IVS is a dataset containing infrared and visible images and their corresponding pedestrian masks (mostly infrared images) that can be used for tasks such as registration and segmentation.
The IVS dataset contains two parts, the images as well as the mask labels. The features of the IVS Dataset include:
- Data format: PNG
- Image resolution: 640 x 480
- Label resolution: 640 x 480
- Dataset size: approximately 9.53 GB
If this dataset is useful for your research, please cite our paper.
@article{xie2023semantics,
title={Semantics lead all: Towards unified image registration and fusion from a semantic perspective},
author={Xie, Housheng and Zhang, Yukuan and Qiu, Junhui and Zhai, Xiangshuai and Liu, Xuedong and Yang, Yang and Zhao, Shan and Luo, Yongfang and Zhong, Jianbo},
journal={Information Fusion},
pages={101835},
year={2023},
publisher={Elsevier}
}