Automated Crack Detection and Measurement Based on Mask R-CNN and Image Analysis with Mobile Application

Early detection and analysis of concrete cracks allow preventive measures to be taken to prevent damage and possible failure of civil structures. However, previous methods of crack analysis require manual labor or costly equipment. Smartphones nowadays have high-quality cameras at an attainable price. This study is about developing an automated crack detection and measurement system with a mobile application for image acquisition. A smartphone camera is used to capture crack images, Mask R-CNN for concrete surface crack detection, and image processing to measure crack width and length. A total of one-hundred and fifty (150) images were tested. The system can successfully perform crack detection, crack measurement, and suggest crack condition if it is safe or unsafe. The system achieved 85% crack detection precision, ±0.15mm error on width measurement, and less than 10% relative error on length measurement.