Vision-Language Consistency Guided Multi-modal Prompt Learning for Blind AI Generated Image Quality Assessment

Introduction

In this letter, we propose vision-language consistency guided multi-modal prompt learning for blind AGIQA, dubbed CLIP-AGIQA. Specifically, we introduce learnable textual and visual prompts in language and vision branches of CLIP models, respectively. Moreover, we design a text-to-image alignment quality prediction task, whose learned vision-language consistency knowledge is used to guide the optimization of the above multi-modal prompts.

Train and Test

First, download datasets AGIQA3k and AGIQA2023.

Second, update path of datasets defined in train_test_clip_auxiliary.py

path = {
    'AGIQA3k': '/home/fujun/datasets/iqa/AGIQA-3K',
    'AGIQA2023': '/home/fujun/datasets/iqa/AIGC2023/DATA/'
}

Third, train and test the model using the following command:

python train_test_clip_auxiliary.py --dataset AGIQA3k --model AGIQA

Finally, check the results in the folder ./log.

Acknowledgement

This project is based on MaPLe, DBCNN, and CLIP-IQA. Thanks for these awesome works.

Citation

Please cite the following paper if you use this repository in your research.

@article{fu2024vision,
  title={Vision-Language Consistency Guided Multi-modal Prompt Learning for Blind AI Generated Image Quality Assessment},
  author={Fu, Jun and Zhou, Wei and Jiang, Qiuping and Liu, Hantao and Zhai, Guangtao},
  journal={IEEE Signal Processing Letters},
  year={2024},
  publisher={IEEE}
}

Contact

For any questions, feel free to contact: [email protected]