Stair dataset with depth maps

We provide a stair dataset with depth maps and fine annotations for stair detection research.
The training set contains 2388 RGB maps and 2388 corresponding depth maps, and the validation set contains 608 RGB maps and 608 corresponding depth maps. Each label contains the classification(convex/concave) and the location of the two endpoints of each stair line. All the images are padded and resized to 512 × 512 to simplify the data loading process. All the images in the dataset are taken in the actual scenes. In detail, we use the RealSense Depth Camera D435i to collect images from actual scenes at Beihang University and some public gardens.
The annotation form of the dataset is as follows:
cls x1 y1 x2 y2/n
...
Each stair line is represented by the above five-tuple data, where cls represents the class of the stair line, 0 represents a convex line and 1 represents a concave line. X1 and y1 represent the coordinates of the left endpoint of the stair line, and x2 and y2 represent the coordinates of the right endpoint of the stair line. The label is stored in a text file and associated with the RGB map and depth map through the file name.
This dataset can be used for the development of stair perception in intelligent autonomous system, such as humanoid robots, rescue robots and exoskeleton robots.