Impact Statement—Overlapping multi-class concrete structural defect classification is the crucial primary onset for structural health inspection/monitoring. Manual inspection of massive infrastructure faces immense challenges due to time-consumption, inaccessibility, weather condition and human error/bias. This paper presents a fully automatic stand-alone composite attention network to improve the multi-target multi-class and singleclass concrete structural defect classification performance. This end-to-end trainable architecture is an endeavour to analyse defect images acquired using unmanned aerial vehicles (UAVs) and it focuses on robust discriminative information selection to leverage high classification accuracy. Utilizing this architecture can ensure efficient classification of major types of concrete structural defects and hence helps to automate the structural health inspection/management/monitoring process.

Automation in structural health monitoring involves a critical step of automatic classification of concrete defect images/videos. Although interdisciplinary research community in AI has responded with some progress, immense challenges are still involved because of the predominant overlapping nature of the defect classes, exacerbated by the large variations in their visual appearance. However, current methodologies mostly consider single-class nonoverlapping defects and emphasize equally over the entire image plane; thus unable to focus on specific defect regions for robust feature selection. Thereby, the classification performance gets degraded and the methodologies became less suitable for real-world scenarios. In this work, we propose a novel stand-alone composite attention network that automatically exerts higher emphasis on the defective regions and less emphasis on the healthy regions to recognize multitarget multiclass and single-class concrete structural defects. This architectu...