#you need to download xception_pytorch_imagenet.pth and put it into the folder,imitmodel-----链接：https://pan.baidu.com/s/187MMiDJUMKvPUnhYwKM8Gg?pwd=kcjn 提取码：kcjn

This repository is a PyTorch implementation for semantic segmentation / scene parsing. The code is easy to use for training and testing on various datasets. The codebase mainly use Xception+ as backbone and can be easily adapted to other basic classification structures.Sample experimented datasets are eTRIMS, PASCAL VOC 2012,CamVid and Cityscapes.

1. Requirement: The proposed EDPNet was trained and predicted based on deep learning server using an NVIDIA GTX 3090 24 GB GPU and a 3.00 GHz Intel Core i9-10980 XE CPU64GB main memory. We implement the programming based on Pytorch-gpu 1.7.1 deep learning framework in Python 3.6 language. The NVIDIA GTX 3090 GPU is powered by CUDA 11.1.
2. Train:

• Download related datasets and symlink the paths to them as follows (you can alternatively modify the relevant paths specified in folder config):

  cd semseg
  mkdir -p dataset
  ln -s /path_to_cityscapes_dataset dataset/cityscapes
  

• Specify the gpu used in config then do training:

  sh tool/train.sh cityscapes EDPNet

3. Test:

• Download trained segmentation models and put them under folder specified in config or modify the specified paths.

  • For full testing (get listed performance):

    sh tool/test.sh cityscapes EDPNet

Code reference PSPNet, Xception+

PSPNet https://github.com/hszhao/semseg, Xception+ https://github.com/bubbliiiing/deeplabv3-plus-pytorch