cd fully_sup
conda create -n blv python=3.7 -y
conda activate blv
conda install pytorch==1.10.0 torchvision==0.11.0 torchaudio==0.10.0 cudatoolkit=11.3 -c pytorch -c conda-forge
pip install openmim
mim install mmcv-full==1.4.0
pip install -e .Please follow this link dataset_prepare.md to setup the datasets.
For some models, you should download the corresponding pretrained checkpoints for the backbone manually.
cd fully_sup
python -u tools/train.py /path/to/the/config/file| Method | Backbone | mIoU | mIoU(tail) | config |
|---|---|---|---|---|
| HRNet-18 | OCRHead | 79.22 | 63.51 | config |
| HRNet-18 | +BLV | 79.94 | 66.70 | config |
| ResNet50 | UperHead | 78.28 | 62.56 | config |
| ResNet50 | +BLV | 78.63 | 64.57 | config |
| ResNet50 | PSPHead | 77.98 | 61.96 | config |
| ResNet50 | +BLV | 78.53 | 63.34 | config |
| ResNet101 | UperHead | 79.41 | 64.68 | config |
| ResNet101 | +BLV | 79.88 | 66.29 | config |
| MiT-b0 | SegformerHead | 76.85 | 67.58 | config |
| MiT-b0 | +BLV | 77.09 | 68.91 | config |
| Swin-T | K-NeT | 79.68 | 71.70 | config |
| Swin-T | +BLV | 80.11 | 72.94 | config |
| Vit-B16 | K-NeT | 76.48 | 68.25 | config |
| Vit-B16 | +BLV | 77.68 | 70.63 | config |
If you find this useful for your research, please cite the following paper.
@inproceedings{wang2023balancing,
title={Balancing Logit Variation for Long-tailed Semantic Segmentation},
author={Wang, Yuchao and Fei, Jingjing and Wang, Haochen and Li, Wei and Bao, Tianpeng and Wu, Liwei and Zhao, Rui and Shen, Yujun},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={19561--19573},
year={2023}
}The implementation of fully-supervised semantic segmentation task is based on mmsegmentation.
The implementation of unsupervised domain adaptived semantic segmentation task is based on HRDA.