Optimization Research of Lightweight YOLOv8 Model in Building Crack Detection

³ Guangxi Zhuang Autonomous Region Prefabricated Building Life Cycle Management and Virtual Simulation Engineering Research Center, Liuzhou 545026, Guangxi, China

⁴ School of Economics and Management, Liuzhou Institute of Technology, Liuzhou 545026, Guangxi, China

BDCI 2025 , 3(1), 29–34; https://doi.org/10.1008611/amcmr.v1i11.65

© 2025 by the Author(s). Licensee Whioce Publishing, USA. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International License ( https://creativecommons.org/licenses/by/4.0/ )

Abstract

In order to improve the detection accuracy and efficiency of building cracks based on the YOLOv8 model, this paper proposes an improved YOLOv8 model. The improved model incorporates the SlimNeck structure, the CSPELAN4 designed based on the GELAN architecture, and the InnerGIoU loss function respectively. Then, an experimental comparative study of this model in building crack detection is carried out. The experimental results show that the precision P increases by 2.1%, the recall rate R increases by 4.2%, mAP@0.5 increases by 2.3%, and mAP@0.5:0.95 increases by 6.0%. At the same time, Params and GFLOPs are reduced by 21.6% and 23.5%, respectively.

Keywords

Crack detection

YOLOv8

Lightweight model

SlimNeck

GELAN

InnerGIoU loss function

References

1. Xu YC, Li HK (2024) Research on Innovation and Application of Civil Engineering Structure Health Monitoring Technology. New Urban Construction Science and Technology, 33(10): 57–59.
2. Zeng W, Gu LS, Chen SJ, et al. (2023) Research on the Influence of Fibers on the Mechanical Properties and Microstructure of Ultra-High Performance Concrete. Journal of Guangxi University of Science and Technology, 34(01): 20–27.
3. Wang ZX (2024) Causes Control and Treatment Methods of Concrete Cracks in Construction Engineering. Engineering Research and Practice, 5(7): 181–183.
4. Li Y, Han Z, Xu H, et al. (2019) YOLOv3-Lite: A Lightweight Crack Detection Network for Aircraft Structure Based on Depthwise Separable Convolutions. Applied Sciences, 9(18): 3781.
5. He Z, Su C, Deng Y (2024) A Novel Hybrid Approach for Concrete Crack Segmentation Based on Deformable Oriented-YOLOv4 and Image Processing Techniques. Applied Sciences, 14(5): 1892.
6. Huang KY, Zhao Y, Hu N, et al. (2024) Research on the Identification of Concrete Bridge Cracks by the YOLOv5 Network Based on the Attention Mechanism. Bulletin of Science and Technology, 40(09): 71–76.
7. Wang D, Zhang Y, Zhang R, et al. (2024) Detection and Assessment of Post-Earthquake Functional Building Ceiling Damage Based on Improved YOLOv8. Journal of Building Engineering, 98: 111315.
8. Dong X, Liu Y, Dai J (2024) Concrete Surface Crack Detection Algorithm Based on Improved YOLOv8. Sensors, 24(16): 5252.
9. Li QF, Zhang YH, Xie PP, et al. (2025) Rainy-Day Pedestrian Detection Method Based on Improved YOLOv8. Journal of Guangxi University of Science and Technology, 2025: 1–11.
10. Zhang C, Chen X, Liu P, et al. (2024) Automated Detection and Segmentation of Tunnel Defects and Objects Using YOLOv8-CM. Tunnelling and Underground Space Technology, 150: 105857.
11. Xu W, Zhu D, Deng R, et al. (2024) Violence-YOLO: Enhanced GELAN Algorithm for Violence Detection. Applied Sciences, 14(15): 6712.
12. Wang CY, Yeh IH, Liao HYM (2024) Yolov9: Learning What You Want to Learn Using Programmable Gradient Information. arXiv preprint, 2402.13616.

Conflict of interest

The authors declare no conflict of interest.

Previous article in this issue

Next article in this issue