Design and Implementation of Multi-Target Detection Based on YOLO and SORT Algorithms

Abstract

Multi-target detection is one of the current directions in the field of computer vision, with broad application scenarios in intelligent transportation, security surveillance, and other domains. Traditional detection systems employ classifiers to evaluate different slices of test images. Previous detection algorithms suffered from slow speeds and difficulties in optimization. Through several generations of development, the YOLO (You Only Look Once) algorithm, with its end-to-end detection architecture, has significantly improved processing speed, making it more suitable for real-time scenarios. This paper adopts the YOLO model for object detection and further incorporates the SORT (Simple Online and Realtime Tracking) algorithm for object tracking. Ultimately, the study implements a vehicle and pedestrian detection system, completing tasks from algorithm implementation to model training, and further constructs a user-friendly graphical interface using the QT framework.