EDTER

EDTER: Edge Detection with Transformer
Mengyang Pu, Yaping Huang, Yuming Liu, Qingji Guan and Haibin Ling
CVPR 2022

Please refer to supplementary material for more results.

Usage

Datasets

BSDS500

Download the augmented BSDS500 data (1.2GB) from here.

|-- data
    |-- BSDS
        |-- ImageSets
        |   |-- train_pair.txt
        |   |-- test.txt
        |   |-- pascal_train_pair.txt
        |-- train
        |   |-- aug_data
        |   |-- aug_data_scale_0.5
        |   |-- aug_data_scale_1.5
        |   |-- aug_gt
        |   |-- aug_gt_scale_0.5
        |   |-- aug_gt_scale_1.5
        |-- test
        |   |-- 2018.jpg
        ......

Pre-trained model

If you are unable to download due to network reasons, you can download the pre-trained model from here and here.

Training

The training of Stage I

./tools/dist_train.sh ${CONFIG_FILE} ${GPU_NUM} 
# For example, train Stage I on BSDS500 dataset with 8 GPUs
./tools/dist_train.sh configs/bsds/EDTER_BIMLA_320x320_80k_bsds_bs_8.py 8

The training of Stage II

Change the '--global-model-path' in train_local.py.

./tools/dist_train_local.sh ${GLOBALCONFIG_FILE} ${CONFIG_FILE} ${GPU_NUM} 
# For example, train Stage II on BSDS500 dataset with 8 GPUs
./tools/dist_train_local.sh configs/bsds/EDTER_BIMLA_320x320_80k_bsds_bs_8.py configs/bsds/EDTER_BIMLA_320x320_80k_bsds_local8x8_bs_8.py 8

Testing

Single-scale testing

Change the '--config', '--checkpoint', and '--tmpdir' in test.py.

python test.py

Multi-scale testing

Change the '--globalconfig', '--config', '--global-checkpoint', '--checkpoint', and '--tmpdir' in test_local.py.
Use the config file ending in _ms.py in configs/EDTER.

python test_local.py

Results

If you want to compare your method with EDTER, you can download the precomputed results here.

Download Pre-trained model.

model	Pre-trained Model
BSDS-StageI	Coming Soon
BSDS-StageII	Coming Soon
BSDS-Aug-StageI	Coming Soon
BSDS-Aug-StageII	Coming Soon

Acknowledgments

We thank the anonymous reviewers for valuable and inspiring comments and suggestions.
Thanks to previous open-sourced repo:
SETR
MMsegmentation

Reference

@InProceedings{Pu_2022_CVPR,
    author    = {Pu, Mengyang and Huang, Yaping and Liu, Yuming and Guan, Qingji and Ling, Haibin},
    title     = {EDTER: Edge Detection With Transformer},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2022},
    pages     = {1402-1412}
}

dungmn / EDTER