Design and implementation of a vehicle and pedestrian detection system based on deep learning

Abstract

With the rapid development of deep learning, using deep learning to achieve efficient and accurate vehicle and pedestrian detection has become a research hotspot in the intelligent transportation field. To improve the application performance of the detection system in practical scenarios, this thesis aims to design and implement a deep learning-based vehicle and pedestrian detection system. The system architecture consists of two main components that are a lightweight detection model optimized by YOLOv8n and a graphical user interface (GUI) for real-time interaction. The system optimizes YOLOv8n model by adding the BiFormer attention mechanism and a 160 × 160 small-scale object detection head to recognize small objects in complex environments. Through comparative experiments with classic object detection models such as SSD and Faster R-CNN, the optimized YOLOv8 model achieves superior detection performance in vehicle and pedestrian detection. In addition, the application layer is developed using Python programming language and PyQt5 framework, providing a user-friendly interface that supports object detection of images, videos, and real-time camera inputs. It also includes real-time statistical analysis, detection confidence display, and category distribution visualization functions. This system provides a complete solution from detection model optimization to interface development. It has good application potential in scenarios such as intelligent transportation, autonomous driving, and safety monitoring.