KITTI Dataset Exploration

Dependencies

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.

Dataset

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.

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