This repository contains the code of semantic segmentation for the paper Vision Transformer with Deformable Attention [arXiv], and DAT++: Spatially Dynamic Vision Transformerwith Deformable Attention (extended version)[arXiv].

This code is based on mmsegmentation and Swin Segmentation. To get started, you can follow the instructions in Swin Transformer.

Other links:

• Classification
• Detection

In addition to the dependencies of the classification codebase, the following packages are required:

• mmcv-full == 1.4.0
• mmsegmentation == 0.29.0

To evaluate a pretrained checkpoint, please download the pretrain weights to your local machine and run the mmsegmentation test scripts as follows:

# single-gpu testing
python tools/test.py <CONFIG_FILE> <SEG_CHECKPOINT_FILE> --eval mIoU

# multi-gpu testing
bash tools/dist_test.sh <CONFIG_FILE> <SEG_CHECKPOINT_FILE> <GPU_NUM> --eval mIoU

# multi-gpu, MS testing
bash tools/dist_test.sh <CONFIG_FILE> <SEG_CHECKPOINT_FILE> <GPU_NUM> --aug-test --eval mIoU

Please notice: Before training or evaluation, please set the data_root variable in configs/_base_/datasets/ade20k.py to the path where ADE20K data stores.

Since evaluating models needs no pretrain weights, you can set the pretrained = None in <CONFIG_FILE>.

To train a segmentor with pre-trained models, run:

# single-gpu training
python tools/train.py <CONFIG_FILE>

# multi-gpu training
bash tools/dist_train.sh <CONFIG_FILE> <GPU_NUM> 

Please notice: Make sure the pretrained variable in <CONFIG_FILE> is correctly set to the path of pretrained DAT model.

This code is developed on the top of Swin Transformer, we thank to their efficient and neat codebase. The computational resources supporting this work are provided by Hangzhou High-Flyer AI Fundamental Research Co.,Ltd.

If you find our work is useful in your research, please consider citing:

@article{xia2023dat,
    title={DAT++: Spatially Dynamic Vision Transformer with Deformable Attention}, 
    author={Zhuofan Xia and Xuran Pan and Shiji Song and Li Erran Li and Gao Huang},
    year={2023},
    journal={arXiv preprint arXiv:2309.01430},
}

@InProceedings{Xia_2022_CVPR,
    author    = {Xia, Zhuofan and Pan, Xuran and Song, Shiji and Li, Li Erran and Huang, Gao},
    title     = {Vision Transformer With Deformable Attention},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2022},
    pages     = {4794-4803}
}

If you have any questions or concerns, please send email to [email protected].