MSCNet: A Framework With a Texture Enhancement Mechanism and Feature Aggregation for Crack Detection

Guanlin Lu,Xiaokang Zhao,Xiaohui He,Jinkang Wang,Faming Shao,Qiang Wang

doi:10.1109/access.2022.3156606

Guanlin Lu, Xiaokang Zhao + Show 4 more

Open Access

https://doi.org/10.1109/access.2022.3156606

Copy DOI

Abstract

Bridge crack is one of the critical optical and visual information to judge the health state of bridges. The bridge crack detection methods based on artificial intelligence are essential in this field, but the current approaches are not satisfactory in terms of speed and accuracy. This study proposes a novel multi-scale crack detection network, called MSCNet, comprising a texture enhancement mechanism and feature aggregation to enhance the visual saliency of the objects in the background for bridge crack detection. We use Res2Net as the backbone network to improve the depth information expression ability of the cracks itself. Because the edge property of bridge cracks is prominent, to make full use of this visual feature, we use a texture enhancement module based on group attention to capturing the detailed information of cracks in low-level features. To further mine the depth information of the network, we use a cascade fusion module to capture crack location information in high-level features. Finally, to fully utilize the characteristic information of the deep network, we fuse the low- and high-level features to obtain the final crack prediction. We evaluate the proposed method compared with other state-of-the-art methods on a large-scale crack dataset. The experimental results demonstrate the effectiveness and superiority of the proposed method, which achieves a precision of 93.5%, recall of 94.2%, and inference speed of over 63 FPS.

Highlights

Research has shown that the surface crack is the most obvious index of possible deterioration or damage of structures; the detection of surface cracks is essential for timely evaluation of the health status of bridges [4]
We present a simple but effective crack texture enhancement module (TEM) to capture the details of cracks in low-level features, which will increase the anti-noise capability of the network
To balance accuracy and consumption of computation resources, we introduce a cascaded fusion module (CFM) in the framework to reduce the complexity of the deep aggregation network and collect location information of cracks from high-level features

Summary

INTRODUCTION

Bridges are an important part of traffic lines and mainly used for railways, highways, channels, pipelines, and people to cross rivers, valleys, or other obstacles. Because the visual detection technology alone is not sufficient to evaluate the internal condition of bridge structural members, other in-depth methods should be introduced for a more comprehensive inspection. Research has shown that the surface crack is the most obvious index of possible deterioration or damage of structures; the detection of surface cracks is essential for timely evaluation of the health status of bridges [4]. Prasanna et al [10] developed an automatic crack detection algorithm called STRUM (spatially tuned robust multifeatured) classifier to detect cracks on the bridge surface. These methods improve the detection accuracy and speed compared with traditional crack detection technologies, they cannot cover all the unforeseen circumstances in complex environments.

RELATED WORKS

FEATURE EXTRACTOR

CASCADED FUSION MODULE

FEATURE AGGREGATION MODULE

EXPERIMENTS AND RESULTS