There have been several recent image-based recognition competitions (such as the PASCAL VOC, ImageNet, and COCO challenges) based on natural objects and scenes. It is time to hold competitions in other areas to develop interdisciplinary interactions with computer vision and machine & deep learning technologies.
As a pioneer study, the Pacific Earthquake Engineering Research (PEER) Center is organizing the first image-based structural damage recognition competition, namely PEER Hub ImageNet (PHI) Challenge, to be held at the end of Summer 2018. In the PHI Challenge, PEER will provide a large image dataset which is relevant to the field of structural engineering, and will design several detection tasks, which will contribute to the establishment of automated vision-based structural health monitoring. The goal of the PHI challenge is to evaluate algorithms for structural image classification using a large-scale dataset based on service conditions and past reconnaissance efforts and laboratory experiments for conditions of extreme events. The state-of-the-art algorithms to be tested in the PHI challenge are expected to enhance the accuracy and the generalization of vision-based approaches. These approaches will aim towards the construction of a big structural image dataset to solve societal-scale problems of structural health monitoring and assessment of the built environment.