Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/Haochen-Wang409/U2PL
[CVPR'22] Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels
https://github.com/Haochen-Wang409/U2PL
cityscapes computer-vision contrastive-learning cvpr2022 deep-learning mechine-learning pascal-voc pytorch semantic-segmentation semi-supervised-learning
Last synced: about 2 months ago
JSON representation
[CVPR'22] Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels
- Host: GitHub
- URL: https://github.com/Haochen-Wang409/U2PL
- Owner: Haochen-Wang409
- License: apache-2.0
- Created: 2022-03-02T04:35:02.000Z (almost 3 years ago)
- Default Branch: main
- Last Pushed: 2023-02-13T04:26:50.000Z (almost 2 years ago)
- Last Synced: 2024-08-02T12:21:58.824Z (5 months ago)
- Topics: cityscapes, computer-vision, contrastive-learning, cvpr2022, deep-learning, mechine-learning, pascal-voc, pytorch, semantic-segmentation, semi-supervised-learning
- Language: Python
- Homepage:
- Size: 8.59 MB
- Stars: 422
- Watchers: 5
- Forks: 59
- Open Issues: 9
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- Awesome-Semi-Supervised-Semantic-Segmentation - Code - Wang409/U2PL)| [Paper](https://arxiv.org/pdf/2203.03884.pdf)| (2022)
README
# Using Unreliable Pseudo Labels
Official PyTorch implementation of [Semi-Supervised Semantic Segmentation Using Unreliable Pseudo Labels](https://arxiv.org/abs/2203.03884), CVPR 2022.
:bell::bell::bell: The extension of this paper ([U2PL+](https://arxiv.org/abs/2306.02314)) has been accepted by **IJCV**! :bell::bell::bell:
Please refer to our **[project page](https://haochen-wang409.github.io/U2PL/)** for qualitative results.
[](https://paperswithcode.com/sota/semi-supervised-semantic-segmentation-on-21?p=semi-supervised-semantic-segmentation-using-2)
[](https://paperswithcode.com/sota/semi-supervised-semantic-segmentation-on-4?p=semi-supervised-semantic-segmentation-using-2)
[](https://paperswithcode.com/sota/semi-supervised-semantic-segmentation-on-9?p=semi-supervised-semantic-segmentation-using-2)
[](https://paperswithcode.com/sota/semi-supervised-semantic-segmentation-on-15?p=semi-supervised-semantic-segmentation-using-2)
> **Abstract.**
> The crux of semi-supervised semantic segmentation is to assign adequate pseudo-labels to the pixels of unlabeled images.
> A common practice is to select the highly confident predictions as the pseudo ground-truth, but it leads to a problem that most pixels may be left unused due to their unreliability.
> We argue that every pixel matters to the model training, even its prediction is ambiguous.
> Intuitively, an unreliable prediction may get confused among the top classes (*i.e*., those with the highest probabilities),
> however, it should be confident about the pixel not belonging to the remaining classes.
> Hence, such a pixel can be convincingly treated as a negative sample to those most unlikely categories.
> Based on this insight, we develop an effective pipeline to make sufficient use of unlabeled data.
> Concretely, we separate reliable and unreliable pixels via the entropy of predictions, push each unreliable pixel to a category-wise queue that consists of negative samples, and manage to train the model with all candidate pixels.
> Considering the training evolution, where the prediction becomes more and more accurate, we adaptively adjust the threshold for the reliable-unreliable partition.
> Experimental results on various benchmarks and training settings demonstrate the superiority of our approach over the state-of-the-art alternatives.
## Results
### PASCAL VOC 2012
Labeled images are selected from the ```train``` set of original VOC, ```1,464``` images in total.
And the remaining ```9,118``` images are all considered as unlabeled ones.
For instance, ```1/2 (732)``` represents ```732``` labeled images
and remaining ```9,850 (9,118 + 732)``` are unlabeled.
| Method | 1/16 (92) | 1/8 (183) | 1/4 (366) | 1/2 (732) | Full (1464) |
| --------------------------- | --------- | --------- | --------- | --------- | ----------- |
| SupOnly | 45.77 | 54.92 | 65.88 | 71.69 | 72.50 |
| U2PL (w/ CutMix) | 67.98 | 69.15 | 73.66 | 76.16 | 79.49 |
Labeled images are selected from the ```train``` set of augmented VOC, ```10,582``` images in total.
Following results are all trained under our own splits.
Training a model on different splits is recommended to measure the performance of a method.
You can train our U2PL on splits provided by [CPS](https://github.com/charlesCXK/TorchSemiSeg/tree/main/DATA/pascal_voc/subset_train_aug) or [ST++](https://github.com/LiheYoung/ST-PlusPlus/tree/master/dataset/splits/pascal).
| Method | 1/16 (662) | 1/8 (1323) | 1/4 (2646) | 1/2 (5291) |
| --------------------------- | ---------- | ---------- | ---------- | ---------- |
| SupOnly | 67.87 | 71.55 | 75.80 | 77.13 |
| U2PL (w/ CutMix) | 77.21 | 79.01 | 79.30 | 80.50 |
### Cityscapes
Labeled images are selected from the ```train``` set, ```2,975``` images in total.
| Method | 1/16 (186) | 1/8 (372) | 1/4 (744) | 1/2 (1488) |
| --------------------------- | ---------- | --------- | --------- | ---------- |
| SupOnly | 65.74 | 72.53 | 74.43 | 77.83 |
| U2PL (w/ CutMix) | 70.30 | 74.37 | 76.47 | 79.05 |
| U2PL (w/ AEL) | 74.90 | 76.48 | 78.51 | 79.12 |
## Checkpoints
- Models on PASCAL VOC 2012 (ResNet101-DeepLabv3+) can be found [here](https://drive.google.com/drive/folders/1_BcixhrEqJEMo3lzKTKPJ_7gCWEjUBWp).
- Models on Cityscapes with AEL (ResNet101-DeepLabv3+)
| 1/16 (186) | 1/8 (372) | 1/4 (744) | 1/2 (1488) |
| -------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------- |
| [Google Drive](https://drive.google.com/file/d/1IZb9cAKSHajAnWYFHDq_qDecKcJMXvl5/view?usp=sharing) | [Google Drive](https://drive.google.com/file/d/1QyG-qPdjaUZ2qEkQTr70ioLeNDOI943p/view?usp=sharing) | [Google Drive](https://drive.google.com/file/d/1xgIICfZw_5bf-FFH9GtdO_85kQCnQmsT/view?usp=sharing) | [Google Drive](https://drive.google.com/file/d/1IlN5IIUMO4pZ_b_9Xo6659AbUV0dllfM/view?usp=sharing) |
| [Baidu Drive](https://pan.baidu.com/s/1DF9WTiV2mLfY2V9SHP14_w)
Fetch Code: rrpd | [Baidu Drive](https://pan.baidu.com/s/1TjtHQM2tXy0H8b-reXMHJw)
Fetch Code: welw | [Baidu Drive](https://pan.baidu.com/s/1LyecRU6rPbrrxgxy2qI65w)
Fetch Code: qwcd | [Baidu Drive](https://pan.baidu.com/s/1ngb7mS0I6UMj1cff_40cYg)
Fetch Code: 4p8r |
## Installation
```bash
git clone https://github.com/Haochen-Wang409/U2PL.git && cd U2PL
conda create -n u2pl python=3.6.9
conda activate u2pl
pip install -r requirements.txt
pip install pip install torch==1.8.1+cu102 torchvision==0.9.1+cu102 -f https://download.pytorch.org/whl/torch_stable.html
```
## Usage
U2PL is evaluated on both Cityscapes and PASCAL VOC 2012 dataset.
### Prepare Data
For Cityscapes
Download "leftImg8bit_trainvaltest.zip" from: https://www.cityscapes-dataset.com/downloads/
Download "gtFine.zip" from: https://drive.google.com/file/d/10tdElaTscdhojER_Lf7XlytiyAkk7Wlg/view?usp=sharing
Next, unzip the files to folder ```data``` and make the dictionary structures as follows:
```angular2html
data/cityscapes
├── gtFine
│ ├── test
│ ├── train
│ └── val
└── leftImg8bit
├── test
├── train
└── val
```
For PASCAL VOC 2012
Refer to [this link](https://github.com/zhixuanli/segmentation-paper-reading-notes/blob/master/others/Summary%20of%20the%20semantic%20segmentation%20datasets.md) and download ```PASCAL VOC 2012 augmented with SBD``` dataset.
And unzip the files to folder ```data``` and make the dictionary structures as follows:
```angular2html
data/VOC2012
├── Annotations
├── ImageSets
├── JPEGImages
├── SegmentationClass
├── SegmentationClassAug
└── SegmentationObject
```
Finally, the structure of dictionary ```data``` should be as follows:
```angular2html
data
├── cityscapes
│ ├── gtFine
│ └── leftImg8bit
├── splits
│ ├── cityscapes
│ └── pascal
└── VOC2012
├── Annotations
├── ImageSets
├── JPEGImages
├── SegmentationClass
├── SegmentationClassAug
└── SegmentationObject
```
### Prepare Pretrained Backbone
Before training, please download ResNet101 pretrained on ImageNet-1K from one of the following:
- [Google Drive](https://drive.google.com/file/d/1nzSX8bX3zoRREn6WnoEeAPbKYPPOa-3Y/view?usp=sharing)
- [Baidu Drive](https://pan.baidu.com/s/1FDQGlhjzQENfPp4HTYfbeA) Fetch Code: 3p9h
After that, modify ```model_urls``` in ```semseg/models/resnet.py``` to ``````
### Train a Fully-Supervised Model
For instance, we can train a model on PASCAL VOC 2012 with only ```1464``` labeled data for supervision by:
```bash
cd experiments/pascal/1464/suponly
# use torch.distributed.launch
sh train.sh
# or use slurm
# sh slurm_train.sh
```
Or for Cityscapes, a model supervised by only ```744``` labeled data can be trained by:
```bash
cd experiments/cityscapes/744/suponly
# use torch.distributed.launch
sh train.sh
# or use slurm
# sh slurm_train.sh
```
After training, the model should be evaluated by
```bash
sh eval.sh
```
### Train a Semi-Supervised Model
We can train a model on PASCAL VOC 2012 with ```1464``` labeled data and ```9118``` unlabeled data for supervision by:
```bash
cd experiments/pascal/1464/ours
# use torch.distributed.launch
sh train.sh
# or use slurm
# sh slurm_train.sh
```
Or for Cityscapes, a model supervised by ```744``` labeled data and ```2231``` unlabeled data can be trained by:
```bash
cd experiments/cityscapes/744/ours
# use torch.distributed.launch
sh train.sh
# or use slurm
# sh slurm_train.sh
```
After training, the model should be evaluated by
```bash
sh eval.sh
```
### Train a Semi-Supervised Model on Cityscapes with AEL
First, you should switch the branch:
```bash
git checkout with_AEL
```
Then, we can train a model supervised by ```744``` labeled data and ```2231``` unlabeled data by:
```bash
cd experiments/city_744
# use torch.distributed.launch
sh train.sh
# or use slurm
# sh slurm_train.sh
```
After training, the model should be evaluated by
```bash
sh eval.sh
```
### Note
`````` means the number of GPUs for training.
To reproduce our results, we recommend you follow the settings:
- Cityscapes: ```4 * V100 (32G)``` for SupOnly and ```8 * V100 (32G)``` for Semi-Supervised
- PASCAL VOC 2012: ```2 * V100 (32G)``` for SupOnly and ```4 * V100 (32G)``` for Semi-Supervised
If you got ```CUDA Out of Memory``` error, please try training our method in [fp16](https://github.com/NVIDIA/apex) mode.
Or, change the ```lr``` in ```config.yaml``` in a linear manner (*e.g.*, if you want to train a SupOnly model on Cityscapes with 8 GPUs,
you are recommended to change the ```lr``` to ```0.02```).
If you want to train a model on other split, you need to modify ```data_list``` and ```n_sup``` in ```config.yaml```.
Due to the randomness of function ```torch.nn.functional.interpolate``` when ```mode="bilinear"```,
the results of semantic segmentation will not be the same EVEN IF a fixed random seed is set.
Therefore, we recommend you run 3 times and get the average performance.
## License
This project is released under the [Apache 2.0](LICENSE) license.
## Acknowledgement
The contrastive learning loss and strong data augmentation (CutMix, CutOut, and ClassMix)
are borrowed from **ReCo**.
We reproduce our U2PL based on **AEL** on branch ```with_AEL```.
- ReCo: https://github.com/lorenmt/reco
- AEL: https://github.com/hzhupku/SemiSeg-AEL
Thanks a lot for their great work!
## Citation
```bibtex
@inproceedings{wang2022semi,
title={Semi-Supervised Semantic Segmentation Using Unreliable Pseudo Labels},
author={Wang, Yuchao and Wang, Haochen and Shen, Yujun and Fei, Jingjing and Li, Wei and Jin, Guoqiang and Wu, Liwei and Zhao, Rui and Le, Xinyi},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2022}
}
```
## Contact
- Yuchao Wang, [email protected]
- Haochen Wang, [email protected]
- Jingjing Fei, [email protected]
- Wei Li, [email protected]