Awesome

URUR: Ultra-High Resolution Segmentation with Ultra-Rich Context

The URUR dataset is for Utra-high Resolution Segmentation, and proposed in CVPR 2023 paper "Ultra-High Resolution Segmentation with Ultra-Rich Context: A Novel Benchmark"

The dataset can be downloaded in Baidu Pan: https://pan.baidu.com/s/14mUaovxu0ha1skWm57pf1A (Password: 75x6)

Class Defination:

classes = [
"others",
"building",
"farmland",
"greenhouse",
"woodland",
"bareland",
"water",
"road"
]

colors = {
"others": [0, 0, 0],
"building": [230, 230, 230],
"greenhouse": [100, 100, 100],
"woodland": [200, 230, 160],
"farmland": [95, 163, 7],
"bareland": [255, 255, 100],
"water": [150, 200, 250],
"road": [240, 100, 80],
}

If you find our work helpful in your research or work, please cite our paper.

@InProceedings{Ji_2023_CVPR,
    author    = {Ji, Deyi and Zhao, Feng and Lu, Hongtao and Tao, Mingyuan and Ye, Jieping},
    title     = {Ultra-High Resolution Segmentation With Ultra-Rich Context: A Novel Benchmark},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2023},
    pages     = {23621-23630}
}