Road crack detection using deep convolutional neural network

• Published in: Proceedings - International Conference on Image Processing pp. 3708 - 3712
• Main Authors: Lei Zhang, Fan Yang, Zhang, Yimin Daniel, Zhu, Ying Julie
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2016
• Subjects: Boosting; convolution neural networks; Deep learning; Feature extraction; Probability; road crack detection; road survey; Roads; Support vector machines; Training
• ISSN: 2381-8549
• DOI: 10.1109/ICIP.2016.7533052

Automatic detection of pavement cracks is an important task in transportation maintenance for driving safety assurance. However, it remains a challenging task due to the intensity inhomogeneity of cracks and complexity of the background, e.g., the low contrast with surrounding pavement and possible shadows with similar intensity. Inspired by recent success on applying deep learning to computer vision and medical problems, a deep-learning based method for crack detection is proposed in this paper. A supervised deep convolutional neural network is trained to classify each image patch in the collected images. Quantitative evaluation conducted on a data set of 500 images of size 3264 χ 2448, collected by a low-cost smart phone, demonstrates that the learned deep features with the proposed deep learning framework provide superior crack detection performance when compared with features extracted with existing hand-craft methods.