Examination of the KITTI dataset.
Switch branches/tags
Nothing to show
Clone or download
Latest commit b8c4fbe Jul 3, 2018
Type Name Latest commit message Commit time
Failed to load latest commit information.
source Add logging routines. Mar 20, 2017
.gitignore Add `.gitignore`. Mar 20, 2017
LICENSE.txt Add license Jul 3, 2018
README.md Remove Binder links. Jan 9, 2018
kitti-dataset.ipynb Include user-customized limits in the interface. Jul 22, 2017
pcl_data.gif Add `README` and images. Mar 20, 2017


KITTI Dataset Exploration


Apart from the common dependencies like numpy and matplotlib notebook requires pykitti. You can install pykitti via pip using:

pip install pykitti

Project structure

File Description
kitti-dataset.ipynb Jupyter Notebook with dataset visualisation routines and output.
parseTrackletXML.py Methods for parsing tracklets (e.g. dataset labels), originally created by Christian Herdtweck.
utilities.py Convenient logging routines.


I have used one of the raw datasets available on KITTI website. See the first one in the list: 2011_09_26_drive_0001 (0.4 GB).

  • Length: 114 frames (00:11 minutes)
  • Image resolution: 1392 x 512 pixels
  • Labels: 12 Cars, 0 Vans, 0 Trucks, 0 Pedestrians, 0 Sitters, 2 Cyclists, 1 Trams, 0 Misc

I mainly focused on point cloud data and plotting labeled tracklets for visualisation. Cars are marked in blue, trams in red and cyclists in green.

Point cloud data with labels

For a more in-depth exploration and implementation details see notebook.