pyiqa.archs.dists_arch

DISTS metric introduced in

@article{ding2020iqa,

title={Image Quality Assessment: Unifying Structure and Texture Similarity}, author={Ding, Keyan and Ma, Kede and Wang, Shiqi and Simoncelli, Eero P.}, journal = {CoRR}, volume = {abs/2004.07728}, year={2020}, url = {https://arxiv.org/abs/2004.07728}

}

Created by: https://github.com/dingkeyan93/DISTS/blob/master/DISTS_pytorch/DISTS_pt.py Re-implemented by: Jiadi Mo (https://github.com/JiadiMo)

Module Contents

pyiqa.archs.dists_arch.default_model_urls[source]
class pyiqa.archs.dists_arch.L2pooling(filter_size=5, stride=2, channels=None, pad_off=0)[source]

Bases: torch.nn.Module

forward(input)[source]
class pyiqa.archs.dists_arch.DISTS(pretrained=True, pretrained_model_path=None, **kwargs)[source]

Bases: torch.nn.Module

DISTS model. :param pretrained_model_path: Pretrained model path. :type pretrained_model_path: String

forward_once(x)[source]
forward(x, y)[source]

Compute IQA using DISTS model.

Parameters:

  • x (-) – An input tensor with (N, C, H, W) shape. RGB channel order for colour images.

  • y (-) – An reference tensor with (N, C, H, W) shape. RGB channel order for colour images.

Returns:

Value of DISTS model.