https://github.com/jdai-cv/fada
(ECCV 2020) Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation
https://github.com/jdai-cv/fada
domain-adaptation semantic-segmentation
Last synced: 6 months ago
JSON representation
(ECCV 2020) Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation
- Host: GitHub
- URL: https://github.com/jdai-cv/fada
- Owner: JDAI-CV
- License: other
- Created: 2020-07-17T06:53:39.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-09-30T13:10:51.000Z (about 5 years ago)
- Last Synced: 2025-03-26T11:03:58.935Z (7 months ago)
- Topics: domain-adaptation, semantic-segmentation
- Language: Python
- Homepage:
- Size: 8.17 MB
- Stars: 141
- Watchers: 8
- Forks: 25
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation (ECCV 2020)
This is a [pytorch](http://pytorch.org/) implementation of [FADA](https://arxiv.org/abs/2007.09222).
### Prerequisites
- Python 3.6
- Pytorch 1.2.0
- torchvision from master
- yacs
- matplotlib
- GCC >= 4.9
- OpenCV
- CUDA >= 9.0
### Step-by-step installation
```bash
conda create --name fada -y python=3.6
conda activate fada
# this installs the right pip and dependencies for the fresh python
conda install -y ipython pip
pip install ninja yacs cython matplotlib tqdm opencv-python imageio mmcv
# follow PyTorch installation in https://pytorch.org/get-started/locally/
# we give the instructions for CUDA 9.2
conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=9.2 -c pytorch
```
### Getting started
- Download [The GTA5 Dataset]( https://download.visinf.tu-darmstadt.de/data/from_games/ )
- Download [The SYNTHIA Dataset]( http://synthia-dataset.net/download/808/ )
- Download [The Cityscapes Dataset]( https://www.cityscapes-dataset.com/ )
- Symlink the required dataset
```bash
ln -s /path_to_gta5_dataset datasets/gta5
ln -s /path_to_synthia_dataset datasets/synthia
ln -s /path_to_cityscapes_dataset datasets/cityscapes
```
- Generate the label statics file for GTA5 and SYNTHIA Datasets by running
```
python datasets/generate_gta5_label_info.py -d datasets/gta5 -o datasets/gta5/
python datasets/generate_synthia_label_info.py -d datasets/synthia -o datasets/synthia/
```
The data folder should be structured as follows:
```
├── datasets/
│ ├── cityscapes/
| | ├── gtFine/
| | ├── leftImg8bit/
│ ├── gta5/
| | ├── images/
| | ├── labels/
| | ├── gtav_label_info.p
│ ├── synthia/
| | ├── RAND_CITYSCAPES/
| | ├── synthia_label_info.p
│ └──
...
```
### Train
We provide the training script using 4 Tesla P40 GPUs. Note that when generating pseudo labels for self distillation, the link to the pseudo label directory should be updated [here](https://github.com/JDAI-CV/FADA/blob/98336a61f0fde633c6d504972fd782688fb8bd3a/core/datasets/dataset_path_catalog.py#L25).
```
bash train_with_sd.sh
```
### Evaluate
```
python test.py -cfg configs/deeplabv2_r101_tgt_self_distill.yaml resume g2c_sd.pth
```
#### Tip: For those who are interested in how performance change during the process of adversarial training, test.py also accepts directory as the input and the results will be stored in a csv file.
### Pretrained weights
Our pretrained models for Synthia -> CityScapes task(s2c) and GTA5 -> CityScapes task(g2c) are available via [Google Drive](https://drive.google.com/drive/folders/1M7mwfSX3fx4W9KUevZwdmCo4JISnfCI_).
### Visualization results
### Acknowledge
Some codes are adapted from [maskrcnn-benchmark](https://github.com/facebookresearch/maskrcnn-benchmark) and [semseg](https://github.com/hszhao/semseg). We thank them for their excellent projects.
### Citation
If you find this code useful please consider citing
```
@InProceedings{Haoran_2020_ECCV,
author = {Wang, Haoran and Shen, Tong and Zhang, Wei and Duan, Lingyu and Mei, Tao},
title = {Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation},
booktitle = {The European Conference on Computer Vision (ECCV)},
month = {August},
year = {2020}
}
```