ANALYSIS OF INFLUENCE OF INSPECTION SKILLS ON DETECTION RESULT OF BRIDGE CRACK DETECTION UNDER IMAGE VISUAL INSPECTION ENVIRONMENT ~EXAMINATION USING CLUSTER ANALYSIS~

Abstract

The aging of many bridges constructed during the period of high economic growth is rapidly progressing. The Ministry of Land, Infrastructure, Transport and Tourism places importance on preventive maintenance of bridges and requires periodic inspections of all bridges. The proximity inspection, which is currently being performed, cannot be expected to be efficient due to budget and personnel issues, and various alternative systems have been studied. We propose an "image visual diagnosis system" that constructs an environment that is visually comparable to current inspections by using a camera, and that allows a person to diagnose cracks on images. In this study, we evaluated the effect of years of experience in bridge inspection on crack detection in visual inspection of images. In this study, the data on the crack detection results of the subjects were classified based on the years of experience and the time required for crack diagnosis from cluster analysis, and the effect of each cluster on the precision, recall, and F value was evaluated.