zwuser1227 / mednet-edge-detection

Edge detection networks inspired by neural mechanisms of selective attention

MEDNet

This repository contains the Implementation details of the paper "Edge detection networks inspired by neural mechanisms of selective attention in biological visual cortex". The address of the paper is at (https://www.frontiersin.org/articles/10.3389/fnins.2022.1073484/full)

Citations

If you are using the code/model/data provided here in a publication, please consider citing our paper:

@article{zhangedge,
  title={Edge detection networks inspired by neural mechanisms of selective attention in biological visual cortex},
  author={Zhang, Zhenguang and Lin, Chuan and Qiao, Yakun and Pan, Yongcai},
  journal={Frontiers in Neuroscience},
  pages={2022},
  publisher={Frontiers}
}

Get Start

1、Download our code.
2、prepare the dataset.
3、Configure the environment.
4、If Windows system, please modify the dataset in cfgs.yaml.
5、Run the "train.py".

文件夹中包含模型文件（model.py） 训练用文件（train.py） 测试用文件(test,py) 以及其他相关文件(cfgs...)

The folder contains model files (model.py)
training files (train.py)
testing files (test.py) and other related files (cfgs ...).

Result

BSDS：Test results of this model on BSDS500 data set.
NYUD：Test results of this model on NYUD data set.
BIPED：Test results of this model on BIPED data set.
Other data and procedures will be made public after the paper is received.

Datsets

We use the links in RCF Repository (really thanks for that). The augmented BSDS500, PASCAL VOC, and NYUD datasets can be downloaded with:

  wget http://mftp.mmcheng.net/liuyun/rcf/data/HED-BSDS.tar.gz
  wget http://mftp.mmcheng.net/liuyun/rcf/data/PASCAL.tar.gz
  wget http://mftp.mmcheng.net/liuyun/rcf/data/NYUD.tar.gz

We use the links in DexiNed Repository (really thanks for that).
We used the first version before the second version came out.
MBIPED Dataset is Here:

  https://drive.google.com/drive/folders/1lZuvJxL4dvhVGgiITmZsjUJPBBrFI_bM?usp=sharing

Test results on dataset MBIPED:

Methods	ODS	ODS	AP
RCF	.849	.861	.906
BDCN	.890	.899	.934
DexiNed-f	.895	.900	.927
DexiNed-a	.893	.897	.940
MEDNet-a-SS	.896	.900	.920

Reference and Acknowledgments

When building our codeWe referenced the repositories as follow:
1.DRC 2.RCF 3.HED 4.DexiNed