Application of deep convolutional neural networks for automated and rapid identification and computation of crack statistics of thin cracks in strain hardening cementitious composites (SHCCs)

Characterization of surface cracks is one of the key steps towards condititional assessment and understanding the durability of strain hardening cementitious composites (SHCCs). Under laboratory conditions, surface crack statistics can be obtained from images of specimen surfaces through manual inspection. This is subjective, time consuming and laborious. In order to automate this process a framework encompassing a deep learning model for rapid identification and computation of crack parameters of thin SHCC cracks in presented in this work. A tailored deep convolutiontional neural network (TDCNN) was trained to detect thin cracks and then crack parameters were computed using image processing technique. The proposed technique does not require optimal lighting conditions, proper surface treatment, and prior (manual) selection of the correct region for proper inference.. The results from the controlled study suggest that the inference ability of TDCNN is reasonably good, resilient against epistemic uncertainty and tunable for completely independent but adverse observations. From the crack pattern computed using TDCCN, crack parameters-average crack width (ACW) and crack density (CD) can be calculated to facilitate conditional and durability assessment in a practical environment.