This is the implementation of CoinSeg (ICCV 2023). Link.

Requirements

All experiments in this paper are done with following environments:

• CUDA 11.6
• python (3.6.13)
• pytorch (1.7.1+cu110)
• torchvision (0.8.2+cu110)
• numpy (1.19.2)
• matplotlib
• pillow

Dataset preparing

Organize datasets in the following structure.

path_to_your_dataset/
    VOC2012/
        Annotations/
        ImageSet/
        JPEGImages/
        SegmentationClassAug/
        proposal100/
        
    ADEChallengeData2016/
        annotations/
            training/
            validation/
        images/
            training/
            validation/
        proposal_adetrain/
        proposal_adeval/

You can get proposal100, proposal_adetrain, proposal_adeval here (provided by previous CISS method MicroSeg).

Startup

We provide a training script script_train.py to facilitate the use of our proposed method. The script enables users to easily train CoinSeg with various settings, for example, the default config of CoinSeg is:

 cd tools 
 python -u script_train.py 15-1 0,1,2,3,4,5 0 --freeze_low  
    --conloss_proposal --conloss_prototype --KDLoss --KDLoss_prelogit --batch 16 
    --name swin_voc

If you want to evaluate model after training , add --test_only.

Cite

If you find our work to be helpful, please consider citing us:

@InProceedings{Zhang_2023_ICCV,
    author    = {Zhang, Zekang and Gao, Guangyu and Jiao, Jianbo and Liu, Chi Harold and Wei, Yunchao},
    title     = {CoinSeg: Contrast Inter- and Intra- Class Representations for Incremental Segmentation},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2023},
    pages     = {843-853}
}