Welcome to the LDIS Dataset

We present the LDIS (Light-Dark Indoor Scene) dataset to study the problem of Computer Vision algorithms under different lighting conditions for indoor scenes.

Overview

LDIS consists of samples collected from different indoor scenes under both well-illuminated and low-light conditions. Our dataset can be used to study a variety of tasks that involve lighting changes. LDIS features:

487 samples where each of them includes a pairs of aligned light and dark images.

For each image in the dataset we provide the corresponding depth images.

A testing set consists of 98 samples with full semantic annotation.

Please refer to our paper for more information about the dataset.

Downloads

To download the dataset, please fill out our end-user license agreement (EULA) and send it to ldis-admin-list@lehigh.edu. Once we review the form we will send you a link to download the dataset.

Citation

If you use the LDIS dataset, please cite the following paper:

@inproceedings{ying2022delving,
    title={Delving into Light-Dark Semantic Segmentation for Indoor Scenes Understanding},
    author={Ying, Xiaowen and Lang, Bo and Zheng, Zhihao and Chuah, Mooi Choo},
    booktitle={Proceedings of the 1st Workshop on Photorealistic Image and Environment Synthesis for Multimedia Experiments},
    pages={3--9},
    year={2022}
    }