Structural Image Classification

As mentioned in Chap.  2 , structural image classification is one of the most fundamental tasks in vision-based SHMStructural health monitoring. However, the introduction of AIArtificial intelligence technologies into the field of SHMStructural health monitoring is not straightforward. Therefore, in this chapter, the feasibility of applying AIArtificial intelligence methods in vision-based SHMStructural health monitoring is explored. This is mainly evaluated by the accuracy and efficiency of the trained AIArtificial intelligence models. MLMachine learning and DLDeep learning can achieve very accurate or promising results through training on a big dataset, but the performance may degrade on a smaller-scale dataset. However, the terms “big” and “small” of the scale of the dataset are ambiguous and subjective, because there is no explicit definition or specific threshold value to clearly separate them. It is also inappropriate and sometimes impossible to infer the results simply by examining the scale of the collected dataset without validation experiments.