This is a road image sequense for lane detection in different scenes, such as the inner city, urban. This is the dataset used to evaluate the algorithm described in [1]. There are 42966 frames which cover: (1) different disturbances, e.g. road paint, vehicles, mud, writings and lens flare; (2) various illumination conditions, including daytime, dusk, night with or without lighting and night with reflecting puddles; (3) different road types, e.g. inter-city road, crowded urban road and beltway.
[1] Zhi Huang, Baozheng Fan, Xiaolin Song: Robust lane detection and tracking using multiple visual cues under stochastic lane shape conditions