Datasets:
You need to agree to share your contact information to access this dataset
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
You agree to use this dataset for non-commercial research purposes only, in accordance with the CC BY-NC 4.0 license, and to cite the accompanying paper in any resulting work.
Log in or Sign Up to review the conditions and access this dataset content.
Liebherr Product (LP) Dataset
Liebherr Product (LP) is a self-collected object-detection dataset of construction machines, containing over 15,000 high-quality images across 23 categories of construction machinery, including articulated dump trucks, bulldozers, combined piling and drilling rigs, various cranes, excavators, loaders, and more.
It accompanies the paper DART: An automated end-to-end object detection pipeline with data Diversification, open-vocabulary bounding box Annotation, pseudo-label Review, and model Training (Expert Systems with Applications, 2024).
Structure
images/
├── articulated dump truck/
├── bulldozer/
├── ...
└── wheel loader/ # 23 class folders in total
LICENSE
The 23 class names and their label ids are listed in
classes.json
in the code repository. Pseudo-labels, metadata, and processing tools also live in
the code repository under Liebherr_Product/.
Usage
Once your access request is approved, log in and download:
hf auth login
hf download Moonxc/Liebherr_Product --repo-type dataset --local-dir Liebherr_Product
or in Python:
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="Moonxc/Liebherr_Product",
repo_type="dataset",
local_dir="Liebherr_Product",
)
Extract / place the images folder at ./Liebherr_Product/images/ in the code
repository to reproduce the pipeline.
License
Liebherr Product dataset by Chen Xin is licensed under CC BY-NC 4.0. Non-commercial use only.
Citation
@article{xin2024dart,
title={DART: An automated end-to-end object detection pipeline with data Diversification, open-vocabulary bounding box Annotation, pseudo-label Review, and model Training},
author={Xin, Chen and Hartel, Andreas and Kasneci, Enkelejda},
journal={Expert Systems with Applications},
pages={125124},
year={2024},
publisher={Elsevier}
}
- Downloads last month
- -