DETR-crowd is all you need

Journal Title: Modern Innovations, Systems and Technologies - Year 2023, Vol 3, Issue 2

Abstract

"Crowded pedestrian detection" is a hot topic in the field of pedestrian detection. To address the issue of missed targets and small pedestrians in crowded scenes, an improved DETR object detection algorithm called DETR-crowd is proposed. The attention model DETR is used as the baseline model to complete object detection in the absence of partial features in crowded pedestrian scenes. The deformable attention encoder is introduced to effectively utilize multi-scale feature maps containing a large amount of small target information to improve the detection accuracy of small pedestrians. To enhance the efficiency of important feature extraction and refinement, the improved EfficientNet backbone network fused with a channel spatial attention module is used for feature extraction. To address the issue of low training efficiency of models that use attention detection modules, Smooth-L1 and GIOU are combined as the loss function during training, allowing the model to converge to higher precision. Experimental results on the Wider-Person crowded pedestrian detection dataset show that the proposed algorithm leads YOLO-X by 0.039 in AP50 accuracy and YOLO-V5 by 0.015 in AP50 accuracy. The proposed algorithm can be effectively applied to crowded pedestrian detection tasks.

Authors and Affiliations

Liu Weijia , Zishen Zheng , Ke Fan , Kun He , Taiqiu Huang , Weijia Liu , Xianlun Ke , Yuming Xu

Keywords

Related Articles

Development of information technologies in the tourism sector

The development of foreign and Russian tourist information centers is characterized. The extensive expansion of the network of Russian centers, the diversity of their activities, the accumulation of a number of unresolve...

Determination of the characteristics of intercooling rocks over hydrocarbons with the use of frequency modulated signals

The article discusses the analysis of the impact of frequency-modulated signals on an anisotropic medium above hydrocarbon accumulations. A quasi-hydrodynamic approach and computer modeling were used to carry out the ana...

Structural and parametric synthesis of a document management system

The structures representing the document flow processes in an organization are considered by synthesizing them from the simplest structures. Presented in the form of graphs the processes of movement of documents become f...

Optimization of the method for determining the fatty acid composition of dairy products

Chromatographic analysis of fatty acid methyl esters is used to characterize the lipid fraction of food products and is an effective method for detecting adulteration and one of the most important applications in food an...

Verification of the mathematical model of the induction soldering technological process

The paper has devoted to the research of the construction and verification of the mathematical model of the process of heating the elements of the thin-walled aluminum waveguide path in the development of the induction s...

Download PDF file
  • EP ID EP716921
  • DOI 10.47813/2782-2818-2023-3-2-0213-0224
  • Views 47
  • Downloads 0

How To Cite

Liu Weijia, Zishen Zheng, Ke Fan, Kun He, Taiqiu Huang, Weijia Liu, Xianlun Ke, Yuming Xu (2023). DETR-crowd is all you need. Modern Innovations, Systems and Technologies, 3(2), -. https://europub.co.uk/articles/-A-716921