Attention‐guided multiscale neural network for defect detection in sewer pipelines

Abstract

Sanitary sewer systems are major infrastructures in every modern city, which are essential in protecting water pollution and preventing urban waterlogging. Since the conditions of sewer systems continuously deteriorate over time due to various defects and extrinsic factors, early intervention in the defects is necessary to prolong the service life of the pipelines. However, prior works for defect inspection are limited by accuracy, efficiency, and economic cost. In addition, the current loss functions in object detection approaches are unable to handle the imbalanced data well. To address the above drawbacks, this paper proposes an automatic defect detection framework that accurately identifies and localizes eight types of defects in closed‐circuit television videos based on a deep neural network. First, an effective attention module is introduced and used in the backbone of the detector for better feature extraction. Then, a novel feature fusion mechanism is presented in the neck to alleviate the problem of feature dilution. After that, an efficient loss function that can reasonably adjust the weight of training samples is proposed to tackle the imbalanced data problem. Also, a publicly available dataset is provided for defect detection tasks. The proposed detection framework is robust against the imbalanced data and achieves a state‐of‐the‐art mean average precision of 73.4%, which is potentially applied in realistic sewer defect inspections.

Urban wind field prediction based on sparse sensors and physics‐informed graph‐assisted auto‐encoder

“…As previously stated, the graph feature exclusively consists of values relating to the sensor in the center and its neighboring points, and the remaining points in $$128 \times 128$$ data matrix are nullified to 0. It can thus be deduced that the application of the graph feature can be interpreted as an attention mechanism (Li et al., 2023), selectively focusing exclusively on the sensor in the center and its neighboring points. The structure of the graph feature encoder can be explicitly visualized and comprehended through Figure 11.…”

An integration–competition network for bridge crack segmentation under complex scenes

“…It is vital to conduct periodical detections on the security and durability of bridges (Mirzaei & Adeli, 2019; Ni et al., 2018; Okazaki et al., 2023; Shim et al., 2023; Sirca & Adeli, 2018). The bridge cracks will cause the structure‐bearing capability to decrease (Q. Kong et al., 2023; Y. Li et al., 2023; Ye et al., 2023; Zhong Zhou et al., 2023). Thus, bridge crack detection has become one of the most fundamental research topics in maintaining bridges (Bang et al., 2021; J. Chen et al., 2023; S. Y. Kong et al., 2021; Yamane et al., 2023).…”

“…The traditional methods mentioned above can compensate for the shortcomings of traditional manual detection, which is highly subjective and labor‐intensive. However, it required manually designing features, and the generalization is unsatisfactory (García‐Aguilar et al., 2023; Hua et al., 2022; Y. Li et al., 2023; Lin et al., 2022; K. Wang et al., 2022).…”

Deep learning‐based automatic classification of three‐level surface information in bridge inspection

“…To address these challenges, artificial intelligence (AI) techniques, such as machine learning (ML) and computer vision (CV), have been introduced to enable autonomous SHM of various structures and infrastructure, including bridges (Mirzazade et al., 2021; Zoubir et al., 2022), pipelines (Y. Li et al., 2023; Ma et al., 2022), tunnel structures (Marasco et al., 2022; Rosso et al., 2023), and pavements (Rodriguez‐Lozano et al., 2023; L. Zhao et al., 2022), among others (Ye et al., 2023). These AI methods provide promising solutions by leveraging their capabilities in data analysis, pattern recognition, and automated decision‐making.…”