CurveLanes is a new benchmark lane detection dataset containing 150K lane images. The dataset includes many complex scenes, such as S-lanes, Y-lanes, nighttime and multi-lane scenes. The training set contains 100,000 images, the validation set contains 20,000 images, and the test set contains 30,000 images. Most images have a resolution of 2650 x 1440 pixels. This dataset uses cubic spline curves to manually annotate all lane lines. CurveLanes is currently the largest and most challenging lane detection dataset.

Citation

text @InProceedings{CurveLane-NAS, Authors: {Hang Xu, Shaoju Wang, Xinyue Cai, Wei Zhang, Xiaodan Liang, Zhenguo Li} title = {CurveLane-NAS: Unifying Lane-Sensitive Architecture Search and Adaptive Point Blending}, booktitle = {ECCV}, year = {2020}, }