Novel Road Crack Detection Technology Based on Deep Learning

Road crack is an important indicator of road detection, and it is necessary to detect pavement cracks. With the continuous progress of science and technology, many methods are applied to crack detection. Traditional detection methods rely on manual identification, which requires technicians to label images one by one, which is inefficient and causes many errors. At the same time, the commonly used image processing methods are easily affected by lighting, pavement stains, and other factors, resulting in unstable detection results. In the research of road surface crack detection, the identification and classification of cracks have been used but the analysis of specific characteristics of cracks is lacking as it is not easy to measure the characteristic value of cracks. Using the deep learning method in computer science and technology, we explored three methods of crack detection and proposed a novel road crack detection technology based on deep learning. Practical Experiments show that our method has excellent performance in road crack detection and crack feature measurement.