This is the official PyTorch codes for the paper:

CiPR: An Efficient Framework with Cross-instance Positive Relations for Generalized Category Discovery
Shaozhe Hao, Kai Han, Kwan-Yee K. Wong
TMLR

TL;DR: We present an efficient GCD framework that designs a novel semi-supervised clustering method to generate reliable and high-purity cross-instance positive relations, incorporated into joint contrastive learning.

Create a conda environment cipr using

conda create -n cipr python=3.8.12
conda activate cipr
conda install pytorch==2.0.1 torchvision==0.15.2 pytorch-cuda=11.7 -c pytorch -c nvidia
pip install -r requirements.txt

The pretrained DINO weights can be downloaded here. Save the pretrained model to ./pretrain.

We use finegrained datasets, including:

• The Semantic Shift Benchmark (SSB) and Herbarium19

We also use generic image recognition datasets, including:

• CIFAR-10/100 and ImageNet

We train the model using

python run.py --dataset_name DATASET

and we test the model using

python run.py --dataset_name DATASET --mode test

The option of DATASET includes: cifar10, cifar100, imgnet100, cub, car, and herb.

We first obtain the extracted features to ./features using

bash scripts/get_feat.sh

With the obtained features, we test the model with our selective neighbor clustering (SNC) using

python eval_snc.py

The implementation of SNC can be found in ./snc/clustering.py. It is an efficient semi-supervised clustering method ready for deployment off the shelf.

Alternatively, we test with semi-supervised k-means using

python eval_sskmeans.py

With all obtained features, we can estimate the number of classes using

python class_estimate.py

If you use this code in your research, please consider citing our paper:

@article{hao2024cipr,
    title={Ci{PR}: An Efficient Framework with Cross-instance Positive Relations for Generalized Category Discovery},
    author={Shaozhe Hao and Kai Han and Kwan-Yee K. Wong},
    journal={Transactions on Machine Learning Research},
    issn={2835-8856},
    year={2024},
    url={https://openreview.net/forum?id=1fNcpcdr1o}}

This project is based on GCD. Thanks for the great work!