The invention relates to a steel rail surface flaw detection system and method based on machine vision and pulse eddy currents. The system comprises a vision flaw detection subsystem, a pulse eddy current flaw detection subsystem, a data transmission subsystem and a data processing subsystem. The system and method are adopted, the detection system based on the machine vision and the pulse eddy currents is constructed through an embedded technology, by scanning and detecting a steel rail, the detection results of the machine vision and the pulse eddy currents are integrated, the size of a flawis judged in combination with characteristics of the flaw area, the length-width ratio and the like, and whether a defect exists or not on the surface of the steel rail is judged; an internet-of-things technology is introduced for wireless signal transmission, the functions of real-time detection, on-line locating, remote control and the like can be achieved, a 4G-LTE transmission function is added, Mobile, Unicom and Telecom 4G high-speed access is achieved, multi-system communication access is supported at the same time, the system and method have high stability, and flaw detection is conducted on the steel rail in real time at a super long distance; a pulse eddy current technology, a machine vision technology and the internet-of-things technology are utilized by the system and method, and the system and method have high application value.