Search

As science and technology advance, the need for novel optimization techniques has led to an increase. The recently proposed metaheuristic algorithm, Gradient-based optimizer (GBO), is rooted in the gradient-based Newton's method. GBO has a more concrete theoretical foundation. However, gradient search rule (GSR) and local escaping operator (LEO) operators in GBO still have some shortcomings. The insufficient updating method and the simple selection process limit the search performance of the algorithm. In this paper, an improved version is proposed to compensate for the above shortcomings, called RL-SDOGBO. First, during the GSR phase, the Spearman rank correlation coefficient is used to determine weak solutions on which to perform dynamic opposite learning. This operation assists the algorithm to escape from local optima and enhance exploration capability. Secondly, to optimize the exploitation capability, reinforcement learning is used to guide the selection of solution update modes in the LEO operator. RL-SDOGBO is tested on 12 classical benchmark functions and 12 CEC2022 benchmark functions with seven representative metaheuristics, respectively. The impact of the improvements, the scalability and running time of the algorithm, and the balance of exploration and exploitation are analyzed and discussed. Combining the experimental results and some statistical results, RL-SDOGBO exhibits excellent numerical optimization performance and provides high-quality solutions in most cases. In addition, RL-SDOGBO is also used to solve the anchor clustering problem for small target detection, making it a more potential and competitive option. [ABSTRACT FROM AUTHOR]

Upgrading ordinary streetlights to smart streetlights to help monitor traffic flow is a low-cost and pragmatic option for cities. Fine-grained classification of vehicles in the sight of smart streetlights is essential for intelligent transportation and smart cities. In order to improve the classification accuracy of distant cars, we propose a reformed YOLOv3 (You Only Look Once, version 3) algorithm to realize the detection of various types of automobiles, such as SUVs, sedans, taxis, commercial vehicles, small commercial vehicles, vans, buses, trucks and pickup trucks. Based on the dataset UA-DETRAC-LITE, manually labeled data is added to improve the data balance. First, data optimization for the vehicle target is performed to improve the generalization ability and position regression loss function of the model. The experimental results show that, within the range of 67 m, and through scale optimization (i.e., by introducing multi-scale training and anchor clustering), the classification accuracies of trucks and pickup trucks are raised by 26.98% and 16.54%, respectively, and the overall accuracy is increased by 8%. Secondly, label smoothing and mixup optimization is also performed to improve the generalization ability of the model. Compared with the original YOLO algorithm, the accuracy of the proposed algorithm is improved by 16.01%. By combining the optimization of the position regression loss function of GIOU (Generalized Intersection Over Union), the overall system accuracy can reach 92.7%, which improves the performance by 21.28% compared with the original YOLOv3 algorithm.

Upgrading ordinary streetlights to smart streetlights to help monitor traffic flow is a low-cost and pragmatic option for cities. Fine-grained classification of vehicles in the sight of smart streetlights is essential for intelligent transportation and smart cities. In order to improve the classification accuracy of distant cars, we propose a reformed YOLOv3 (You Only Look Once, version 3) algorithm to realize the detection of various types of automobiles, such as SUVs, sedans, taxis, commercial vehicles, small commercial vehicles, vans, buses, trucks and pickup trucks. Based on the dataset UA-DETRAC-LITE, manually labeled data is added to improve the data balance. First, data optimization for the vehicle target is performed to improve the generalization ability and position regression loss function of the model. The experimental results show that, within the range of 67 m, and through scale optimization (i.e., by introducing multi-scale training and anchor clustering), the classification accuracies of trucks and pickup trucks are raised by 26.98% and 16.54%, respectively, and the overall accuracy is increased by 8%. Secondly, label smoothing and mixup optimization is also performed to improve the generalization ability of the model. Compared with the original YOLO algorithm, the accuracy of the proposed algorithm is improved by 16.01%. By combining the optimization of the position regression loss function of GIOU (Generalized Intersection Over Union), the overall system accuracy can reach 92.7%, which improves the performance by 21.28% compared with the original YOLOv3 algorithm. [ABSTRACT FROM AUTHOR]