Vison based metric for quality control by comparing built reality to BIM

This paper presents a method that exploits the opportunities site imagery offers in construction monitoring practices. While their power and that of derived point cloud data is acknowledged for progress monitoring purposes, (raw) imagery is often overlooked when assessing locational construction quality. This paper's framework addresses this shortcoming and harnesses the images' significant potential towards metric quality control of freshly built structural elements. Recorded images are assessed against virtually created BIM coun-terpart pictures. The discrepancies between both are exploited to determine construction element deviations. The proposed framework is able to output qualitative results with average differences between actual and determined element deviations ranging to approximately 5 mm with a standard deviation of 5 mm. The results unveil the method's large potential to aid in augmenting construction quality and controlling failure costs, while simultaneously allowing regular updates and conversions of the as-design BIM model into an as-built one.