This paper studies the machine vision platform equipped with micropthon interpreter and related applications, studies the design of mobile platform for object recognition in complex environment; the experimental method uses Hough circle detection algorithm to realize the recognition of circle; and uses the quaternion detection algorithm to realize the recognition of rectangle. Finally, when the camera recognizes the rectangular object, it drives the car motor. At the same time, if the system recognizes a circular object, it will drive the car motor to produce the effect of backward obstacle avoidance. The experimental results show that The mobile detection platform based on openmv can get the color and shape information of the target object according to the camera, and control the system to search and track the object with specific shape or color, and avoid the nontarget with the characteristics of high efficiency, accuracy and real-time.