Integration of YOLOv8 and ResNet-50 to Improve Road Damage Detection Performance

Rendi Andrea Pramana; M. Arief Soeleman

doi:10.15408/jti.v19i1.46941

Authors

Rendi Andrea Pramana Computer Science, Dian Nuswantoro University, Semarang, Central Java, Indonesia
M. Arief Soeleman Computer Science, Dian Nuswantoro University, Semarang, Central Java, Indonesia

DOI:

https://doi.org/10.15408/jti.v19i1.46941

Keywords:

Detection, Road Damage, Yolov8, Resnet50, Backbone

Abstract

Automatic road damage detection is an important solution for more effective and efficient transportation infrastructure maintenance. This study proposes the implementation of the You Only Look Once version 8 (YOLOv8) method with ResNet50 as a backbone to improve feature extraction capabilities in detecting various types of road damage. The model was trained using a road damage image dataset that has gone through preprocessing and data augmentation stages to enrich image variations. Test results show that the proposed model is able to achieve excellent performance, with an accuracy value of 95.2%, a precision of 0.979, a recall of 0.968, and an F1-score of 0.974. This achievement proves that the integration of YOLOv8 with ResNet50 as a backbone can improve the reliability of the road damage detection system compared to the original model. With this performance, this method has the potential to be applied in a real-time road monitoring system to support more optimal transportation infrastructure maintenance planning.

References

[1] M. Hussain, “Yolov5, yolov8 and yolov10: The go-to detectors for real-time vision,”arXiv Prepr. arXiv2407.02988, 2024.

[2] RS Wijaya, S. Santonius, A. Wibisana, ER Jamzuri, and MAB Nugroho, “Comparative Study of YOLOv5, YOLOv7and YOLOv8 for Robust Outdoor Detection," J. Appl. Electr. Eng., vol. 8, no. 1, pp. 37–43, 2024.

[3] MA Ghofur and AB Ulum, “Comparative Analysis of the Performance of ResNet50 and EfficientNet-B0 Models in Road Damage Classification,”J. Media Inform., vol. 6, no. 5, pp. 2504–2511, 2025.

[4] S. Sharma, S. Dhakal, and M. Bhavsar, “Transfer Learning for Wildlife Classification: Evaluating YOLOv8 against DenseNet, ResNet, and VGGNet on a Custom Dataset,”arXiv Prepr. arXiv2408.00002, 2024.

[5] V. Alfiansyah, “Implementation of YOLOv8 for Detecting and Mapping Road Damage Points Using Google Street View (Case Study in Caturtunggal Village).” Gadjah Mada University, 2024.

[6] P. Singh, B. Likhitha, DG Reddy, D. Keerthana, BT Students, and AR Inspection, “Intelligent Road Monitoring : Advanced Damage And Pothole Detection With Yolov8,” no. 1, pp. 674–685, 2025.

[7] F. Wan, C. Sun, H. He, G. Lei, L. Xu, and T. Xiao, “YOLO-LRDD: a lightweight method for road damage detection based on improved YOLOv5s,”EURASIP J. Adv. Signal Process., vol. 2022, no. 1, 2022, doi: 10.1186/s13634-022-00931-x.

[8] ERMBA Sucipto, RRM Putri, and BD Setiawan, “Development of a Pothole Detection System on Roads Using the Yolo Algorithm Based on Esp32-Cam,”J. Development of Information Technology and Computer Science., vol. 9, no. 4, 2025.

[9] M. Surahmanto, S. Aras, M. Rifki Idhan Adhim, and P. Ussalama, “Pothole Detection Using the Yolov5 Algorithm,”J. Digit. Bus. Inf. Technol., vol. 1, no. 1, pp. 1–8, 2024, doi: 10.23971/jobit.v1i1.198.

[10] LG Denaro and R. Lim, “Analysis of Asphalt Road Damage Detection using Deep Learning to Support Cost and Time Efficiency in Continuous Monitoring,”J. Dimens. Ins. Prof., vol. 3, no. 1, pp. 16–25, 2025, doi: 10.9744/jdip.3.1.16-25.

[11] U. Satchithanantham, “Advanced pothole detection using neural network model-VGG16,”Int. J. Commun. Inf. Technol., vol. 5, no. 2, pp. 11–16, 2024, doi: 10.33545/2707661x.2024.v5.i2a.86.

[12] J. Zeng and H. Zhong, “YOLOv8-PD: an improved road damage detection algorithm based on YOLOv8n model,”Sci. Rep., vol. 14, no. 1, pp. 1–14, 2024, doi: 10.1038/s41598-024-62933-z.

[13] J. Terven, D.-M. Córdova-Esparza, and J.-A. Romero-González, “A comprehensive review of yolo architectures in computer vision: From yolov1 to yolov8 and yolo-nas,”Mach. Learn. Knowl. Extr., vol. 5, no. 4, pp. 1680–1716, 2023.

[14] Y. Lin, T. Yu, and Z. Lin, “FTN-ResNet50: flexible transformer network model with ResNet50 for road crack detection,”Evol. Syst., vol. 16, March. 2025, doi: 10.1007/s12530-025-09667-z.

[15] DBD Hanggoro, “Comparative Analysis of Deep Learning Architectures for Computer Vision Applications: A Literature Review Study,”J. Comput. Technol. Inf. Sis. Inf., vol. 4, no. 2, pp. 1001–1008, 2025.

[16] A. Setiyadi, E. Utami, and D. Ariatmanto, “Analysis of the capability of the YOLOv8 algorithm in human object detection using the architectural modification method,”J-SAKTI (Journal of Computer Science and Information)., vol. 7, no. 2, pp. 891–901, 2023.

[17] MH Rais, A. Musnansyah, and H. Fakhrurroja, “Application of Yolo V8 Algorithm for Recognizing Motorcycle Riders Without Helmets in Traffic Violation Monitoring System,”eProceedings Eng., vol. 12, no. 1, 2025.

[18] RG Wijanarko, AI Pradana, and D. Hartanti, “Implementation of Drone Detection Using YOLO (You Only Look Once),”J. Faculty of Computer Science, vol. 14, no. 2, pp. 437–442, 2024.

[19] I. Andi, M. Muchtar, and JY Sari, "Mask Detection Using the YOLO (You Only Look Once) Method,"J. Media Inf. Technol., vol. 1, no. 1, pp. 1–12, 2024.

[20] ZS Hidayat, YA Wijaya, and R. Kurniawan, "Optimizing YOLOv8 for Autonomous Driving: Batch Size for Best Mean Average Precision (mAP),"J. Tech. Inform., vol. 5, no. 4, pp. 1147–1153, 2024.