EfficientDet Model for Accurate Detection and Classification of Road Damages in Indonesia

Journal: Journal of Machine Learning and Deep Learning (JMLDL), Volume:2, Issue:1, Pages: 11-17 Download pdf

Authors: Bernard Hugo, Alfi Yusrotis Zakiyyah, Kelvin Asclepius Minor

Date: 05-2025

Abstract: This electronic Road infrastructure is an important aspect that must be maintained to ensure public road safety. Traditional road damage detection methods are labor-intensive, costly, and inefficient, highlighting the need for an automated solution. By utilizing the EfficientDet model, we tried to assess the model’s performance in detecting and classifying diverse types of road damage in Indonesia. After that, we leverage a built-in data augmentation technique to improve the model’s succession. We achieved the best result with the validation F1 Score of 59.7%, bypassing the performance of the previous work. The moderate performance of the model is caused by the complexity in learning road damages features and challenges in generalizing on unseen data.

Keywords: EfficientDet, road damage, road damage detection and classification, road infrastructure, Indonesia.

References:

[1] Y. Jiang, "Road Damage Detection and Classification Using Deep Neural Networks," Discover Applied Sciences, vol. 6, no. 421, p. 1, 2024.

[2] E. Suhartono, M. A, R. BP, M. M. D. R. Kusumastuti, D. B. Setiawan and N. S. P, "Estimated Cost of Repairing Road Pavement Damage Assessment: A Case Study of Indonesia," IJASRE, vol. 5, no. 6, p. 1-2, 2019.

[3] H. Kusumah, M. R. Nurholik, C. P. Riani and I. R. N. Rahman, "Deep Learning for Pothole Detection on Indonesian Roadways," Journal Sensi, vol. 09, no. 02, pp. 175-176, 2023.

[4] G. Guo and Z. Zhang, "Road Damage Detection Algorithm for Improved YOLOv5," scientific reports, vol. 12, no. 15523, p. 1, 2022.

[5] D. Arya, H. Maeda, S. K. Ghosh, D. Toshniwal, A. Mraz, T. Kashiyama and Y. Sekimoto, "Deep Learning-based Road Damage Detection and Classification for Multiple Countries," Automation in Construction, vol. 132, p. 1, 2021.

[6] V. Pham, L. D. T. Ngoc and D.-L. Bui, "Optimizing YOLO Architectures for Optimal Road Damage Detection and Classification: A Comparative Study from YOLOv7 to YOLOv10," IEEE, p. 1, 2024.

[7] S. Naddaf-Sh, M.-M. Naddaf-Sh, A. R. Kashani and H. Zargarzadeh, "An Efficient and Scalable Deep Learning Approach for Road Damage Detection," IEEE, pp. 1-2, 5, 2020.

[8] Basily, A. (2020, August). Road Damage, Version 1. Retrieved January 11, 2025, from https://www.kaggle.com/datasets/alvarobasily/roaddamage/ data.

[9] M. Tan, R. Pang and Q. V. Le, "EfficientDet: Scalable and Efficient Object Detection," IEEE, pp. 1-8, 2020.

[10] M. Tan and Q. V. Le, "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks," Proceedings of the 36th International Conference on Machine Learning, pp. 6105-6114, 2019.

[11] JRA. Maintenance and repair guide book of the pavement 2013. Japan Road Association, 1st. edition, 2013.

[12] H. Rezatofighi, N. Tsoi, J. Gwak, A. Sadeghian, I. Reid and S. Savarese, "Generalized Intersection Over Union: A Metric and a Loss for Bounding Box Regression," IEEE, pp. 658-659, 2019.

[13] T.-Y. Lin and P. G. R. G. K. H. P. Doll´ar, "Focal Loss for Dense Object Detection," IEEE, pp. 2980-2982, 2017.

[14] G. P. Meyer, "An Alternative Probabilistic Interpretation of the Huber Loss," IEEE, pp. 5257-5259, 2021.

[15] M. u. I. Arif, M. Jameel and L. Schmidt-Thieme, "Directly Optimizing IoU for Bounding Box Localization," in Pattern Recognition, Auckland, Springer, 2019, p. 545.

[16] H. Luo, C. Li, M. Wu and L. Cai, "An Enhanced Lightweight Network for Road Damage Detection Based on Deep Learning," Electronics, pp. 1-20, 2023.

[17] V. Mandal, A. R. Mussah and Y. Adu-Gyamfi, "Deep Learning Frameworks for Pavement Distress Classification: A Comparative Analysis," IEEE, pp. 1-7, 2020.

[18] V. Pham, D. Nguyen and C. Donan, "Road Damages Detection and Classification with YOLOv7," IEEE, pp. 1-8, 2022.

[19] M. S. Arman, M. M. Hasan, F. Sadia, A. K. Shaki, K. Sarker and F. A. Himu, "Detection and Classification of Road Damage Using R-CNN and Faster R-CNN: A Deep Learning Approach," LNICST 325, pp. 730-741, 2020.

[20] R. Vishwakarma and R. Vennelakanti, "CNN Model & Tuning for Global Road Damage Detection," IEEE, pp. 1-7, 2021.

[21] V. Pham, C. Pham and T. Dang, "Road Damage Detection and Classification with Detectron2 and Faster R-CNN," IEEE, pp. 1-10, 2020.

[22] F. Kortmann, K. Talits, P. Fassmeyer, A. Warnecke, N. Meier, J. Heger, P. Drews and B. Funk, "Detecting Various Road Damage Types in Global Countries Utilizing Faster R-CNN," IEEE, pp. 1-9, 2020.

[23] 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, vol. 2022, no. 98, pp. 1-18, 2022.

[24] A. Anand, A. Shinde, H. Borgave, and S. S. Thigale, "Road Damage Detection using CNN in Machine Learning," International Journal of Engineering Research & Technology (IJERT), vol. 12, no. 04, pp. 505-509, 2023.

[25] M.-M. Naddaf-Sh, "roadDamageDetection2020," GitHub, 26 October 2020. [Online]. Available: https://github.com/mahdi65/roadDamageDetection2020/blob/main/README.md. [Accessed 11 January 2025].