To read more about the dataset or download it, please visit https://www.man.eu/truckscenes
Our devkit is available and can be installed via pip:
pip install truckscenes-devkit
If you also want to install all the (optional) dependencies for running the visualizations:
pip install "truckscenes-devkit[all]"
For more details on the installation see installation
Download all archives from our download page or the AWS Open Data Registry.
Unpack the archives to the /data/man-truckscenes folder without overwriting folders that occur in multiple archives.
Eventually you should have the following folder structure:
/data/man-truckscenes
samples - Sensor data for keyframes.
sweeps - Sensor data for intermediate frames.
v1.0-* - JSON tables that include all the meta data and annotations. Each split (trainval, test, mini) is provided in a separate folder.
Please follow these steps to make yourself familiar with the MAN TruckScenes dataset:
- Read the dataset description.
- Explore the dataset videos.
- Download the dataset from our website.
- Make yourself familiar with the dataset schema
- Run the tutorial to get started:
- Read the MAN TruckScenes paper for a detailed analysis of the dataset.
@inproceedings{truckscenes2024,
title = {MAN TruckScenes: A multimodal dataset for autonomous trucking in diverse conditions},
author = {Fent, Felix and Kuttenreich, Fabian and Ruch, Florian and Rizwin, Farija and Juergens, Stefan and Lechermann, Lorenz and Nissler, Christian and Perl, Andrea and Voll, Ulrich and Yan, Min and Lienkamp, Markus},
booktitle = {Advances in Neural Information Processing Systems},
editor = {A. Globerson and L. Mackey and D. Belgrave and A. Fan and U. Paquet and J. Tomczak and C. Zhang},
pages = {62062--62082},
publisher = {Curran Associates, Inc.},
url = {https://proceedings.neurips.cc/paper_files/paper/2024/file/71ac06f0f8450e7d49063c7bfb3257c2-Paper-Datasets_and_Benchmarks_Track.pdf},
volume = {37},
year = {2024}
}
Copied and adapted from nuscenes-devkit
