Multiple data sources and domain generalization learning method for road surface defect classification

TL;DR

The paper tackles the problem of generalizing road surface defect classification across multiple data sources and unseen domains using camera images. It presents a two-part approach: contrastive learning to learn invariant embeddings across data sources and a domain-generalization algorithm that aligns unseen test feature distributions to the training distribution via Mahalanobis-distance-based adjustments. The method is evaluated on the six-country RDD2022 dataset with four defect classes (D00, D10, D20, D40) and multiple backbones, showing superior generalization to unseen data compared with transfer-learning baselines. This work advances practical road maintenance AI by enabling robust cross-domain defect classification without retraining on new data sources.

Abstract

Roads are an essential mode of transportation, and maintaining them is critical to economic growth and citizen well-being. With the continued advancement of AI, road surface inspection based on camera images has recently been extensively researched and can be performed automatically. However, because almost all of the deep learning methods for detecting road surface defects were optimized for a specific dataset, they are difficult to apply to a new, previously unseen dataset. Furthermore, there is a lack of research on training an efficient model using multiple data sources. In this paper, we propose a method for classifying road surface defects using camera images. In our method, we propose a scheme for dealing with the invariance of multiple data sources while training a model on multiple data sources. Furthermore, we present a domain generalization training algorithm for developing a generalized model that can work with new, completely unseen data sources without requiring model updates. We validate our method using an experiment with six data sources corresponding to six countries from the RDD2022 dataset. The results show that our method can efficiently classify road surface defects on previously unseen data.

Figures (1)

• Figure 1: The diagram illustrates our method for training domain generalization and contrastive learning on multiple data sources.