What is here

Three scenes from the KITTI Road/Lane benchmark, included so that the lab handouts have something to illustrate. This is not the dataset. Run fetch-kitti.sh to download the full 289-image training split.

Each scene has a left image (image_2/), the corresponding right stereo image (image_3/), and a calibration file (calib/).

Ground truth encoding

Ground truth images are colour coded:

• magenta — the positive region (road surface, or the ego lane)
• red — the negative region
• black — “do not care” pixels, excluded from scoring

Files named *_road_* cover the whole road surface; files named *_lane_* cover the ego lane only and exist for the “um” category only.

Calibration

Each calib/*.txt contains the projection matrices P0–P3 (P2 is the left colour camera, P3 the right), the rectifying rotation R0_rect, and the rigid transforms between the sensors. The stereo baseline can be recovered from the translation components of P2 and P3.

Licence

The KITTI Vision Benchmark Suite is copyright Andreas Geiger, Philip Lenz and Raquel Urtasun, and is released under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 licence. These images are reproduced here for coursework only.

If you refer to this data in a report, cite:

A. Geiger, P. Lenz and R. Urtasun, “Are we ready for autonomous driving? The KITTI vision benchmark suite,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012.