A general file client to access files in different backend. The client loads a file or text in a specified backend from its path and retur

Conv and residual block used in BasicVSR. Args: num_in_ch (int): Number of input channels. Default: 3. num_out_ch (int): Numb

ResNet block that uses SPADE. It differs from the ResNet block of pix2pixHD in that it takes in the segmentation map as input, learns the s

Conv Layer used in StyleGAN2 Discriminator. Args: in_channels (int): Channel number of the input. out_channels (int): Channel

The registry that provides name -> object mapping, to support third-party users' custom modules. To create a registry (e.g. a backbo

Residual block without BN. It has a style of: ---Conv-ReLU-Conv-+- |________________| Args: num_feat (int):

Equalized Linear as StyleGAN2. Args: in_channels (int): Size of each sample. out_channels (int): Size of each output sample.

InceptionC block patched for FID computation

Spatial feature transform (SFT) with upsampling block. Args: in_channel (int): Number of input channels. out_channel (int): N

The re-parameterizable block used in the ECBSR architecture. Paper: Edge-oriented Convolution Block for Real-time Super Resolution on Mobile De

This JPEG algorithm result is slightly different from cv2. DiffJPEG supports batch processing. Args: differentiable(bool): If Tru

The ECB block used in the ECBSR architecture. Paper: Edge-oriented Convolution Block for Real-time Super Resolution on Mobile Devices Ref

InceptionA block patched for FID computation

Residual block recommended in: http://torch.ch/blog/2016/02/04/resnets.html

Residual Dense Block. Used in RRDB block in ESRGAN. Args: num_feat (int): Channel number of intermediate features. num

SpyNet architecture. Args: load_path (str): path for pretrained SpyNet. Default: None.

Style conv used in StyleGAN2. Args: in_channels (int): Channel number of the input. out_channels (int): Channel number of the o

Style conv. Args: in_channels (int): Channel number of the input. out_channels (int): Channel number of the output.

Upsample, FIR filter, and downsample (smooth version). Args: resample_kernel (list[int]): A list indicating the 1D resample kernel

Modulated deformable conv for deformable alignment. Different from the official DCNv2Pack, which generates offsets and masks from the prec

Data normalization with mean and std. Args: rgb_range (int): Maximum value of RGB. rgb_mean (list[float]): Mean for RGB chan

Modulated Conv2d used in StyleGAN2. There is no bias in ModulatedConv2d. Args: in_channels (int): Channel number of the input.

Modulated Conv2d used in StyleGAN2. There is no bias in ModulatedConv2d. Args: in_channels (int): Channel number of the input.

Alignment module using Pyramid, Cascading and Deformable convolution (PCD). It is used in EDVR. Ref: EDVR: Video Restoration with

r""" Image to Patch Embedding Args: img_size (int): Image size. Default: 224. patch_size (int): Patch token size. Default: 4

r""" Image to Patch Unembedding Args: img_size (int): Image size. Default: 224. patch_size (int): Patch token size. Default:

Residual block with bilinear upsampling/downsampling. Args: in_channels (int): Channel number of the input. out_channels (int):

Temporal Spatial Attention (TSA) fusion module. Temporal: Calculate the correlation between center frame and neighboring frames;

To RGB (image space) from features. Args: in_channels (int): Channel number of input. num_style_feat (int): Channel number of s

To RGB from features. Args: in_channels (int): Channel number of input. num_style_feat (int): Channel number of style feature

Upsample module. Args: scale (int): Scale factor. Supported scales: 2^n and 3. num_feat (int): Channel number of intermediate

VGG network for feature extraction. In this implementation, we allow users to choose whether use normalization in the input feature and th

A basic Swin Transformer layer for one stage. Args: dim (int): Number of input channels. input_resolution (tuple[int]): Inpu

Basic Module for SpyNet.

Basic module of SPyNet. Note that unlike the architecture in spynet_arch.py, the basic module here contains batch normalization.

Merge patches into image

Splitting image into patches

Dequantize CbCr channel

CPU prefetcher. Args: loader: Dataloader.

JPEG Quantization for CbCr channels Args: rounding(function): rounding function to use

CUDA prefetcher. Ref: https://github.com/NVIDIA/apex/issues/304# It may consums more GPU memory. Args: loader: Data

Channel attention. Args: num_feat (int): Channel number of intermediate features. squeeze_factor (int): Channel squeeze facto

Channel attention used in RCAN. Args: num_feat (int): Channel number of intermediate features. squeeze_factor (int): Channel

Chroma subsampling on CbCr channels

Upsample chroma layers

Full JPEG compression algorithm Args: rounding(function): rounding function to use

Constant input. Args: num_channel (int): Channel number of constant input. size (int): Spatial size of constant input.

Constant input. Args: num_channel (int): Channel number of constant input. size (int): Spatial size of constant input.

Discrete Cosine Transformation

Full JPEG decompression algorithm Args: rounding(function): rounding function to use

A concatenation of N dense blocks. Args: num_feat (int): Number of channels in the blocks. Default: 64. num_grow_ch (int): G

A concatenation of 3 dense blocks with reduction in temporal dimension. Note that the output temporal dimension is 6 fewer the input temporal d

Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). From: https://github.com/rwightman/pytorch-image-mode

Dynamic upsampling filter used in DUF. Ref: https://github.com/yhjo09/VSR-DUF. It only supports input with 3 channels. And it applies the

EDVR feature extractor used in IconVSR. Args: num_input_frame (int): Number of input frames. num_feat (int): Number of featur

Enhanced Residual block without BN. There are three convolution layers in residual branch. It has a style of: ---Conv-ReLU-Conv

Sampler that restricts data loading to a subset of the dataset. Modified from torch.utils.data.distributed.DistributedSampler Support enla

Equalized Linear as StyleGAN2. Args: in_channels (int): Channel number of the input. out_channels (int): Channel number of th

First InceptionE block patched for FID computation

Second InceptionE block patched for FID computation

Helper for the face restoration pipeline (base class).

The GFPGAN architecture: Unet + StyleGAN2 decoder with SFT. It is the clean version without custom compiled CUDA extensions used in StyleGAN2.

Pretrained InceptionV3 network returning feature maps

Local Importance-based Pooling (Ziteng Gao et.al.,ICCV 2019)

Multi-scale dilation block.