Pedestrian Occlusion Detection and Tracking Method Based on Improved PP-YOLOE+DeepSORT algorithm

This paper proposes a pedestrian target detection and tracking algorithm based on the improved PP-YOLOE and DeepSORT. It aims to address the problems of low accuracy and poor robustness of traditional target detection and tracking algorithms in scenes where human bodies occlude each other. In the detection part, the spatial attention mechanism is applied to the PP-YOLOE algorithm so that the model can focus on key regions in the image and extract more expressive features. In addition, this paper's network model is improved to reduce the computational complexity and speed up the detection speed. In the tracking part, we choose the DeepSORT algorithm to track the detected pedestrian features in real time. Finally, experimental results show that the model's design and improvement can accurately detect and track pedestrians in complex environments with occlusion and high target density.