Welcome to Francis Academic Press

Academic Journal of Computing & Information Science, 2024, 7(1); doi: 10.25236/AJCIS.2024.070102.

A Method of Lightweight Pedestrian Detection in Rainy and Snowy Weather Based on Improved YOLOv5


Liuyang Gao, Yun Yang

Corresponding Author:
Liuyang Gao

School of Electronic Information and Artificial Intelligence, Shaanxi University of Science and Technology, Xi’an, China


With the rapid development of artificial intelligence technology, pedestrian detection tasks have been widely used in intelligent security and other fields. In rainy and snowy weather, pedestrian detection models may be severely affected, resulting in missed detections and false alarms. This article proposes a lightweight pedestrian detection method based on improved YOLOv5 for rainy and snowy weather conditions. Firstly, to address the issue of insufficient datasets, artificial rain and snow noise is added to the collected pedestrian data. This increases the diversity of the data and improves the robustness of the model in handling pedestrian detection tasks in rainy and snowy weather. Next, a lightweight pedestrian detection model called MAC_YOLOv5s is designed based on the YOLOv5 network model. The MAC_YOLOv5s model replaces the backbone network of the original network with the Mobilenetv3 network to achieve light weight. In addition, the CBAM attention mechanism is integrated to enhance the model's anti-interference ability. The progressive feature pyramid network AFPN is introduced to fuse feature information between adjacent layers, extracting more important features. Experimental results show that the improved MAC_YOLOv5s lightweight pedestrian detection network model reduces the model size to one-third of the original YOLOv5s model, with a file size of only 5.35MB. The mAP value reaches 0.683. These results indicate that the improved MAC_YOLOv5s model has fewer parameters and a smaller model size compared to the original model, making it suitable for pedestrian detection tasks in rainy and snowy weather conditions.


Pedestrian Detection, YOLOv5s network, Mobilenetv3 Network, Attention Mechanism

Cite This Paper

Liuyang Gao, Yun Yang. A Method of Lightweight Pedestrian Detection in Rainy and Snowy Weather Based on Improved YOLOv5. Academic Journal of Computing & Information Science (2024), Vol. 7, Issue 1: 9-17. https://doi.org/10.25236/AJCIS.2024.070102.


[1] Ragesh N K, Rajesh R. Pedestrian Detection in Automotive Safety: Understanding State-of-the-Art [J]. IEEE Access, 2019, 7:47864-47890.

[2] Bootorabi F, Haapasalo J, Smith E, et al. Carbonic anhydrase VII–a potential prognostic marker in gliomas [J]. Health, 2011, 3(01): 6.

[3] Gao Y, WU Y H. YOLOv5s-based Lightweight Pedestrian Detection Algorithm [J]. Modern Electronics Technology, 2023, 46(22): 151-158. 

[4] Chen Juan, Li Yan, Kan Xi, et al. Based on EfficientNet, a lightweight pedestrian detection algorithm [J]. Foreign Electronic Measurement Technology, 2023, 42(06): 1-9.

[5] Yi Z, Yu S Z, Xiong W L, et al. In-Field Tobacco Leaf Maturity Detection with an Enhanced MobileNetV1: Incorporating a Feature Pyramid Network and Attention Mechanism [J]. Sensors (Basel, Switzerland), 2023, 23(13).

[6] Liu Y X, Gao Chen, Wu Zhen, et al. Improved YOLOv3 Embedded Real-time Vehicle Detection Algorithm [J]. Single-Chip Microcomputer and Embedded System Application, 2023, 23(02): 33-37.

[7] Su T, Wang Y, Deng Q Y, et al. Improved foggy weather pedestrian and vehicle detection algorithm based on YOLOv5 [J]. Journal of System Simulation, 1-11 [2023-12-14]. 

[8] Yang H, Wang J. Efficient Detection of Forest Fire Smoke in UAV Aerial Imagery Based on an Improved Yolov5 Model and Transfer Learning[J]. Remote Sensing,2023,15(23).

[9] Wang M, Mei Q S, Song X Y, et al. A Machine Anomalous Sound Detection Method Using the lMS Spectrogram and ES-MobileNetV3 Network [J]. Applied Sciences, 2023, 13(23).

[10] Gui X Q, Liu S Q, Li L, et al. Based on the improved YOLOv8 algorithm, a pedestrian detection algorithm for scenic areas [J]. Computer Engineering: 1-11 [2023-12-14]. 

[11] Niu S Q, Ma R, Xun X L, et al. Research on corn seed variety recognition based on improved CBAM attention mechanism in MobileNetV2 [J]. Journal of Chinese Cereals and Oils: 1-12 [2023-12-14]. 

[12] Jia L, Wang Y, Zang Y, Li Q, Leng H, Xiao Z & Jiang L. (2022). MobileNetV3 with CBAM for bamboo stick counting. IEEE Access, 10, 53963-53971.

[13] Yang G, Lei J, Zhu Z, Cheng S, Feng Z, & Liang R. AFPN: Asymptotic Feature Pyramid Network for Object Detection. arXiv 2023. arXiv preprint arXiv:2306.15988.

[14] He J, Erfani S, Ma X, Bailey J, Chi Y, & Hua X S. (2021). α-IoU: A family of power intersection over union losses for bounding box regression. Advances in Neural Information Processing Systems, 34, 20230-20242.