Official implementation of our CVPR2025 paper "GET: Unlocking the Multi-modal Potential of CLIP for Generalized Category Discovery"

Given unlabelled datasets containing both old and new categories, generalized category discovery (GCD) aims to accurately discover new classes while correctly classifying old classes. Current GCD methods only use a single visual modality of information, resulting in a poor classification of visually similar classes. As a different modality, text information can provide complementary discriminative information, which motivates us to introduce it into the GCD task. However, the lack of class names for unlabelled data makes it impractical to utilize text information. To tackle this challenging problem, in this paper, we propose a Text Embedding Synthesizer (TES) to generate pseudo text embeddings for unlabelled samples. Specifically, our TES leverages the property that CLIP can generate aligned vision-language features, converting visual embeddings into tokens of the CLIP’s text encoder to generate pseudo text embeddings. Besides, we employ a dual-branch framework, through the joint learning and instance consistency of different modality branches, visual and semantic information mutually enhance each other, promoting the interaction and fusion of visual and text knowledge. Our method unlocks the multi-modal potentials of CLIP and outperforms the baseline methods by a large margin on all GCD benchmarks, achieving new state-of-the-art.

We use the standard benchmarks in this paper, including:

• Generic datasets：CIFAR-10/100 and ImageNet-100/1K

• Fine-grained datasets: The Semantic Shift Benchmark (SSB) and Herbarium19

pip install -r requirements.txt

Set paths to datasets and desired log directories in config.py

Train GET on the Stanford Cars dataset:

bash scripts/run_scars.sh

If you find this repo useful for your research, please consider citing our paper:

@article{wang2024get,
  title={GET: Unlocking the Multi-modal Potential of CLIP for Generalized Category Discovery},
  author={Wang, Enguang and Peng, Zhimao and Xie, Zhengyuan and Yang, Fei and Liu, Xialei and Cheng, Ming-Ming},
  journal={arXiv preprint arXiv:2403.09974},
  year={2024}
}

Our codebase relies on SimGCD. Thanks for their excellent work!

This project is licensed under the MIT License - see the LICENSE file for details.

If you have further questions or discussions, feel free to contact me:

Enguang Wang ([email protected])