Evaluation of the Crack Severity in Squared Timber Using CNN.

Natural woods for building materials, such as cedars used in wooden construction, have high moisture. It is essential to dry them to use as building materials. Currently, high-temperature drying is commonly employed. However, this process sometimes results in internal cracks that are not apparent on the surface. It takes considerable time and effort to measure the length and area of the cracks in the cross-section of square timber using calipers. A system to automatically estimate the crack severity is required to facilitate researchers to derive the conditions and methodology so that internal cracks do not occur while drying. Therefore, we tried to rank the severity of internal cracks in this study using CNN. The authors experimented with verifying CNN's ability to evaluate the severity of cracks in cross-sectional images. The classification accuracy was 90%. In this paper, we describe the experimental detail and future works.