Improved YOLOv4 Approach: A Real Time Occluded Vehicle Detection

Abstract

A major challenge in computer vision is detecting and tracking vehicles in real-time. However, existing algorithms fail to detect vehicles at high speed and accuracy. Therefore, an algorithm that detects vehicles with higher accuracy is required for surveillance in traffic scenarios. This paper proposed an improved algorithm for vehicle detection based on YOLO (You Only Look Once) Version 4 through Convolution Neural Network (CNN) and Hard Negative Example Mining (HNEM) dataset in the training process to improve the accuracy of the vehicle detection. In the end, videos are used to detect vehicles using a deep learning technique called You Only Look Once (YOLO). The test results indicate good real-time performance and high detection accuracy of the proposed algorithm. Several parameters such as accuracy, precision, recognition recall, F1, and mAP have been used to measure the proposed algorithm’s performance. The experiments have proved that the proposed algorithm achieved satisfactory performance in real-time due to occlusion and change in viewpoint. Finally, our proposed algorithm achieves improved precision, recall and mAP compared to the existing algorithms for occluded vehicle detection.