A Real-Time Road Anomaly Detection via Masked Autoencoder-Enhanced Vision Transformers

Ahmad Khalid Hussain; Fati Oiza Ochepa

doi:10.55524/ijircst.2025.13.3.25

Abstract

This study presents a deep learning solution for detecting road anomalies via a hybrid architecture consisting of a Masked Autoencoder (MAE) and a Vision Transformer (ViT) model. It presented a framework for dual road classification, namely an intact road (good) and defected road (bad) where defected roads are characterized by anomalies such as potholes or cracks. The target road anomaly classification model was trained and tested using publicly available datasets of road condition images. The model demonstrated good feature extraction as well as good generalization with a training accuracy of 99.79% and a test accuracy of 98.29%. Furthermore, we integrated the road anomaly detection model into a web-application providing real-time road anomaly detection, exemplifying the possible benefits of applying computer vision and machine learning algorithms to improve road maintenance in Nigeria.

Keywords

Anomaly detection, Deep learning, Image processing, Masked autoencoders, Transport safety, Vision Transformers

References

T. Tahir, H. Hassam, J. Choi, and E.-S. Jung, "Lightweight deep learning model for road pothole detection," Applied Intelligence, vol. 56, pp. 1234–1245, 2022.
K. Muhammad, J. Ahmad, Z. Lv, P. Bellavista, P. Yang, and S. W. Baik, "Automated Road defect and anomaly detection for traffic safety," Sensors, vol. 23, no. 12, p. 5656, 2023. Available from: https://doi.org/10.3390/s23125656
M. M. Q. U. H. T. Anwar, S. M., "Deep learning for road anomaly detection: A survey," arXiv preprint, 2022.
A. Martinez-Ríos, M. R. Bustamante-Bello, and L. A. Arce-Saénz, "A review of road surface anomaly detection and classification systems based on vibration-based techniques," Applied Sciences, vol. 12, no. 19, p. 9413, 2022. Available from: https://doi.org/10.3390/app12199413
R. Bibi, Y. Saeed, A. Zeb, T. M. Ghazal, T. Rahman, R. A. Said, S. Abbas, M. Ahmad, and M. A. Khan, "Edge AI-Based Automated Detection and Classification of Road Anomalies in VANET Using Deep Learning," Computational Intelligence and Neuroscience, vol. 2021, Article ID 6262194, 2021. Available from: https://doi.org/10.1155/2021/6262194
R. Basher, A. R. Ayon, A. Gharamy, A. A. Zayed, and M. S. Y. Ibna Zaman, "Real-time pothole detection using deep learning models: Mobilenet, Inception-V3, and YOLOv5," B.Sc. thesis, Dept. of Computer Science and Engineering, BRAC University, Dhaka, Bangladesh, 2022. Available from: https://tinyurl.com/hutawa5r
Y.-M. Kim, Y.-G. Kim, S.-Y. Son, S.-Y. Lim, B.-Y. Choi, and D.-H. Choi, "Review of recent automated pothole-detection methods," Applied Sciences, vol. 12, no. 11, p. 5320, 2022. Available from: https://doi.org/10.3390/app12115320
R. Ben-Shoushan and A. Brook, “Fused Thermal and RGB Imagery for Robust Detection and Classification of Dynamic Objects in Mixed Datasets via Pre-Trained High-Level CNN,” Remote Sensing, vol. 15, no. 3, p. 723, 2023. Available from: https://doi.org/10.3390/rs15030723
Ramesh, D. Nikam, V. N. Balachandran, L. Guo, R. Wang, L. Hu, G. Comert, and Y. Jia, “Cloud-Based Collaborative Road-Damage Monitoring with Deep Learning and Smartphones,” Sustainability, vol. 14, no. 14, p. 8682, 2022. Available from: https://doi.org/10.3390/su14148682
M. U. Hassan, O.-M. H. Steinnes, E. G. Gustafsson, S. Løken, and I. A. Hameed, “Predictive Maintenance of Norwegian Road Network Using Deep Learning Models,” Sensors, vol. 23, no. 6, p. 2935, 2023. Available from: https://doi.org/10.3390/s23062935
Bu?ko, E. Lieskovská, K. Zábovská, and M. Zábovský, "Computer vision-based pothole detection under challenging conditions," Sensors, vol. 22, no. 22, p. 8878, 2022. Available from: https://doi.org/10.3390/s22228878
H. Salaudeen and E. Çelebi, "Pothole detection using image enhancement GAN and object detection network," Electronics, vol. 11, no. 12, p. 1882, 2022. Available from: https://doi.org/10.3390/electronics11121882
M. Jakubec, E. Lieskovská, B. Bu?ko, and K. Zábovská, “Comparison of CNN-Based Models for Pothole Detection in Real-World Adverse Conditions: Overview and Evaluation,” Applied Sciences, vol. 13, no. 9, p. 5810, 2023. Available from: https://doi.org/10.3390/app13095810
Ezeibe, C. Ilo, C. Oguonu, A. Ali, I. Abada, E. Ezeibe, C. Oguonu, F. Abada, E. Izueke, and H. Agbo, “The impact of traffic sign deficit on road traffic accidents in Nigeria,” International Journal of Injury Control and Safety Promotion, vol. 26, no. 1, pp. 1–9, 2018. Available from: https://doi.org/10.1080/17457300.2018.1456470
S. S. Potluru, R. Mohammad, and R. Mande, "Road anomaly detection utilizing Swin transformer and deep convolutional neural networks with yolov8," in Innovations in Computational Intelligence and Computer Vision, Springer, 2024, pp. 375–393. Available from: https://link.springer.com/chapter/10.1007/978-981-97-6992-6_28
K. He, X. Chen, S. Xie, Y. Li, P. Dollár, and R. Girshick, "Masked autoencoders are scalable vision learners," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. Available from: https://tinyurl.com/52bnm4z4
Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, ... & N. Houlsby, "An image is worth 16x16 words: Transformers for image recognition at scale," International Conference on Learning Representations (ICLR), 2021. Available from: https://arxiv.org/pdf/2010.11929/1000

Cites this article as

A. K. Hussain, F. O. Ochepa, "A Real-Time Road Anomaly Detection via Masked Autoencoder-Enhanced Vision Transformers", International Journal of Innovative Research in Computer Science and Technology (IJIRCST), vol. 13, no. 3, pp. 179-187, 2025. doi: https://doi.org/10.55524/ijircst.2025.13.3.25

Corresponding Author

Ahmad Khalid Hussain

Graduate Student, Computer Science, Federal University Lokoja, Lokoja, Nigeria

Download Full Paper

Download PDF

No. of Downloads: 75 | No. of Views: 1916

A Real-Time Road Anomaly Detection via Masked Autoencoder-Enhanced Vision Transformers

Citations

Total View 1916

Total Download 75