Joint Adversarial Learning for Domain Adaptation in Semantic Segmentation (AAAI2020)

A Pytorch Implementation of Joint Adversarial Learning for Domain Adaptation in Semantic Segmentation.

Framework

Example Results

Quantitave Results

Introduction

In this project, we use Pytorch 1.1.0 and CUDA version is 10.0.

Datasets

Datasets Preparation

Download the GTA5 Dataset as the source domain, and put it in the data/GTA5 folder
Download the Cityscapes Dataset as the target domain, and put it in the data/Cityscapes folder

Models

Pre-trained Models

In our experiments, we used two pre-trained models on ImageNet, i.e., VGG16 and ResNet101. Please download these two models from:

VGG16: Dropbox VT Server
ResNet101: Dropbox VT Server

Trained Models

Train

bash ./experiments/scripts/jal_train.sh train ./configs/jal/GTA_Citsycapes_vgg16_train.xml

Test

bash ./experiments/scripts/jal_train.sh test ./configs/jal/GTA_Citsycapes_vgg16_test.xml

Citation

If you find this repository useful, please cite our paper:

@inproceedings{zhang2020joint,
  title={Joint Adversarial Learning for Domain Adaptation in Semantic Segmentation},
  author={Zhang, Yixin and Wang, Zilei},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={34},
  number={04},
  pages={6877--6884},
  year={2020}
}

zhyx12 / JAL