In order to solve the problem of chip damage caused by improper use of vacuum nozzles in semiconductor chip packaging, a vacuum nozzles model identification system based on OpenCV and Qt platform for semiconductor chip packaging was developed by using the theory and method of image recognition and visual detection. The shape, outer contour size, inner contour size, color and other characteristic parameters of the vacuum nozzle were extracted by the system. For the shape of the vacuum nozzle, the shape recognition algorithm based on rectangle was used. For the color, a color recognition algorithm based on the average gray value in the region of interest was proposed. For the dimension measurement, the edge point detection algorithm based on gray jump of one-dimensional pixel sequence was used. Then the characteristic parameters were taken as search conditions and compared in the vacuum nozzle database to achieve the classification effect. The experimental results show that the recognition accuracy of the system is 9885%, and the recognition time of a vacuum nozzle is about 1 s, which meets the practical application requirements. It provides reference for the management and classification of vacuum nozzle.