EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Road pedestrian detection and tracking algorithm based on improved YOLOv5s and DeepSORT

Guofeng Qin, Rongting Pan iD, Yi Deng iD, Peiwen Mi, Yongjian Zhu

PLOS One · 04.11.2025 · 20 (11) · e0334786

DOI: 10.1371/journal.pone.0334786 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 04.11.2025. Verfasst von Guofeng Qin und 4 weiteren Autor:innen. Beteiligte Einrichtung: Teachers College for Vocational and Technical Education, Guangxi Normal University, Guilin, China. Hauptthemen laut MEDLINE: Algorithms, Pedestrians. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Innovative Research Group Project of the National Natural Science Foundation of China, Project for Enhancing Young and Middle-aged Teacher's Research Basis Ability in Colleges of Guangxi.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

To address the challenges of low accuracy, high miss detection rate, and poor tracking stability in pedestrian detection and tracking under dense occlusion and small object scenarios on traffic roads, this paper proposes a pedestrian detection and tracking algorithm based on improved YOLOv5s and DeepSORT. For the improvements in the YOLOv5s detection network, first, the Focal-EIoU loss function is used to replace the CIoU loss function. Second, a 160 × 160-pixel Small Object (SO) detection layer is added to the Neck structure. Finally, the Multi-Head Self-Attention (MHSA) mechanism is introduced into the Backbone network to enhance the model’s detection performance. Regarding the improvements in the DeepSORT tracking framework, a lightweight ShuffleNetV2 network is integrated into the appearance feature extraction network, reducing the number of model parameters while maintaining accuracy. Experimental results show that the improved YOLOv5s achieves an mAP0.5 of 80.8% and an mAP0.5:0.95 of 49.7%, representing increases of 4.4% and 3.9%, respectively, compared to the original YOLOv5s. The enhanced YOLOv5s-DeepSORT achieves an MOTA of 50.7% and an MOTP of 77.3%, improving by 3.3% and 0.5%, respectively, over the original YOLOv5s-DeepSORT. Additionally, the number of identity switches (IDs) is reduced by 11.3%, and the model size is reduced to 20% of the original algorithm, enhancing its portability. The proposed method demonstrates strong robustness and can effectively track targets of different sizes.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

AlgorithmsPedestriansHumans

Autor:innen und Institutionen

  • Teachers College for Vocational and Technical Education, Guangxi Normal University, Guilin, China.
  • Guangxi Key Laboratory of Brain-inspired Computing and Intelligent Chips, School of Electronic and Information Engineering, Guangxi Normal University, Guilin, China.
  • School of Information Engineering, Guangxi Vocational College of Water Resources and Electric Power, Nanning, China.
  • Guilin Normal University, Guilin, China.
  • College of Engineering Physics, Shenzhen Technology University, Shenzhen, China.

ORCID-Profile: Rongting Pan, Yi Deng

Förderung

  • Innovative Research Group Project of the National Natural Science Foundation of China
  • Project for Enhancing Young and Middle-aged Teacher's Research Basis Ability in Colleges of Guangxi

Zitieren

APA

Qin, G., Pan, R., Deng, Y., Mi, P., & Zhu, Y. (2025). Road pedestrian detection and tracking algorithm based on improved YOLOv5s and DeepSORT. PLOS One, 20(11), e0334786. https://doi.org/10.1371/journal.pone.0334786

Vancouver

Qin G, Pan R, Deng Y, Mi P, Zhu Y. Road pedestrian detection and tracking algorithm based on improved YOLOv5s and DeepSORT. PLOS One. 2025;20(11):e0334786. doi:10.1371/journal.pone.0334786

BibTeX
@article{Qin2025,
  author = {Qin, Guofeng and Pan, Rongting and Deng, Yi and Mi, Peiwen and Zhu, Yongjian},
  title = {Road pedestrian detection and tracking algorithm based on improved YOLOv5s and DeepSORT},
  journal = {PLOS One},
  year = {2025},
  volume = {20},
  number = {11},
  pages = {e0334786},
  doi = {10.1371/journal.pone.0334786}
}
Kostenlos starten →