pyiqa.archs.tres_arch

TReS model.

Reference:

No-Reference Image Quality Assessment via Transformers, Relative Ranking, and Self-Consistency. S. Alireza Golestaneh, Saba Dadsetan, Kris M. Kitani WACV2022

Official code: https://github.com/isalirezag/TReS

Module Contents

pyiqa.archs.tres_arch.default_model_urls

class pyiqa.archs.tres_arch.Transformer(d_model=256, nhead=8, num_encoder_layers=6, num_decoder_layers=6, dim_feedforward=2048, dropout=0.1, activation='relu', normalize_before=False, return_intermediate_dec=False)

Bases: torch.nn.Module

forward(src, pos_embed)

class pyiqa.archs.tres_arch.TransformerEncoder(encoder_layer, num_layers, norm=None)

Bases: torch.nn.Module

forward(src, mask: torch.Tensor | None = None, src_key_padding_mask: torch.Tensor | None = None, pos: torch.Tensor | None = None)

class pyiqa.archs.tres_arch.TransformerEncoderLayer(d_model, nhead, dim_feedforward=2048, dropout=0.1, activation='relu', normalize_before=False)

Bases: torch.nn.Module

with_pos_embed(tensor, pos: torch.Tensor | None)

forward_post(src, src_mask: torch.Tensor | None = None, src_key_padding_mask: torch.Tensor | None = None, pos: torch.Tensor | None = None)

forward_pre(src, src_mask: torch.Tensor | None = None, src_key_padding_mask: torch.Tensor | None = None, pos: torch.Tensor | None = None)

forward(src, src_mask: torch.Tensor | None = None, src_key_padding_mask: torch.Tensor | None = None, pos: torch.Tensor | None = None)

class pyiqa.archs.tres_arch.PositionEmbeddingSine(num_pos_feats=64, temperature=10000, normalize=False, scale=None)

Bases: torch.nn.Module

This is a more standard version of the position embedding, very similar to the one used by the Attention is all you need paper, generalized to work on images.

forward(tensor_val)

class pyiqa.archs.tres_arch.L2pooling(filter_size=5, stride=1, channels=None, pad_off=0)

Bases: torch.nn.Module

forward(input)

class pyiqa.archs.tres_arch.TReS(network='resnet50', train_dataset='koniq', nheadt=16, num_encoder_layerst=2, dim_feedforwardt=64, test_sample=50, default_mean=[0.485, 0.456, 0.406], default_std=[0.229, 0.224, 0.225], pretrained=True, pretrained_model_path=None)

Bases: torch.nn.Module

forward_backbone(model, x)

forward(x)