Vehicle detection from road image sequences for intelligent traffic scheduling. (October 2021)

• Record Type: Journal Article
• Authors: Li, Yaochen; Chen, Yuting; Yuan, Sheng; Liu, Jingle; Zhao, Xi; Yang, Yang; Liu, Yuehu
• Is Part Of: Computers & electrical engineering. Volume 95(2021)
• Publication Date: 2021-10
• Subjects: Intelligent transportation -- Asymmetric convolution -- Global context attention -- Small object detection -- Deep reinforcement learning; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2021.107406
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: With the rapid development of unmanned aerial vehicle (UAV) technology, the UAV surveillance system has attracted extensive attention in the intelligent transportation community. In this paper, an object detection model with global context cross (YOLO-GCC) is proposed for identifying small sized traffic elements in UAV image sequences. The concept of the asymmetric convolution is introduced to increase the robustness of the object detection model. Moreover, a global context attention module is added to extract more efficient features to ensure the real-time performance while improving the detection accuracy of small objects. The evaluation and comparison results on multiple UAV datasets demonstrate the effectiveness of the proposed model. Furthermore, an intelligent traffic signal scheduling algorithm named Traffic Deep Q-Network(Traffic-DQN) using deep reinforcement learning is introduced, which utilizes the traffic flow data obtained from YOLO-GCC as the benchmark for traffic scheduling. The experimental results demonstrate that the proposed algorithm can effectively alleviate traffic congestion compared with other methods. Graphical abstract: Highlights: A feature extraction backbone network based on DarkNet-53 optimization is proposed, which can obtain more efficient features by integrating global context-based attention modules at the right location. The concept of asymmetric cross convolution is introduced to improve the small object identification accuracy under the premise of ensuring real-time performance. An intelligent traffic signal scheduling algorithm based on deep reinforcement learning is proposed, which can alleviate traffic congestions in complex roads through deep reinforcement learning.