alongside the rails, these devices face data distortion and functional failure caused by safety risks resulting from natural environment, human and external interference, and equipment operation and maintenance, which affects the accuracy and reliability of the 5T system in detecting rolling stock faults. Currently, 5T trackside equipment mainly relies on regular manual inspection. With the increase in railway speed and load, problems of manual inspection in terms of operation efficiency and equipment operation safety have become increasingly prominent. This paper designs and implements a 5T trackside equipment video monitoring system based on AI algorithms. Combining edge computing and AI technology, the system collects video images of 5T trackside equipment through high-definition trackside cameras, and uses AI algorithms to realize intelligent detection and automatic alarm for three types of abnormalities: personnel intrusion, tools left-behind, and equipment loosening. Field experiment results show that the system achieves an early warning accuracy rate of 93.3%, a false alarm rate of less than 2.2%, and an average response time of no more than 30 seconds. With good scalability and flexibility, the system can provide effective technical support for the safe operation and maintenance of 5T systems.