期刊
AUTOMATION IN CONSTRUCTION
卷 158, 期 -, 页码 -出版社
ELSEVIER
DOI: 10.1016/j.autcon.2023.105217
关键词
Pavement crack detection; Semantic segmentation; Convolutional neural network; Vision transformer; Deep learning
This study presents a dual-path network for pavement crack segmentation, combining Convolutional Neural Network (CNN) and transformer. A lightweight CNN encoder is used for local feature extraction, while a novel transformer encoder integrates high-low frequency attention mechanism and efficient feedforward network for global feature extraction. Additionally, a complementary fusion module is introduced to aggregate intermediate features extracted from both encoders. Evaluation on three datasets confirms the superior performance of the proposed network.
Cracks are one of the most common pavement surface diseases. Timely repair of these cracks is imperative to prevent a substantial reduction in the pavement's service life. However, the persistent challenges in crack segmentation arise from factors such as thin and shallow crack characteristics, a cluttered background, and foreground distractors. In response to these challenges, a dual-path network for pavement crack segmentation is introduced, leveraging a synergistic combination of Convolutional Neural Network (CNN) and transformer. First, the proposed approach involves a lightweight CNN encoder for local feature extraction and a novel transformer encoder integrating a fully convolutional high-low frequency attention (FCHiLo) mechanism and an efficient feedforward network for global feature extraction. Second, a complementary fusion module (CFM) is introduced to aggregate intermediate features extracted from both encoders. The multi-scale fusion outputs are systematically conveyed to the decoder, facilitating gradual image recovery and segmentation result acquisition. Evaluation on three publicly available datasets-DeepCrack, CrackForest, and CrackTree 260-affirms the superior performance of the proposed network compared to ten established models.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据