Abstract:
With the aim of addressing the existing challenges related to vehicle target detection across various environments, this study introduces an enhanced YOLOv5-based algorit...Show MoreMetadata
Abstract:
With the aim of addressing the existing challenges related to vehicle target detection across various environments, this study introduces an enhanced YOLOv5-based algorithm for vehicle detection. By incorporating a small target detection layer, the algorithm effectively addresses the issues of missed detections and false positives in the experimental settings, thus enhancing the model’s robustness in diverse environments. In this research, BiFPN is employed as the feature fusion network instead of PANet, and an adaptive weighted fusion strategy along with jump connections is adopted to tackle the challenge of integrating multi-scale information. Regarding image detection accuracy, the experimental outcomes demonstrate that the proposed improved YOLOv5 algorithm exhibits promising performance on vehicle datasets, achieving a detection accuracy of 94%. This notable enhancement in performance compared to the original algorithm confirms the feasibility of real-time vehicle detection using the algorithm proposed in this study.
Published in: 2023 3rd International Conference on Frontiers of Electronics, Information and Computation Technologies (ICFEICT)
Date of Conference: 26-29 May 2023
Date Added to IEEE Xplore: 01 September 2023
ISBN Information: