Exploring CLIP for Assessing the Look and Feel of Images (AAAI 2023)

Paper

Jianyi Wang, Kelvin C.K. Chan, Chen Change Loy

S-Lab, Nanyang Technological University

TODO

• Website release
• Code release

Dependencies and Installation

The same as MMEditing, currently is using an old version.

conda create -n clipiqa python=3.6
pip install torch==1.8.1+cu101 torchvision==0.9.1+cu101 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html # For CUDA 10.1
pip install -r requirements.txt
pip install -v -e .

Running Examples

Test CLIP-IQA on KonIQ-10k

python demo/clipiqa_koniq_demo.py

Test CLIP-IQA on Live-iWT

python demo/clipiqa_liveiwt_demo.py

Train CLIP-IQA+ on KonIQ-10k

# Support dist training as MMEditing
python tools/train.py configs/clipiqa/clipiqa_coop_koniq.py

Test CLIP-IQA+ on KonIQ-10k (Checkpoint)

python demo/clipiqa_koniq_demo.py --config configs/clipiqa/clipiqa_coop_koniq.py --checkpoint ./iter_80000.pth

[Note] You may change prompts for different datasets, please refer to config files for details.

[Note] For testing on a single image, please refer to here for details.

Demo

✨ Versatile Quality Assessment

✨ Demo for IQA on SPAQ

✨ Demo for Abstract Perception on AVA

For more evaluation, please refer to our paper for details.

Citation

If our work is useful for your research, please consider citing:

@inproceedings{wang2022exploring,
    author = {Wang, Jianyi and Chan, Kelvin CK and Loy, Chen Change},
    title = {Exploring CLIP for Assessing the Look and Feel of Images},
    booktitle = {AAAI},
    year = {2023}
}

License

This project is licensed under NTU S-Lab License 1.0. Redistribution and use should follow this license.

Acknowledgement

This project is based on MMEditing and CLIP. Thanks for their awesome works.

Contact

If you have any question, please feel free to reach me out at [email protected].