Research on Multi-target Recognition and Classification Strategy Based on Yolo v5 Framework

Abstract

AbstractWith the ever-increasing social labor cost and automation level, the research on the recognition and automatic classification and sorting of multi-target objects has important practical significance. To this end, a set of multi-target object recognition and sorting system based on Yolo v5 is designed, which is composed of three parts: image acquisition part, image target object recognition and positioning part, and sorting execution of different types of targets. First, the sorting system collects the original image of the target object by controlling the movement of the camera. Next, the Yolo v5 target detection algorithm is used to realize the identification of the target object, and at the same time return the position information of the target object to the control host. Finally, the upper host computer controls the movement of the mechanical claws to realize sorting by controlling the decision-making algorithm. In order to improve the recognition accuracy in the target detection process and compress the model size to adapt to the Jetson Nano embedded system, an improved Yolo v5 target detection algorithm is proposed. The experimental results show that the designed multi-target detection system can accurately identify, locate and classify target objects; at the same time, the improved Yolo v5 target detection algorithm proposed has a smaller model and faster recognition speed, which can be realized in Rapid deployment of embedded systems.KeywordYolo frameworkTarget recognition and positioningDeep learningMachine visionControl strategy