Overview of Lightweight Technology for Pedestrian Detection Based on YOLOv8

DOI: https://doi.org/10.62517/jbdc.202601206

Author(s)

Tan Li

Affiliation(s)

Electronic Information Engineering, Hubei University, Hubei, China

Abstract

Pedestrian detection, as one of the core tasks in current computer vision, is a foundational technology for achieving goals such as autonomous driving and intelligent surveillance. With the increase in edge computing and real-time requirements, various detection models must meet stringent demands for low latency and low power consumption while maintaining high accuracy. YOLOv8 has become the mainstream framework for object detection, especially pedestrian detection tasks, due to its excellent accuracy-speed balance and reasonable modular design. This paper aims to systematically review the progress of lightweight research on pedestrian detection based on YOLOv8. First, it outlines the architectural features of YOLOv8 and its advantages as a choice for lightweight models; second, it categorizes and summarizes representative work in recent years on algorithm improvements and model compression, comparing their performance through analysis; then, it delves into core lightweight techniques such as pruning, quantization, and knowledge distillation, and their applicability in pedestrian detection; finally, it discusses current challenges like handling small targets and occlusions, balancing accuracy and speed, and looks forward to future directions including multi-modal fusion, dynamic inference, and new versions of the YOLO series. This paper provides a reference for researchers to fully understand the landscape of this field and choose technical pathways.

Keywords

YOLOv8; Lightweight; Pedestrian Detection

References

[1] Ultralytics, "YOLOv8 - Supported tasks and modes," Ultralytics Docs, 2023. [Online]. Available: https://docs.ultralytics.com/zh/models/yolov8/#supported-tasks-and-modes. [Accessed: Jan. 29, 2026]. [2] J. Sang et al., "YOLOv8-GDI: A lightweight YOLOv8 for real-time pedestrian detection," in *2024 7th International Conference on Robotics, Control and Automation Engineering (RCAE)*, 2024, pp. 422-428. [3] Zhu G L, Yuan C X, Jiang F, et al. Lightweight YOLOv8 real-time object detection via progressive pruning and feature-aware knowledge distillation[C]//2025 IEEE 8th Advanced Information Technology and Electronic Information Engineering Conference. IEEE, 2025: 1-6. [4] Y. Gu, Y. Zheng, T. Ding, X. Song and X. Zhang, "Urban Road Autonomous Driving Vehicle Target Detection Algorithm Based on Improved YOLOv8," 2024 9th International Conference on Intelligent Computing and Signal Processing (ICSP), Xian, China, 2024, pp. 167-173, [5] G. Zhu, C. Yuan and F. Jiang, "Lightweight YOLOv8 Real-Time Object Detection via Progressive Pruning and Feature-Aware Knowledge Distillation," 2025 IEEE 8th Advanced Information Technology, Electronic and Automation Control Conference (IAEAC), Guiyang, China, 2025, pp. 925-933. [6] K. Li, S. Wang and Z. Zhang, "Research on Road Object Detection Method Based on Lightweight YOLOv8s," 2025 7th International Conference on Information Science, Electrical and Automation Engineering (ISEAE), Harbin, China, 2025, pp. 1205-1209. [7] C. Hu, Y. Wei and X. Tao, "Lightweight YOLOv8 Pedestrian Detection Model Based on FasterNet," 2024 4th International Symposium on Artificial Intelligence and Intelligent Manufacturing (AIIM), Chengdu, China, 2024, pp. 787-790. [8] J. Cui, H. Zheng, X. Huang and Y. Zhang, "A Lightweight Detection Algorithm YOLOv8-L Based on YOLOv8n," 2024 6th International Academic Exchange Conference on Science and Technology Innovation (IAECST), Guangzhou, China, 2024, pp. 853-857. [9] W. Zhao, X. Yang, X. Ma and Y. Wang, "A Lightweight Multi Object Detection Algorithm for Complex Road Scenes Based on CMF-YOLO," 2024 IEEE 22nd International Conference on Industrial Informatics (INDIN), Beijing, China, 2024, pp. 1-6.