Accurately detecting pavement cracks is essential to apply preventive and effective pavement treatments in a timely manner. In this paper, we proposed the Dense Attention U-Net (DAU-Net) to achieve pixel-wise segmentation of cracks on 3D pavement images. The encoder of the DAU-Net consists of multi-stage dense blocks to improve its capability of extracting informative contextual features. To achieve precise localization of cracks in the decoder, a novel channel attention block (CAB) is proposed, which reduces noisy responses and highlight salient encoder features using the channel attention mechanism. The DAU-Net is evaluated on large-scale, real-world 3D asphalt pavement images. In the ablation study, the proposed CAB demonstrates its effectiveness with a large boost on crack segmentation precision. In the comparative study, the DAU-Net outperforms state-of-the-art semantic segmentation models from previous works. With both qualitative and quantitative evaluations, the effectiveness of the DAU-Net is verified.
DAU-Net: Dense Attention U-Net for Pavement Crack Segmentation
19.09.2021
599394 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
CrackFormer Network for Pavement Crack Segmentation
IEEE | 2023
|Deep Learning-Based Real-Time Crack Segmentation for Pavement Images
Springer Verlag | 2021
|