Guoxi Liu

Automated detection of expressway pavement cracks is critically important for infrastructure inspection and maintenance. In this task, range information provides a key geometric cue for distinguishing genuine cracks from pavement textures, shadows, stains, and other visual interference. However, industrial-grade pavement data acquisition equipment is expensive, and pixel-wise annotation is labor intensive. Consequently, most existing pavement crack datasets contain only visual imagery and cannot precisely represent the geometric morphology of the pavement surface. GTCrack was developed to alleviate this data bottleneck. A survey vehicle equipped with two laser line profilers was used to…