Pedestrian Detection Based on Improved YOLOv8

To address the high false positive of pedestrian detection in complex environments, an improved model based on YOLOv8 is proposed. A baseline network is optimized in this model to improve pedestrian detection accuracy and robustness in complex scenarios. Our method utilizes the WiderPerson dataset, known for its diverse and challenging scenes, for pedestrian target detection. The results show significant improvements in detection performance, making our model highly suitable for applications in smart tourism.