search

chaofengc / IQA-PyTorch

👁️ 🖼️ 🔥PyTorch Toolbox for Image Quality Assessment, including PSNR, SSIM, LPIPS, FID, NIQE, NRQM(Ma), MUSIQ, TOPIQ, NIMA, DBCNN, BRISQUE, PI and more...
https://iqa-pytorch.readthedocs.io/
Other
2.03k stars 177 forks source link

question about the 4-parameter logistic function in TOPIQ #103

Closed cyclopssssss closed 1 year ago

cyclopssssss commented 1 year ago

Hi, thank you for such an excellent repository! In your paper TOPIQ, you mentioned before calculating the PLCC, the scores need to go through a logistic func FJ_F7WO@QYHB~8@ C$2PV15 but i don't see any operations like this in the topiq_arch.py, how to deal with it?

chaofengc commented 1 year ago

Please see explanations in readme: https://github.com/chaofengc/IQA-PyTorch#performance-evaluation-protocol

In this repository, we did not use logistic correction because most deep learning methods already achieve high performance. The paper just follow previous work for fair comparison.

If you want to use logistic correction, please refer to codes in https://github.com/chaofengc/IQA-PyTorch/blob/main/pyiqa/metrics/correlation_coefficient.py