ChangeSim is a dataset mainly used for online scene change detection. By collecting data in simulated environments, researchers obtain multimodal sensor data and accurate ground truth annotations, such as RGB images, depth images, semantic segmentation, change segmentation, camera pose, and 3D reconstruction. These simulated environments have non-target changes, such as changes in air turbidity and lighting conditions, as well as changes in target objects in industrial indoor environments. Different from previous online SCD datasets where models are evaluated given only well-aligned image pairs, ChangeSim also provides the original unpaired sequences, providing an opportunity to develop online SCD models in an end-to-end manner.

Citation

@inproceedings{park2021changesim,
author = {Park, Jin-Man and Jang, Jae-hyuk, and Yoo, Sahng-Min and Lee, Sun-Kyung and Kim, Ue-hwan and Kim, Jong-Hwan},
title = {ChangeSim: Towards End-to-End Online Scene Change Detection in Industrial Indoor Environments},
booktitle={2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
year = {2021},
organization = {IEEE},
url = {https://arxiv.org/abs/2103.05368},
}