/feature-self-reinforcement

Primary LanguagePython

Progressive Uncertain Feature Self-reinforcement for Weakly Supervised Semantic Segmentation

This repo is a PyTorch implementation for paper: Progressive Uncertain Feature Self-reinforcement for Weakly Supervised Semantic Segmentation

overview

Data Preparation

PASCAL VOC 2012

1. Download

wget http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar

2. Segmentation Labels

The augmented annotations are from SBD dataset. Here is a download link of the augmented annotations at DropBox. After downloading SegmentationClassAug.zip, you should unzip it and move it to VOCdevkit/VOC2012/.

VOCdevkit/
└── VOC2012
    ├── Annotations
    ├── ImageSets
    ├── JPEGImages
    ├── SegmentationClass
    ├── SegmentationClassAug
    └── SegmentationObject

MSCOCO 2014

1. Download

wget http://images.cocodataset.org/zips/train2014.zip
wget http://images.cocodataset.org/zips/val2014.zip

2. Segmentation Labels

To generate VOC style segmentation labels for COCO, you could use the scripts provided at this repo, or just download the generated masks from Google Drive.

COCO/
├── JPEGImages
│    ├── train2014
│    └── val2014
└── SegmentationClass
     ├── train2014
     └── val2014

Requirement

Please refer to requirements.txt

Our implementation incorporates a regularization term for segmentation. Please download and compile the python extension.

Train

bash train_voc.sh

bash train_coco.sh

Evaluation

bash infer_voc.sh

bash infer_coco.sh

Checkpoints

Coming soon.

Acknowledgement

This repo is built upon ToCo. Our work is greatly inspired by DINO. Many thanks to their brilliant works!