"GO Stanford" Dataset
In order to train VUNet, we need two kinds of dataset to train SNet and DNet, individually. For SNet, which can achieve the static transformation by the robot pose change, we use previous dataset, "GO Stanford 2" (GS2), because most of GS2 doesn't include the dynamic objects. For the training of DNet, we collected new dataset, "GO Stanford 3" (GS3) at the Stanford University campus to predict the motion of the dynamic objects.
Download "GO Stanford 2" and "GO Stanford 3"
We open our dataset, "GO Stanford 2" and "GO Stanford 3".
GO Stanford 2 (GS2)
GS2 includes left and right images of fish eye image from Ricoh THETA S and, corresponding robot linear and angular velocities. In the directory, there are some files named as "dataset_'X'_'Y'pre_'Z'.txt". These are the list files to link the images and the corresponding robot velocities. Here, 'X' is R or L or "refres", which indicate right or left side of camera or the robot velocities. 'Y' is number of building in Stanford campus, and 'Z' is the identification number in building of 'Y'. If 'Z' includes 'F', its dataset indicates the augumented images and velocities by horizontal flipping.
GO Stanford 3 (GS3)
GS3 is dataset, which includes the pedesrians in the camera view with and without robot motion. The list file name in the directory is "dataset_'A'_'B'dyn'C'.txt". Here 'A' is R or L or "refres", which indicate right or left side of camera or the robot velocities. 'B' is the identification number for the environment. And, 'C' is none or "F". "F" indicates the augumented images and velocities by horizontal flipping. Note that the dataset of 'B' from 43 to 48 is with the robot motion.
GO Stanford 3
We place the mobile robot with fish eye camera in indoor and outdoor environment and collect the fisheye camera images at 46 fixed points in Stanford University. Total time length is 4.6 hours. The collected images include the dynamic object, e.g. pedesrian and vehicle.
Points
Hours
Images
Pedestrian Dataset
We collected small dataset "Pedestrian Dataset" as the part of GS3, which the robot travels in the dynamic environment only to evaluate our method not for the training. In this dataset, the pedestrian often crosses the future trajectroy of the robot.
scenes
Hours
Images
License
The datasets provided on this page are published under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License. This means that you must attribute the work in the manner specified by the authors, you may not use this work for commercial purposes and if you alter, transform, or build upon this work, you may distribute the resulting work only under the same license. If you are interested in commercial usage you can contact us for further options.