Chaofeng Chen

They are available in the release section: https://github.com/chaofengc/ITER/releases, and will be downloaded automatically if you tried the codes.

Thanks for your interest! I checked the code, and did not find anything wrong currently. The `topiq_nr-face` metric have the following steps for test: 1. Detect and extract face in...

It seems that your images are too large. Reading and processing 48 megapixel image might be slow. You may try to resize all images (**keep aspect ratio**) to less than...

Thanks to your information. There is a bug in the following shape permutation code: https://github.com/chaofengc/IQA-PyTorch/blob/aac09d01a38f06cae0d9048c945b10e83bae9f21/pyiqa/archs/liqe_arch.py#L105 After unfold the shape should be `(b, c, ph, pw, h, w)` and permutation should...

The issue is likely due to the large size of your image, which can cause an OOM error. To take full advantage of the GPU’s parallel processing power, the `entropy`...

您提到的这些方法是通过random sample一些patch来做的，所以每次结果会不太一样。在最新的版本里面，random seed应该是固定的。请更新版本在尝试一下。

Thanks to your information. Please note that the implementation follows the codes below: - Official matlab codes: https://github.com/utlive/BRISQUE - Python re-implementation: https://github.com/bukalapak/pybrisque **And there is no strict guarantee of [0,...

That's not what I mean and not true. Let's make it clearer: - Brisque has no strict guarantee of [0, 100] score range, even when you use official matlab codes....

I may consider to support it when I have time

Inception score is added, you now may use it like `pyiqa.create_metric("inception_score")` The results are calibrated with [torch-fidelity](https://github.com/toshas/torch-fidelity) using the command `fidelity --gpu 0 --samples-find-ext bmp,BMP --isc --isc-splits 1 --input1 ResultsCalibra/dist_dir/`