First-person view RGB-D dataset

We provide a varied dataset containing RGB-D data with the goal to stablish a benchmark for the evaluation of systems performing NAVI (Navigation Assistance for the Visually Impaired) based on RGB-D devices. Our dataset contains the color and depth data of an Asus XtionPro Live camera along different indoor scenarios. The data was recorded at 30Hz (full frame rate) with a 640x480 sensor resolution. All scenarios have been recorded in rosbag files and compressed in format 7z. All files contain XYZRGB point-cloud data and most of them have recorded raw images too. The RGB-D device was carried by a person in order to simulate a realistic situation.

We also provide a set of 32 calibration images and a calibration report of the RGB-D device. The calibration was carried out with the Kinect-stereo-calib software developed by the Mobile Robot Programming Toolkit. The calibration used a 8x10 chessboard of 22.8x24.3mm squares. More details can be found in webpage:
http://webdiis.unizar.es/%7Eglopez

More information about the dataset can be found in the related publication:
A. Aladren, G. Lopez-Nicolas, L. Puig, and J. J. Guerrero. Navigation Assistance for the Visually Impaired Using RGB-D Sensor with Range Expansion. IEEE Systems Journal, vol. 10, no. 3, pp. 922-932, Sept. 2016.