diff options
Diffstat (limited to 'extensions-builtin/SwinIR')
| -rw-r--r-- | extensions-builtin/SwinIR/preload.py | 6 | ||||
| -rw-r--r-- | extensions-builtin/SwinIR/scripts/swinir_model.py | 172 | ||||
| -rw-r--r-- | extensions-builtin/SwinIR/swinir_model_arch.py | 867 | ||||
| -rw-r--r-- | extensions-builtin/SwinIR/swinir_model_arch_v2.py | 1017 |
4 files changed, 2062 insertions, 0 deletions
diff --git a/extensions-builtin/SwinIR/preload.py b/extensions-builtin/SwinIR/preload.py new file mode 100644 index 00000000..567e44bc --- /dev/null +++ b/extensions-builtin/SwinIR/preload.py @@ -0,0 +1,6 @@ +import os
+from modules import paths
+
+
+def preload(parser):
+ parser.add_argument("--swinir-models-path", type=str, help="Path to directory with SwinIR model file(s).", default=os.path.join(paths.models_path, 'SwinIR'))
diff --git a/extensions-builtin/SwinIR/scripts/swinir_model.py b/extensions-builtin/SwinIR/scripts/swinir_model.py new file mode 100644 index 00000000..9a74b253 --- /dev/null +++ b/extensions-builtin/SwinIR/scripts/swinir_model.py @@ -0,0 +1,172 @@ +import contextlib +import os + +import numpy as np +import torch +from PIL import Image +from basicsr.utils.download_util import load_file_from_url +from tqdm import tqdm + +from modules import modelloader, devices, script_callbacks, shared +from modules.shared import cmd_opts, opts +from swinir_model_arch import SwinIR as net +from swinir_model_arch_v2 import Swin2SR as net2 +from modules.upscaler import Upscaler, UpscalerData + + +device_swinir = devices.get_device_for('swinir') + + +class UpscalerSwinIR(Upscaler): + def __init__(self, dirname): + self.name = "SwinIR" + self.model_url = "https://github.com/JingyunLiang/SwinIR/releases/download/v0.0" \ + "/003_realSR_BSRGAN_DFOWMFC_s64w8_SwinIR" \ + "-L_x4_GAN.pth " + self.model_name = "SwinIR 4x" + self.user_path = dirname + super().__init__() + scalers = [] + model_files = self.find_models(ext_filter=[".pt", ".pth"]) + for model in model_files: + if "http" in model: + name = self.model_name + else: + name = modelloader.friendly_name(model) + model_data = UpscalerData(name, model, self) + scalers.append(model_data) + self.scalers = scalers + + def do_upscale(self, img, model_file): + model = self.load_model(model_file) + if model is None: + return img + model = model.to(device_swinir, dtype=devices.dtype) + img = upscale(img, model) + try: + torch.cuda.empty_cache() + except: + pass + return img + + def load_model(self, path, scale=4): + if "http" in path: + dl_name = "%s%s" % (self.model_name.replace(" ", "_"), ".pth") + filename = load_file_from_url(url=path, model_dir=self.model_path, file_name=dl_name, progress=True) + else: + filename = path + if filename is None or not os.path.exists(filename): + return None + if filename.endswith(".v2.pth"): + model = net2( + upscale=scale, + in_chans=3, + img_size=64, + window_size=8, + img_range=1.0, + depths=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="nearest+conv", + resi_connection="1conv", + ) + params = None + else: + model = net( + upscale=scale, + in_chans=3, + img_size=64, + window_size=8, + img_range=1.0, + depths=[6, 6, 6, 6, 6, 6, 6, 6, 6], + embed_dim=240, + num_heads=[8, 8, 8, 8, 8, 8, 8, 8, 8], + mlp_ratio=2, + upsampler="nearest+conv", + resi_connection="3conv", + ) + params = "params_ema" + + pretrained_model = torch.load(filename) + if params is not None: + model.load_state_dict(pretrained_model[params], strict=True) + else: + model.load_state_dict(pretrained_model, strict=True) + return model + + +def upscale( + img, + model, + tile=None, + tile_overlap=None, + window_size=8, + scale=4, +): + tile = tile or opts.SWIN_tile + tile_overlap = tile_overlap or opts.SWIN_tile_overlap + + + img = np.array(img) + img = img[:, :, ::-1] + img = np.moveaxis(img, 2, 0) / 255 + img = torch.from_numpy(img).float() + img = img.unsqueeze(0).to(device_swinir, dtype=devices.dtype) + with torch.no_grad(), devices.autocast(): + _, _, h_old, w_old = img.size() + h_pad = (h_old // window_size + 1) * window_size - h_old + w_pad = (w_old // window_size + 1) * window_size - w_old + img = torch.cat([img, torch.flip(img, [2])], 2)[:, :, : h_old + h_pad, :] + img = torch.cat([img, torch.flip(img, [3])], 3)[:, :, :, : w_old + w_pad] + output = inference(img, model, tile, tile_overlap, window_size, scale) + output = output[..., : h_old * scale, : w_old * scale] + output = output.data.squeeze().float().cpu().clamp_(0, 1).numpy() + if output.ndim == 3: + output = np.transpose( + output[[2, 1, 0], :, :], (1, 2, 0) + ) # CHW-RGB to HCW-BGR + output = (output * 255.0).round().astype(np.uint8) # float32 to uint8 + return Image.fromarray(output, "RGB") + + +def inference(img, model, tile, tile_overlap, window_size, scale): + # test the image tile by tile + b, c, h, w = img.size() + tile = min(tile, h, w) + assert tile % window_size == 0, "tile size should be a multiple of window_size" + sf = scale + + stride = tile - tile_overlap + h_idx_list = list(range(0, h - tile, stride)) + [h - tile] + w_idx_list = list(range(0, w - tile, stride)) + [w - tile] + E = torch.zeros(b, c, h * sf, w * sf, dtype=devices.dtype, device=device_swinir).type_as(img) + W = torch.zeros_like(E, dtype=devices.dtype, device=device_swinir) + + with tqdm(total=len(h_idx_list) * len(w_idx_list), desc="SwinIR tiles") as pbar: + for h_idx in h_idx_list: + for w_idx in w_idx_list: + in_patch = img[..., h_idx: h_idx + tile, w_idx: w_idx + tile] + out_patch = model(in_patch) + out_patch_mask = torch.ones_like(out_patch) + + E[ + ..., h_idx * sf: (h_idx + tile) * sf, w_idx * sf: (w_idx + tile) * sf + ].add_(out_patch) + W[ + ..., h_idx * sf: (h_idx + tile) * sf, w_idx * sf: (w_idx + tile) * sf + ].add_(out_patch_mask) + pbar.update(1) + output = E.div_(W) + + return output + + +def on_ui_settings(): + import gradio as gr + + shared.opts.add_option("SWIN_tile", shared.OptionInfo(192, "Tile size for all SwinIR.", gr.Slider, {"minimum": 16, "maximum": 512, "step": 16}, section=('upscaling', "Upscaling"))) + shared.opts.add_option("SWIN_tile_overlap", shared.OptionInfo(8, "Tile overlap, in pixels for SwinIR. Low values = visible seam.", gr.Slider, {"minimum": 0, "maximum": 48, "step": 1}, section=('upscaling', "Upscaling"))) + + +script_callbacks.on_ui_settings(on_ui_settings) diff --git a/extensions-builtin/SwinIR/swinir_model_arch.py b/extensions-builtin/SwinIR/swinir_model_arch.py new file mode 100644 index 00000000..863f42db --- /dev/null +++ b/extensions-builtin/SwinIR/swinir_model_arch.py @@ -0,0 +1,867 @@ +# ----------------------------------------------------------------------------------- +# SwinIR: Image Restoration Using Swin Transformer, https://arxiv.org/abs/2108.10257 +# Originally Written by Ze Liu, Modified by Jingyun Liang. +# ----------------------------------------------------------------------------------- + +import math +import torch +import torch.nn as nn +import torch.nn.functional as F +import torch.utils.checkpoint as checkpoint +from timm.models.layers import DropPath, to_2tuple, trunc_normal_ + + +class Mlp(nn.Module): + def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.): + super().__init__() + out_features = out_features or in_features + hidden_features = hidden_features or in_features + self.fc1 = nn.Linear(in_features, hidden_features) + self.act = act_layer() + self.fc2 = nn.Linear(hidden_features, out_features) + self.drop = nn.Dropout(drop) + + def forward(self, x): + x = self.fc1(x) + x = self.act(x) + x = self.drop(x) + x = self.fc2(x) + x = self.drop(x) + return x + + +def window_partition(x, window_size): + """ + Args: + x: (B, H, W, C) + window_size (int): window size + + Returns: + windows: (num_windows*B, window_size, window_size, C) + """ + B, H, W, C = x.shape + x = x.view(B, H // window_size, window_size, W // window_size, window_size, C) + windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C) + return windows + + +def window_reverse(windows, window_size, H, W): + """ + Args: + windows: (num_windows*B, window_size, window_size, C) + window_size (int): Window size + H (int): Height of image + W (int): Width of image + + Returns: + x: (B, H, W, C) + """ + B = int(windows.shape[0] / (H * W / window_size / window_size)) + x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1) + x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1) + return x + + +class WindowAttention(nn.Module): + r""" Window based multi-head self attention (W-MSA) module with relative position bias. + It supports both of shifted and non-shifted window. + + Args: + dim (int): Number of input channels. + window_size (tuple[int]): The height and width of the window. + num_heads (int): Number of attention heads. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set + attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0 + proj_drop (float, optional): Dropout ratio of output. Default: 0.0 + """ + + def __init__(self, dim, window_size, num_heads, qkv_bias=True, qk_scale=None, attn_drop=0., proj_drop=0.): + + super().__init__() + self.dim = dim + self.window_size = window_size # Wh, Ww + self.num_heads = num_heads + head_dim = dim // num_heads + self.scale = qk_scale or head_dim ** -0.5 + + # define a parameter table of relative position bias + self.relative_position_bias_table = nn.Parameter( + torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads)) # 2*Wh-1 * 2*Ww-1, nH + + # get pair-wise relative position index for each token inside the window + coords_h = torch.arange(self.window_size[0]) + coords_w = torch.arange(self.window_size[1]) + coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww + coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww + relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # 2, Wh*Ww, Wh*Ww + relative_coords = relative_coords.permute(1, 2, 0).contiguous() # Wh*Ww, Wh*Ww, 2 + relative_coords[:, :, 0] += self.window_size[0] - 1 # shift to start from 0 + relative_coords[:, :, 1] += self.window_size[1] - 1 + relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1 + relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww + self.register_buffer("relative_position_index", relative_position_index) + + self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) + self.attn_drop = nn.Dropout(attn_drop) + self.proj = nn.Linear(dim, dim) + + self.proj_drop = nn.Dropout(proj_drop) + + trunc_normal_(self.relative_position_bias_table, std=.02) + self.softmax = nn.Softmax(dim=-1) + + def forward(self, x, mask=None): + """ + Args: + x: input features with shape of (num_windows*B, N, C) + mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None + """ + B_, N, C = x.shape + qkv = self.qkv(x).reshape(B_, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) + q, k, v = qkv[0], qkv[1], qkv[2] # make torchscript happy (cannot use tensor as tuple) + + q = q * self.scale + attn = (q @ k.transpose(-2, -1)) + + relative_position_bias = self.relative_position_bias_table[self.relative_position_index.view(-1)].view( + self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1) # Wh*Ww,Wh*Ww,nH + relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # nH, Wh*Ww, Wh*Ww + attn = attn + relative_position_bias.unsqueeze(0) + + if mask is not None: + nW = mask.shape[0] + attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0) + attn = attn.view(-1, self.num_heads, N, N) + attn = self.softmax(attn) + else: + attn = self.softmax(attn) + + attn = self.attn_drop(attn) + + x = (attn @ v).transpose(1, 2).reshape(B_, N, C) + x = self.proj(x) + x = self.proj_drop(x) + return x + + def extra_repr(self) -> str: + return f'dim={self.dim}, window_size={self.window_size}, num_heads={self.num_heads}' + + def flops(self, N): + # calculate flops for 1 window with token length of N + flops = 0 + # qkv = self.qkv(x) + flops += N * self.dim * 3 * self.dim + # attn = (q @ k.transpose(-2, -1)) + flops += self.num_heads * N * (self.dim // self.num_heads) * N + # x = (attn @ v) + flops += self.num_heads * N * N * (self.dim // self.num_heads) + # x = self.proj(x) + flops += N * self.dim * self.dim + return flops + + +class SwinTransformerBlock(nn.Module): + r""" Swin Transformer Block. + + Args: + dim (int): Number of input channels. + input_resolution (tuple[int]): Input resolution. + num_heads (int): Number of attention heads. + window_size (int): Window size. + shift_size (int): Shift size for SW-MSA. + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. + drop (float, optional): Dropout rate. Default: 0.0 + attn_drop (float, optional): Attention dropout rate. Default: 0.0 + drop_path (float, optional): Stochastic depth rate. Default: 0.0 + act_layer (nn.Module, optional): Activation layer. Default: nn.GELU + norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm + """ + + def __init__(self, dim, input_resolution, num_heads, window_size=7, shift_size=0, + mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., drop_path=0., + act_layer=nn.GELU, norm_layer=nn.LayerNorm): + super().__init__() + self.dim = dim + self.input_resolution = input_resolution + self.num_heads = num_heads + self.window_size = window_size + self.shift_size = shift_size + self.mlp_ratio = mlp_ratio + if min(self.input_resolution) <= self.window_size: + # if window size is larger than input resolution, we don't partition windows + self.shift_size = 0 + self.window_size = min(self.input_resolution) + assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size" + + self.norm1 = norm_layer(dim) + self.attn = WindowAttention( + dim, window_size=to_2tuple(self.window_size), num_heads=num_heads, + qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop) + + self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() + self.norm2 = norm_layer(dim) + mlp_hidden_dim = int(dim * mlp_ratio) + self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) + + if self.shift_size > 0: + attn_mask = self.calculate_mask(self.input_resolution) + else: + attn_mask = None + + self.register_buffer("attn_mask", attn_mask) + + def calculate_mask(self, x_size): + # calculate attention mask for SW-MSA + H, W = x_size + img_mask = torch.zeros((1, H, W, 1)) # 1 H W 1 + h_slices = (slice(0, -self.window_size), + slice(-self.window_size, -self.shift_size), + slice(-self.shift_size, None)) + w_slices = (slice(0, -self.window_size), + slice(-self.window_size, -self.shift_size), + slice(-self.shift_size, None)) + cnt = 0 + for h in h_slices: + for w in w_slices: + img_mask[:, h, w, :] = cnt + cnt += 1 + + mask_windows = window_partition(img_mask, self.window_size) # nW, window_size, window_size, 1 + mask_windows = mask_windows.view(-1, self.window_size * self.window_size) + attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) + attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0)) + + return attn_mask + + def forward(self, x, x_size): + H, W = x_size + B, L, C = x.shape + # assert L == H * W, "input feature has wrong size" + + shortcut = x + x = self.norm1(x) + x = x.view(B, H, W, C) + + # cyclic shift + if self.shift_size > 0: + shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) + else: + shifted_x = x + + # partition windows + x_windows = window_partition(shifted_x, self.window_size) # nW*B, window_size, window_size, C + x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # nW*B, window_size*window_size, C + + # W-MSA/SW-MSA (to be compatible for testing on images whose shapes are the multiple of window size + if self.input_resolution == x_size: + attn_windows = self.attn(x_windows, mask=self.attn_mask) # nW*B, window_size*window_size, C + else: + attn_windows = self.attn(x_windows, mask=self.calculate_mask(x_size).to(x.device)) + + # merge windows + attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) + shifted_x = window_reverse(attn_windows, self.window_size, H, W) # B H' W' C + + # reverse cyclic shift + if self.shift_size > 0: + x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)) + else: + x = shifted_x + x = x.view(B, H * W, C) + + # FFN + x = shortcut + self.drop_path(x) + x = x + self.drop_path(self.mlp(self.norm2(x))) + + return x + + def extra_repr(self) -> str: + return f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, " \ + f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}" + + def flops(self): + flops = 0 + H, W = self.input_resolution + # norm1 + flops += self.dim * H * W + # W-MSA/SW-MSA + nW = H * W / self.window_size / self.window_size + flops += nW * self.attn.flops(self.window_size * self.window_size) + # mlp + flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio + # norm2 + flops += self.dim * H * W + return flops + + +class PatchMerging(nn.Module): + r""" Patch Merging Layer. + + Args: + input_resolution (tuple[int]): Resolution of input feature. + dim (int): Number of input channels. + norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm + """ + + def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm): + super().__init__() + self.input_resolution = input_resolution + self.dim = dim + self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False) + self.norm = norm_layer(4 * dim) + + def forward(self, x): + """ + x: B, H*W, C + """ + H, W = self.input_resolution + B, L, C = x.shape + assert L == H * W, "input feature has wrong size" + assert H % 2 == 0 and W % 2 == 0, f"x size ({H}*{W}) are not even." + + x = x.view(B, H, W, C) + + x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C + x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C + x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C + x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C + x = torch.cat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C + x = x.view(B, -1, 4 * C) # B H/2*W/2 4*C + + x = self.norm(x) + x = self.reduction(x) + + return x + + def extra_repr(self) -> str: + return f"input_resolution={self.input_resolution}, dim={self.dim}" + + def flops(self): + H, W = self.input_resolution + flops = H * W * self.dim + flops += (H // 2) * (W // 2) * 4 * self.dim * 2 * self.dim + return flops + + +class BasicLayer(nn.Module): + """ A basic Swin Transformer layer for one stage. + + Args: + dim (int): Number of input channels. + input_resolution (tuple[int]): Input resolution. + depth (int): Number of blocks. + num_heads (int): Number of attention heads. + window_size (int): Local window size. + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. + drop (float, optional): Dropout rate. Default: 0.0 + attn_drop (float, optional): Attention dropout rate. Default: 0.0 + drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 + norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm + downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None + use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. + """ + + def __init__(self, dim, input_resolution, depth, num_heads, window_size, + mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., + drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False): + + super().__init__() + self.dim = dim + self.input_resolution = input_resolution + self.depth = depth + self.use_checkpoint = use_checkpoint + + # build blocks + self.blocks = nn.ModuleList([ + SwinTransformerBlock(dim=dim, input_resolution=input_resolution, + num_heads=num_heads, window_size=window_size, + shift_size=0 if (i % 2 == 0) else window_size // 2, + mlp_ratio=mlp_ratio, + qkv_bias=qkv_bias, qk_scale=qk_scale, + drop=drop, attn_drop=attn_drop, + drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path, + norm_layer=norm_layer) + for i in range(depth)]) + + # patch merging layer + if downsample is not None: + self.downsample = downsample(input_resolution, dim=dim, norm_layer=norm_layer) + else: + self.downsample = None + + def forward(self, x, x_size): + for blk in self.blocks: + if self.use_checkpoint: + x = checkpoint.checkpoint(blk, x, x_size) + else: + x = blk(x, x_size) + if self.downsample is not None: + x = self.downsample(x) + return x + + def extra_repr(self) -> str: + return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}" + + def flops(self): + flops = 0 + for blk in self.blocks: + flops += blk.flops() + if self.downsample is not None: + flops += self.downsample.flops() + return flops + + +class RSTB(nn.Module): + """Residual Swin Transformer Block (RSTB). + + Args: + dim (int): Number of input channels. + input_resolution (tuple[int]): Input resolution. + depth (int): Number of blocks. + num_heads (int): Number of attention heads. + window_size (int): Local window size. + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. + drop (float, optional): Dropout rate. Default: 0.0 + attn_drop (float, optional): Attention dropout rate. Default: 0.0 + drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 + norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm + downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None + use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. + img_size: Input image size. + patch_size: Patch size. + resi_connection: The convolutional block before residual connection. + """ + + def __init__(self, dim, input_resolution, depth, num_heads, window_size, + mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., + drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False, + img_size=224, patch_size=4, resi_connection='1conv'): + super(RSTB, self).__init__() + + self.dim = dim + self.input_resolution = input_resolution + + self.residual_group = BasicLayer(dim=dim, + input_resolution=input_resolution, + depth=depth, + num_heads=num_heads, + window_size=window_size, + mlp_ratio=mlp_ratio, + qkv_bias=qkv_bias, qk_scale=qk_scale, + drop=drop, attn_drop=attn_drop, + drop_path=drop_path, + norm_layer=norm_layer, + downsample=downsample, + use_checkpoint=use_checkpoint) + + if resi_connection == '1conv': + self.conv = nn.Conv2d(dim, dim, 3, 1, 1) + elif resi_connection == '3conv': + # to save parameters and memory + self.conv = nn.Sequential(nn.Conv2d(dim, dim // 4, 3, 1, 1), nn.LeakyReLU(negative_slope=0.2, inplace=True), + nn.Conv2d(dim // 4, dim // 4, 1, 1, 0), + nn.LeakyReLU(negative_slope=0.2, inplace=True), + nn.Conv2d(dim // 4, dim, 3, 1, 1)) + + self.patch_embed = PatchEmbed( + img_size=img_size, patch_size=patch_size, in_chans=0, embed_dim=dim, + norm_layer=None) + + self.patch_unembed = PatchUnEmbed( + img_size=img_size, patch_size=patch_size, in_chans=0, embed_dim=dim, + norm_layer=None) + + def forward(self, x, x_size): + return self.patch_embed(self.conv(self.patch_unembed(self.residual_group(x, x_size), x_size))) + x + + def flops(self): + flops = 0 + flops += self.residual_group.flops() + H, W = self.input_resolution + flops += H * W * self.dim * self.dim * 9 + flops += self.patch_embed.flops() + flops += self.patch_unembed.flops() + + return flops + + +class PatchEmbed(nn.Module): + r""" Image to Patch Embedding + + Args: + img_size (int): Image size. Default: 224. + patch_size (int): Patch token size. Default: 4. + in_chans (int): Number of input image channels. Default: 3. + embed_dim (int): Number of linear projection output channels. Default: 96. + norm_layer (nn.Module, optional): Normalization layer. Default: None + """ + + def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): + super().__init__() + img_size = to_2tuple(img_size) + patch_size = to_2tuple(patch_size) + patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]] + self.img_size = img_size + self.patch_size = patch_size + self.patches_resolution = patches_resolution + self.num_patches = patches_resolution[0] * patches_resolution[1] + + self.in_chans = in_chans + self.embed_dim = embed_dim + + if norm_layer is not None: + self.norm = norm_layer(embed_dim) + else: + self.norm = None + + def forward(self, x): + x = x.flatten(2).transpose(1, 2) # B Ph*Pw C + if self.norm is not None: + x = self.norm(x) + return x + + def flops(self): + flops = 0 + H, W = self.img_size + if self.norm is not None: + flops += H * W * self.embed_dim + return flops + + +class PatchUnEmbed(nn.Module): + r""" Image to Patch Unembedding + + Args: + img_size (int): Image size. Default: 224. + patch_size (int): Patch token size. Default: 4. + in_chans (int): Number of input image channels. Default: 3. + embed_dim (int): Number of linear projection output channels. Default: 96. + norm_layer (nn.Module, optional): Normalization layer. Default: None + """ + + def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): + super().__init__() + img_size = to_2tuple(img_size) + patch_size = to_2tuple(patch_size) + patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]] + self.img_size = img_size + self.patch_size = patch_size + self.patches_resolution = patches_resolution + self.num_patches = patches_resolution[0] * patches_resolution[1] + + self.in_chans = in_chans + self.embed_dim = embed_dim + + def forward(self, x, x_size): + B, HW, C = x.shape + x = x.transpose(1, 2).view(B, self.embed_dim, x_size[0], x_size[1]) # B Ph*Pw C + return x + + def flops(self): + flops = 0 + return flops + + +class Upsample(nn.Sequential): + """Upsample module. + + Args: + scale (int): Scale factor. Supported scales: 2^n and 3. + num_feat (int): Channel number of intermediate features. + """ + + def __init__(self, scale, num_feat): + m = [] + if (scale & (scale - 1)) == 0: # scale = 2^n + for _ in range(int(math.log(scale, 2))): + m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1)) + m.append(nn.PixelShuffle(2)) + elif scale == 3: + m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1)) + m.append(nn.PixelShuffle(3)) + else: + raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.') + super(Upsample, self).__init__(*m) + + +class UpsampleOneStep(nn.Sequential): + """UpsampleOneStep module (the difference with Upsample is that it always only has 1conv + 1pixelshuffle) + Used in lightweight SR to save parameters. + + Args: + scale (int): Scale factor. Supported scales: 2^n and 3. + num_feat (int): Channel number of intermediate features. + + """ + + def __init__(self, scale, num_feat, num_out_ch, input_resolution=None): + self.num_feat = num_feat + self.input_resolution = input_resolution + m = [] + m.append(nn.Conv2d(num_feat, (scale ** 2) * num_out_ch, 3, 1, 1)) + m.append(nn.PixelShuffle(scale)) + super(UpsampleOneStep, self).__init__(*m) + + def flops(self): + H, W = self.input_resolution + flops = H * W * self.num_feat * 3 * 9 + return flops + + +class SwinIR(nn.Module): + r""" SwinIR + A PyTorch impl of : `SwinIR: Image Restoration Using Swin Transformer`, based on Swin Transformer. + + Args: + img_size (int | tuple(int)): Input image size. Default 64 + patch_size (int | tuple(int)): Patch size. Default: 1 + in_chans (int): Number of input image channels. Default: 3 + embed_dim (int): Patch embedding dimension. Default: 96 + depths (tuple(int)): Depth of each Swin Transformer layer. + num_heads (tuple(int)): Number of attention heads in different layers. + window_size (int): Window size. Default: 7 + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4 + qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float): Override default qk scale of head_dim ** -0.5 if set. Default: None + drop_rate (float): Dropout rate. Default: 0 + attn_drop_rate (float): Attention dropout rate. Default: 0 + drop_path_rate (float): Stochastic depth rate. Default: 0.1 + norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm. + ape (bool): If True, add absolute position embedding to the patch embedding. Default: False + patch_norm (bool): If True, add normalization after patch embedding. Default: True + use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False + upscale: Upscale factor. 2/3/4/8 for image SR, 1 for denoising and compress artifact reduction + img_range: Image range. 1. or 255. + upsampler: The reconstruction reconstruction module. 'pixelshuffle'/'pixelshuffledirect'/'nearest+conv'/None + resi_connection: The convolutional block before residual connection. '1conv'/'3conv' + """ + + def __init__(self, img_size=64, patch_size=1, in_chans=3, + embed_dim=96, depths=[6, 6, 6, 6], num_heads=[6, 6, 6, 6], + window_size=7, mlp_ratio=4., qkv_bias=True, qk_scale=None, + drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1, + norm_layer=nn.LayerNorm, ape=False, patch_norm=True, + use_checkpoint=False, upscale=2, img_range=1., upsampler='', resi_connection='1conv', + **kwargs): + super(SwinIR, self).__init__() + num_in_ch = in_chans + num_out_ch = in_chans + num_feat = 64 + self.img_range = img_range + if in_chans == 3: + rgb_mean = (0.4488, 0.4371, 0.4040) + self.mean = torch.Tensor(rgb_mean).view(1, 3, 1, 1) + else: + self.mean = torch.zeros(1, 1, 1, 1) + self.upscale = upscale + self.upsampler = upsampler + self.window_size = window_size + + ##################################################################################################### + ################################### 1, shallow feature extraction ################################### + self.conv_first = nn.Conv2d(num_in_ch, embed_dim, 3, 1, 1) + + ##################################################################################################### + ################################### 2, deep feature extraction ###################################### + self.num_layers = len(depths) + self.embed_dim = embed_dim + self.ape = ape + self.patch_norm = patch_norm + self.num_features = embed_dim + self.mlp_ratio = mlp_ratio + + # split image into non-overlapping patches + self.patch_embed = PatchEmbed( + img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim, + norm_layer=norm_layer if self.patch_norm else None) + num_patches = self.patch_embed.num_patches + patches_resolution = self.patch_embed.patches_resolution + self.patches_resolution = patches_resolution + + # merge non-overlapping patches into image + self.patch_unembed = PatchUnEmbed( + img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim, + norm_layer=norm_layer if self.patch_norm else None) + + # absolute position embedding + if self.ape: + self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim)) + trunc_normal_(self.absolute_pos_embed, std=.02) + + self.pos_drop = nn.Dropout(p=drop_rate) + + # stochastic depth + dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule + + # build Residual Swin Transformer blocks (RSTB) + self.layers = nn.ModuleList() + for i_layer in range(self.num_layers): + layer = RSTB(dim=embed_dim, + input_resolution=(patches_resolution[0], + patches_resolution[1]), + depth=depths[i_layer], + num_heads=num_heads[i_layer], + window_size=window_size, + mlp_ratio=self.mlp_ratio, + qkv_bias=qkv_bias, qk_scale=qk_scale, + drop=drop_rate, attn_drop=attn_drop_rate, + drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # no impact on SR results + norm_layer=norm_layer, + downsample=None, + use_checkpoint=use_checkpoint, + img_size=img_size, + patch_size=patch_size, + resi_connection=resi_connection + + ) + self.layers.append(layer) + self.norm = norm_layer(self.num_features) + + # build the last conv layer in deep feature extraction + if resi_connection == '1conv': + self.conv_after_body = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1) + elif resi_connection == '3conv': + # to save parameters and memory + self.conv_after_body = nn.Sequential(nn.Conv2d(embed_dim, embed_dim // 4, 3, 1, 1), + nn.LeakyReLU(negative_slope=0.2, inplace=True), + nn.Conv2d(embed_dim // 4, embed_dim // 4, 1, 1, 0), + nn.LeakyReLU(negative_slope=0.2, inplace=True), + nn.Conv2d(embed_dim // 4, embed_dim, 3, 1, 1)) + + ##################################################################################################### + ################################ 3, high quality image reconstruction ################################ + if self.upsampler == 'pixelshuffle': + # for classical SR + self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), + nn.LeakyReLU(inplace=True)) + self.upsample = Upsample(upscale, num_feat) + self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) + elif self.upsampler == 'pixelshuffledirect': + # for lightweight SR (to save parameters) + self.upsample = UpsampleOneStep(upscale, embed_dim, num_out_ch, + (patches_resolution[0], patches_resolution[1])) + elif self.upsampler == 'nearest+conv': + # for real-world SR (less artifacts) + self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), + nn.LeakyReLU(inplace=True)) + self.conv_up1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1) + if self.upscale == 4: + self.conv_up2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1) + self.conv_hr = nn.Conv2d(num_feat, num_feat, 3, 1, 1) + self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) + self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True) + else: + # for image denoising and JPEG compression artifact reduction + self.conv_last = nn.Conv2d(embed_dim, num_out_ch, 3, 1, 1) + + self.apply(self._init_weights) + + def _init_weights(self, m): + if isinstance(m, nn.Linear): + trunc_normal_(m.weight, std=.02) + if isinstance(m, nn.Linear) and m.bias is not None: + nn.init.constant_(m.bias, 0) + elif isinstance(m, nn.LayerNorm): + nn.init.constant_(m.bias, 0) + nn.init.constant_(m.weight, 1.0) + + @torch.jit.ignore + def no_weight_decay(self): + return {'absolute_pos_embed'} + + @torch.jit.ignore + def no_weight_decay_keywords(self): + return {'relative_position_bias_table'} + + def check_image_size(self, x): + _, _, h, w = x.size() + mod_pad_h = (self.window_size - h % self.window_size) % self.window_size + mod_pad_w = (self.window_size - w % self.window_size) % self.window_size + x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), 'reflect') + return x + + def forward_features(self, x): + x_size = (x.shape[2], x.shape[3]) + x = self.patch_embed(x) + if self.ape: + x = x + self.absolute_pos_embed + x = self.pos_drop(x) + + for layer in self.layers: + x = layer(x, x_size) + + x = self.norm(x) # B L C + x = self.patch_unembed(x, x_size) + + return x + + def forward(self, x): + H, W = x.shape[2:] + x = self.check_image_size(x) + + self.mean = self.mean.type_as(x) + x = (x - self.mean) * self.img_range + + if self.upsampler == 'pixelshuffle': + # for classical SR + x = self.conv_first(x) + x = self.conv_after_body(self.forward_features(x)) + x + x = self.conv_before_upsample(x) + x = self.conv_last(self.upsample(x)) + elif self.upsampler == 'pixelshuffledirect': + # for lightweight SR + x = self.conv_first(x) + x = self.conv_after_body(self.forward_features(x)) + x + x = self.upsample(x) + elif self.upsampler == 'nearest+conv': + # for real-world SR + x = self.conv_first(x) + x = self.conv_after_body(self.forward_features(x)) + x + x = self.conv_before_upsample(x) + x = self.lrelu(self.conv_up1(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest'))) + if self.upscale == 4: + x = self.lrelu(self.conv_up2(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest'))) + x = self.conv_last(self.lrelu(self.conv_hr(x))) + else: + # for image denoising and JPEG compression artifact reduction + x_first = self.conv_first(x) + res = self.conv_after_body(self.forward_features(x_first)) + x_first + x = x + self.conv_last(res) + + x = x / self.img_range + self.mean + + return x[:, :, :H*self.upscale, :W*self.upscale] + + def flops(self): + flops = 0 + H, W = self.patches_resolution + flops += H * W * 3 * self.embed_dim * 9 + flops += self.patch_embed.flops() + for i, layer in enumerate(self.layers): + flops += layer.flops() + flops += H * W * 3 * self.embed_dim * self.embed_dim + flops += self.upsample.flops() + return flops + + +if __name__ == '__main__': + upscale = 4 + window_size = 8 + height = (1024 // upscale // window_size + 1) * window_size + width = (720 // upscale // window_size + 1) * window_size + model = SwinIR(upscale=2, img_size=(height, width), + window_size=window_size, img_range=1., depths=[6, 6, 6, 6], + embed_dim=60, num_heads=[6, 6, 6, 6], mlp_ratio=2, upsampler='pixelshuffledirect') + print(model) + print(height, width, model.flops() / 1e9) + + x = torch.randn((1, 3, height, width)) + x = model(x) + print(x.shape) diff --git a/extensions-builtin/SwinIR/swinir_model_arch_v2.py b/extensions-builtin/SwinIR/swinir_model_arch_v2.py new file mode 100644 index 00000000..0e28ae6e --- /dev/null +++ b/extensions-builtin/SwinIR/swinir_model_arch_v2.py @@ -0,0 +1,1017 @@ +# -----------------------------------------------------------------------------------
+# Swin2SR: Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration, https://arxiv.org/abs/
+# Written by Conde and Choi et al.
+# -----------------------------------------------------------------------------------
+
+import math
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.utils.checkpoint as checkpoint
+from timm.models.layers import DropPath, to_2tuple, trunc_normal_
+
+
+class Mlp(nn.Module):
+ def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):
+ super().__init__()
+ out_features = out_features or in_features
+ hidden_features = hidden_features or in_features
+ self.fc1 = nn.Linear(in_features, hidden_features)
+ self.act = act_layer()
+ self.fc2 = nn.Linear(hidden_features, out_features)
+ self.drop = nn.Dropout(drop)
+
+ def forward(self, x):
+ x = self.fc1(x)
+ x = self.act(x)
+ x = self.drop(x)
+ x = self.fc2(x)
+ x = self.drop(x)
+ return x
+
+
+def window_partition(x, window_size):
+ """
+ Args:
+ x: (B, H, W, C)
+ window_size (int): window size
+ Returns:
+ windows: (num_windows*B, window_size, window_size, C)
+ """
+ B, H, W, C = x.shape
+ x = x.view(B, H // window_size, window_size, W // window_size, window_size, C)
+ windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
+ return windows
+
+
+def window_reverse(windows, window_size, H, W):
+ """
+ Args:
+ windows: (num_windows*B, window_size, window_size, C)
+ window_size (int): Window size
+ H (int): Height of image
+ W (int): Width of image
+ Returns:
+ x: (B, H, W, C)
+ """
+ B = int(windows.shape[0] / (H * W / window_size / window_size))
+ x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)
+ x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
+ return x
+
+class WindowAttention(nn.Module):
+ r""" Window based multi-head self attention (W-MSA) module with relative position bias.
+ It supports both of shifted and non-shifted window.
+ Args:
+ dim (int): Number of input channels.
+ window_size (tuple[int]): The height and width of the window.
+ num_heads (int): Number of attention heads.
+ qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+ attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0
+ proj_drop (float, optional): Dropout ratio of output. Default: 0.0
+ pretrained_window_size (tuple[int]): The height and width of the window in pre-training.
+ """
+
+ def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0.,
+ pretrained_window_size=[0, 0]):
+
+ super().__init__()
+ self.dim = dim
+ self.window_size = window_size # Wh, Ww
+ self.pretrained_window_size = pretrained_window_size
+ self.num_heads = num_heads
+
+ self.logit_scale = nn.Parameter(torch.log(10 * torch.ones((num_heads, 1, 1))), requires_grad=True)
+
+ # mlp to generate continuous relative position bias
+ self.cpb_mlp = nn.Sequential(nn.Linear(2, 512, bias=True),
+ nn.ReLU(inplace=True),
+ nn.Linear(512, num_heads, bias=False))
+
+ # get relative_coords_table
+ relative_coords_h = torch.arange(-(self.window_size[0] - 1), self.window_size[0], dtype=torch.float32)
+ relative_coords_w = torch.arange(-(self.window_size[1] - 1), self.window_size[1], dtype=torch.float32)
+ relative_coords_table = torch.stack(
+ torch.meshgrid([relative_coords_h,
+ relative_coords_w])).permute(1, 2, 0).contiguous().unsqueeze(0) # 1, 2*Wh-1, 2*Ww-1, 2
+ if pretrained_window_size[0] > 0:
+ relative_coords_table[:, :, :, 0] /= (pretrained_window_size[0] - 1)
+ relative_coords_table[:, :, :, 1] /= (pretrained_window_size[1] - 1)
+ else:
+ relative_coords_table[:, :, :, 0] /= (self.window_size[0] - 1)
+ relative_coords_table[:, :, :, 1] /= (self.window_size[1] - 1)
+ relative_coords_table *= 8 # normalize to -8, 8
+ relative_coords_table = torch.sign(relative_coords_table) * torch.log2(
+ torch.abs(relative_coords_table) + 1.0) / np.log2(8)
+
+ self.register_buffer("relative_coords_table", relative_coords_table)
+
+ # get pair-wise relative position index for each token inside the window
+ coords_h = torch.arange(self.window_size[0])
+ coords_w = torch.arange(self.window_size[1])
+ coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww
+ coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww
+ relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # 2, Wh*Ww, Wh*Ww
+ relative_coords = relative_coords.permute(1, 2, 0).contiguous() # Wh*Ww, Wh*Ww, 2
+ relative_coords[:, :, 0] += self.window_size[0] - 1 # shift to start from 0
+ relative_coords[:, :, 1] += self.window_size[1] - 1
+ relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
+ relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww
+ self.register_buffer("relative_position_index", relative_position_index)
+
+ self.qkv = nn.Linear(dim, dim * 3, bias=False)
+ if qkv_bias:
+ self.q_bias = nn.Parameter(torch.zeros(dim))
+ self.v_bias = nn.Parameter(torch.zeros(dim))
+ else:
+ self.q_bias = None
+ self.v_bias = None
+ self.attn_drop = nn.Dropout(attn_drop)
+ self.proj = nn.Linear(dim, dim)
+ self.proj_drop = nn.Dropout(proj_drop)
+ self.softmax = nn.Softmax(dim=-1)
+
+ def forward(self, x, mask=None):
+ """
+ Args:
+ x: input features with shape of (num_windows*B, N, C)
+ mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None
+ """
+ B_, N, C = x.shape
+ qkv_bias = None
+ if self.q_bias is not None:
+ qkv_bias = torch.cat((self.q_bias, torch.zeros_like(self.v_bias, requires_grad=False), self.v_bias))
+ qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias)
+ qkv = qkv.reshape(B_, N, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
+ q, k, v = qkv[0], qkv[1], qkv[2] # make torchscript happy (cannot use tensor as tuple)
+
+ # cosine attention
+ attn = (F.normalize(q, dim=-1) @ F.normalize(k, dim=-1).transpose(-2, -1))
+ logit_scale = torch.clamp(self.logit_scale, max=torch.log(torch.tensor(1. / 0.01)).to(self.logit_scale.device)).exp()
+ attn = attn * logit_scale
+
+ relative_position_bias_table = self.cpb_mlp(self.relative_coords_table).view(-1, self.num_heads)
+ relative_position_bias = relative_position_bias_table[self.relative_position_index.view(-1)].view(
+ self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1) # Wh*Ww,Wh*Ww,nH
+ relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # nH, Wh*Ww, Wh*Ww
+ relative_position_bias = 16 * torch.sigmoid(relative_position_bias)
+ attn = attn + relative_position_bias.unsqueeze(0)
+
+ if mask is not None:
+ nW = mask.shape[0]
+ attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)
+ attn = attn.view(-1, self.num_heads, N, N)
+ attn = self.softmax(attn)
+ else:
+ attn = self.softmax(attn)
+
+ attn = self.attn_drop(attn)
+
+ x = (attn @ v).transpose(1, 2).reshape(B_, N, C)
+ x = self.proj(x)
+ x = self.proj_drop(x)
+ return x
+
+ def extra_repr(self) -> str:
+ return f'dim={self.dim}, window_size={self.window_size}, ' \
+ f'pretrained_window_size={self.pretrained_window_size}, num_heads={self.num_heads}'
+
+ def flops(self, N):
+ # calculate flops for 1 window with token length of N
+ flops = 0
+ # qkv = self.qkv(x)
+ flops += N * self.dim * 3 * self.dim
+ # attn = (q @ k.transpose(-2, -1))
+ flops += self.num_heads * N * (self.dim // self.num_heads) * N
+ # x = (attn @ v)
+ flops += self.num_heads * N * N * (self.dim // self.num_heads)
+ # x = self.proj(x)
+ flops += N * self.dim * self.dim
+ return flops
+
+class SwinTransformerBlock(nn.Module):
+ r""" Swin Transformer Block.
+ Args:
+ dim (int): Number of input channels.
+ input_resolution (tuple[int]): Input resulotion.
+ num_heads (int): Number of attention heads.
+ window_size (int): Window size.
+ shift_size (int): Shift size for SW-MSA.
+ mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+ qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+ drop (float, optional): Dropout rate. Default: 0.0
+ attn_drop (float, optional): Attention dropout rate. Default: 0.0
+ drop_path (float, optional): Stochastic depth rate. Default: 0.0
+ act_layer (nn.Module, optional): Activation layer. Default: nn.GELU
+ norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+ pretrained_window_size (int): Window size in pre-training.
+ """
+
+ def __init__(self, dim, input_resolution, num_heads, window_size=7, shift_size=0,
+ mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0.,
+ act_layer=nn.GELU, norm_layer=nn.LayerNorm, pretrained_window_size=0):
+ super().__init__()
+ self.dim = dim
+ self.input_resolution = input_resolution
+ self.num_heads = num_heads
+ self.window_size = window_size
+ self.shift_size = shift_size
+ self.mlp_ratio = mlp_ratio
+ if min(self.input_resolution) <= self.window_size:
+ # if window size is larger than input resolution, we don't partition windows
+ self.shift_size = 0
+ self.window_size = min(self.input_resolution)
+ assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size"
+
+ self.norm1 = norm_layer(dim)
+ self.attn = WindowAttention(
+ dim, window_size=to_2tuple(self.window_size), num_heads=num_heads,
+ qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop,
+ pretrained_window_size=to_2tuple(pretrained_window_size))
+
+ self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+ self.norm2 = norm_layer(dim)
+ mlp_hidden_dim = int(dim * mlp_ratio)
+ self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)
+
+ if self.shift_size > 0:
+ attn_mask = self.calculate_mask(self.input_resolution)
+ else:
+ attn_mask = None
+
+ self.register_buffer("attn_mask", attn_mask)
+
+ def calculate_mask(self, x_size):
+ # calculate attention mask for SW-MSA
+ H, W = x_size
+ img_mask = torch.zeros((1, H, W, 1)) # 1 H W 1
+ h_slices = (slice(0, -self.window_size),
+ slice(-self.window_size, -self.shift_size),
+ slice(-self.shift_size, None))
+ w_slices = (slice(0, -self.window_size),
+ slice(-self.window_size, -self.shift_size),
+ slice(-self.shift_size, None))
+ cnt = 0
+ for h in h_slices:
+ for w in w_slices:
+ img_mask[:, h, w, :] = cnt
+ cnt += 1
+
+ mask_windows = window_partition(img_mask, self.window_size) # nW, window_size, window_size, 1
+ mask_windows = mask_windows.view(-1, self.window_size * self.window_size)
+ attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)
+ attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))
+
+ return attn_mask
+
+ def forward(self, x, x_size):
+ H, W = x_size
+ B, L, C = x.shape
+ #assert L == H * W, "input feature has wrong size"
+
+ shortcut = x
+ x = x.view(B, H, W, C)
+
+ # cyclic shift
+ if self.shift_size > 0:
+ shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))
+ else:
+ shifted_x = x
+
+ # partition windows
+ x_windows = window_partition(shifted_x, self.window_size) # nW*B, window_size, window_size, C
+ x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # nW*B, window_size*window_size, C
+
+ # W-MSA/SW-MSA (to be compatible for testing on images whose shapes are the multiple of window size
+ if self.input_resolution == x_size:
+ attn_windows = self.attn(x_windows, mask=self.attn_mask) # nW*B, window_size*window_size, C
+ else:
+ attn_windows = self.attn(x_windows, mask=self.calculate_mask(x_size).to(x.device))
+
+ # merge windows
+ attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
+ shifted_x = window_reverse(attn_windows, self.window_size, H, W) # B H' W' C
+
+ # reverse cyclic shift
+ if self.shift_size > 0:
+ x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2))
+ else:
+ x = shifted_x
+ x = x.view(B, H * W, C)
+ x = shortcut + self.drop_path(self.norm1(x))
+
+ # FFN
+ x = x + self.drop_path(self.norm2(self.mlp(x)))
+
+ return x
+
+ def extra_repr(self) -> str:
+ return f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, " \
+ f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}"
+
+ def flops(self):
+ flops = 0
+ H, W = self.input_resolution
+ # norm1
+ flops += self.dim * H * W
+ # W-MSA/SW-MSA
+ nW = H * W / self.window_size / self.window_size
+ flops += nW * self.attn.flops(self.window_size * self.window_size)
+ # mlp
+ flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio
+ # norm2
+ flops += self.dim * H * W
+ return flops
+
+class PatchMerging(nn.Module):
+ r""" Patch Merging Layer.
+ Args:
+ input_resolution (tuple[int]): Resolution of input feature.
+ dim (int): Number of input channels.
+ norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+ """
+
+ def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm):
+ super().__init__()
+ self.input_resolution = input_resolution
+ self.dim = dim
+ self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False)
+ self.norm = norm_layer(2 * dim)
+
+ def forward(self, x):
+ """
+ x: B, H*W, C
+ """
+ H, W = self.input_resolution
+ B, L, C = x.shape
+ assert L == H * W, "input feature has wrong size"
+ assert H % 2 == 0 and W % 2 == 0, f"x size ({H}*{W}) are not even."
+
+ x = x.view(B, H, W, C)
+
+ x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C
+ x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C
+ x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C
+ x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C
+ x = torch.cat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C
+ x = x.view(B, -1, 4 * C) # B H/2*W/2 4*C
+
+ x = self.reduction(x)
+ x = self.norm(x)
+
+ return x
+
+ def extra_repr(self) -> str:
+ return f"input_resolution={self.input_resolution}, dim={self.dim}"
+
+ def flops(self):
+ H, W = self.input_resolution
+ flops = (H // 2) * (W // 2) * 4 * self.dim * 2 * self.dim
+ flops += H * W * self.dim // 2
+ return flops
+
+class BasicLayer(nn.Module):
+ """ A basic Swin Transformer layer for one stage.
+ Args:
+ dim (int): Number of input channels.
+ input_resolution (tuple[int]): Input resolution.
+ depth (int): Number of blocks.
+ num_heads (int): Number of attention heads.
+ window_size (int): Local window size.
+ mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+ qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+ drop (float, optional): Dropout rate. Default: 0.0
+ attn_drop (float, optional): Attention dropout rate. Default: 0.0
+ drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
+ norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+ downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
+ use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
+ pretrained_window_size (int): Local window size in pre-training.
+ """
+
+ def __init__(self, dim, input_resolution, depth, num_heads, window_size,
+ mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0.,
+ drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False,
+ pretrained_window_size=0):
+
+ super().__init__()
+ self.dim = dim
+ self.input_resolution = input_resolution
+ self.depth = depth
+ self.use_checkpoint = use_checkpoint
+
+ # build blocks
+ self.blocks = nn.ModuleList([
+ SwinTransformerBlock(dim=dim, input_resolution=input_resolution,
+ num_heads=num_heads, window_size=window_size,
+ shift_size=0 if (i % 2 == 0) else window_size // 2,
+ mlp_ratio=mlp_ratio,
+ qkv_bias=qkv_bias,
+ drop=drop, attn_drop=attn_drop,
+ drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path,
+ norm_layer=norm_layer,
+ pretrained_window_size=pretrained_window_size)
+ for i in range(depth)])
+
+ # patch merging layer
+ if downsample is not None:
+ self.downsample = downsample(input_resolution, dim=dim, norm_layer=norm_layer)
+ else:
+ self.downsample = None
+
+ def forward(self, x, x_size):
+ for blk in self.blocks:
+ if self.use_checkpoint:
+ x = checkpoint.checkpoint(blk, x, x_size)
+ else:
+ x = blk(x, x_size)
+ if self.downsample is not None:
+ x = self.downsample(x)
+ return x
+
+ def extra_repr(self) -> str:
+ return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}"
+
+ def flops(self):
+ flops = 0
+ for blk in self.blocks:
+ flops += blk.flops()
+ if self.downsample is not None:
+ flops += self.downsample.flops()
+ return flops
+
+ def _init_respostnorm(self):
+ for blk in self.blocks:
+ nn.init.constant_(blk.norm1.bias, 0)
+ nn.init.constant_(blk.norm1.weight, 0)
+ nn.init.constant_(blk.norm2.bias, 0)
+ nn.init.constant_(blk.norm2.weight, 0)
+
+class PatchEmbed(nn.Module):
+ r""" Image to Patch Embedding
+ Args:
+ img_size (int): Image size. Default: 224.
+ patch_size (int): Patch token size. Default: 4.
+ in_chans (int): Number of input image channels. Default: 3.
+ embed_dim (int): Number of linear projection output channels. Default: 96.
+ norm_layer (nn.Module, optional): Normalization layer. Default: None
+ """
+
+ def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None):
+ super().__init__()
+ img_size = to_2tuple(img_size)
+ patch_size = to_2tuple(patch_size)
+ patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]]
+ self.img_size = img_size
+ self.patch_size = patch_size
+ self.patches_resolution = patches_resolution
+ self.num_patches = patches_resolution[0] * patches_resolution[1]
+
+ self.in_chans = in_chans
+ self.embed_dim = embed_dim
+
+ self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)
+ if norm_layer is not None:
+ self.norm = norm_layer(embed_dim)
+ else:
+ self.norm = None
+
+ def forward(self, x):
+ B, C, H, W = x.shape
+ # FIXME look at relaxing size constraints
+ # assert H == self.img_size[0] and W == self.img_size[1],
+ # f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})."
+ x = self.proj(x).flatten(2).transpose(1, 2) # B Ph*Pw C
+ if self.norm is not None:
+ x = self.norm(x)
+ return x
+
+ def flops(self):
+ Ho, Wo = self.patches_resolution
+ flops = Ho * Wo * self.embed_dim * self.in_chans * (self.patch_size[0] * self.patch_size[1])
+ if self.norm is not None:
+ flops += Ho * Wo * self.embed_dim
+ return flops
+
+class RSTB(nn.Module):
+ """Residual Swin Transformer Block (RSTB).
+
+ Args:
+ dim (int): Number of input channels.
+ input_resolution (tuple[int]): Input resolution.
+ depth (int): Number of blocks.
+ num_heads (int): Number of attention heads.
+ window_size (int): Local window size.
+ mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+ qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+ drop (float, optional): Dropout rate. Default: 0.0
+ attn_drop (float, optional): Attention dropout rate. Default: 0.0
+ drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
+ norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+ downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
+ use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
+ img_size: Input image size.
+ patch_size: Patch size.
+ resi_connection: The convolutional block before residual connection.
+ """
+
+ def __init__(self, dim, input_resolution, depth, num_heads, window_size,
+ mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0.,
+ drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False,
+ img_size=224, patch_size=4, resi_connection='1conv'):
+ super(RSTB, self).__init__()
+
+ self.dim = dim
+ self.input_resolution = input_resolution
+
+ self.residual_group = BasicLayer(dim=dim,
+ input_resolution=input_resolution,
+ depth=depth,
+ num_heads=num_heads,
+ window_size=window_size,
+ mlp_ratio=mlp_ratio,
+ qkv_bias=qkv_bias,
+ drop=drop, attn_drop=attn_drop,
+ drop_path=drop_path,
+ norm_layer=norm_layer,
+ downsample=downsample,
+ use_checkpoint=use_checkpoint)
+
+ if resi_connection == '1conv':
+ self.conv = nn.Conv2d(dim, dim, 3, 1, 1)
+ elif resi_connection == '3conv':
+ # to save parameters and memory
+ self.conv = nn.Sequential(nn.Conv2d(dim, dim // 4, 3, 1, 1), nn.LeakyReLU(negative_slope=0.2, inplace=True),
+ nn.Conv2d(dim // 4, dim // 4, 1, 1, 0),
+ nn.LeakyReLU(negative_slope=0.2, inplace=True),
+ nn.Conv2d(dim // 4, dim, 3, 1, 1))
+
+ self.patch_embed = PatchEmbed(
+ img_size=img_size, patch_size=patch_size, in_chans=dim, embed_dim=dim,
+ norm_layer=None)
+
+ self.patch_unembed = PatchUnEmbed(
+ img_size=img_size, patch_size=patch_size, in_chans=dim, embed_dim=dim,
+ norm_layer=None)
+
+ def forward(self, x, x_size):
+ return self.patch_embed(self.conv(self.patch_unembed(self.residual_group(x, x_size), x_size))) + x
+
+ def flops(self):
+ flops = 0
+ flops += self.residual_group.flops()
+ H, W = self.input_resolution
+ flops += H * W * self.dim * self.dim * 9
+ flops += self.patch_embed.flops()
+ flops += self.patch_unembed.flops()
+
+ return flops
+
+class PatchUnEmbed(nn.Module):
+ r""" Image to Patch Unembedding
+
+ Args:
+ img_size (int): Image size. Default: 224.
+ patch_size (int): Patch token size. Default: 4.
+ in_chans (int): Number of input image channels. Default: 3.
+ embed_dim (int): Number of linear projection output channels. Default: 96.
+ norm_layer (nn.Module, optional): Normalization layer. Default: None
+ """
+
+ def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None):
+ super().__init__()
+ img_size = to_2tuple(img_size)
+ patch_size = to_2tuple(patch_size)
+ patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]]
+ self.img_size = img_size
+ self.patch_size = patch_size
+ self.patches_resolution = patches_resolution
+ self.num_patches = patches_resolution[0] * patches_resolution[1]
+
+ self.in_chans = in_chans
+ self.embed_dim = embed_dim
+
+ def forward(self, x, x_size):
+ B, HW, C = x.shape
+ x = x.transpose(1, 2).view(B, self.embed_dim, x_size[0], x_size[1]) # B Ph*Pw C
+ return x
+
+ def flops(self):
+ flops = 0
+ return flops
+
+
+class Upsample(nn.Sequential):
+ """Upsample module.
+
+ Args:
+ scale (int): Scale factor. Supported scales: 2^n and 3.
+ num_feat (int): Channel number of intermediate features.
+ """
+
+ def __init__(self, scale, num_feat):
+ m = []
+ if (scale & (scale - 1)) == 0: # scale = 2^n
+ for _ in range(int(math.log(scale, 2))):
+ m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1))
+ m.append(nn.PixelShuffle(2))
+ elif scale == 3:
+ m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1))
+ m.append(nn.PixelShuffle(3))
+ else:
+ raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.')
+ super(Upsample, self).__init__(*m)
+
+class Upsample_hf(nn.Sequential):
+ """Upsample module.
+
+ Args:
+ scale (int): Scale factor. Supported scales: 2^n and 3.
+ num_feat (int): Channel number of intermediate features.
+ """
+
+ def __init__(self, scale, num_feat):
+ m = []
+ if (scale & (scale - 1)) == 0: # scale = 2^n
+ for _ in range(int(math.log(scale, 2))):
+ m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1))
+ m.append(nn.PixelShuffle(2))
+ elif scale == 3:
+ m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1))
+ m.append(nn.PixelShuffle(3))
+ else:
+ raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.')
+ super(Upsample_hf, self).__init__(*m)
+
+
+class UpsampleOneStep(nn.Sequential):
+ """UpsampleOneStep module (the difference with Upsample is that it always only has 1conv + 1pixelshuffle)
+ Used in lightweight SR to save parameters.
+
+ Args:
+ scale (int): Scale factor. Supported scales: 2^n and 3.
+ num_feat (int): Channel number of intermediate features.
+
+ """
+
+ def __init__(self, scale, num_feat, num_out_ch, input_resolution=None):
+ self.num_feat = num_feat
+ self.input_resolution = input_resolution
+ m = []
+ m.append(nn.Conv2d(num_feat, (scale ** 2) * num_out_ch, 3, 1, 1))
+ m.append(nn.PixelShuffle(scale))
+ super(UpsampleOneStep, self).__init__(*m)
+
+ def flops(self):
+ H, W = self.input_resolution
+ flops = H * W * self.num_feat * 3 * 9
+ return flops
+
+
+
+class Swin2SR(nn.Module):
+ r""" Swin2SR
+ A PyTorch impl of : `Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration`.
+
+ Args:
+ img_size (int | tuple(int)): Input image size. Default 64
+ patch_size (int | tuple(int)): Patch size. Default: 1
+ in_chans (int): Number of input image channels. Default: 3
+ embed_dim (int): Patch embedding dimension. Default: 96
+ depths (tuple(int)): Depth of each Swin Transformer layer.
+ num_heads (tuple(int)): Number of attention heads in different layers.
+ window_size (int): Window size. Default: 7
+ mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4
+ qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True
+ drop_rate (float): Dropout rate. Default: 0
+ attn_drop_rate (float): Attention dropout rate. Default: 0
+ drop_path_rate (float): Stochastic depth rate. Default: 0.1
+ norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.
+ ape (bool): If True, add absolute position embedding to the patch embedding. Default: False
+ patch_norm (bool): If True, add normalization after patch embedding. Default: True
+ use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False
+ upscale: Upscale factor. 2/3/4/8 for image SR, 1 for denoising and compress artifact reduction
+ img_range: Image range. 1. or 255.
+ upsampler: The reconstruction reconstruction module. 'pixelshuffle'/'pixelshuffledirect'/'nearest+conv'/None
+ resi_connection: The convolutional block before residual connection. '1conv'/'3conv'
+ """
+
+ def __init__(self, img_size=64, patch_size=1, in_chans=3,
+ embed_dim=96, depths=[6, 6, 6, 6], num_heads=[6, 6, 6, 6],
+ window_size=7, mlp_ratio=4., qkv_bias=True,
+ drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1,
+ norm_layer=nn.LayerNorm, ape=False, patch_norm=True,
+ use_checkpoint=False, upscale=2, img_range=1., upsampler='', resi_connection='1conv',
+ **kwargs):
+ super(Swin2SR, self).__init__()
+ num_in_ch = in_chans
+ num_out_ch = in_chans
+ num_feat = 64
+ self.img_range = img_range
+ if in_chans == 3:
+ rgb_mean = (0.4488, 0.4371, 0.4040)
+ self.mean = torch.Tensor(rgb_mean).view(1, 3, 1, 1)
+ else:
+ self.mean = torch.zeros(1, 1, 1, 1)
+ self.upscale = upscale
+ self.upsampler = upsampler
+ self.window_size = window_size
+
+ #####################################################################################################
+ ################################### 1, shallow feature extraction ###################################
+ self.conv_first = nn.Conv2d(num_in_ch, embed_dim, 3, 1, 1)
+
+ #####################################################################################################
+ ################################### 2, deep feature extraction ######################################
+ self.num_layers = len(depths)
+ self.embed_dim = embed_dim
+ self.ape = ape
+ self.patch_norm = patch_norm
+ self.num_features = embed_dim
+ self.mlp_ratio = mlp_ratio
+
+ # split image into non-overlapping patches
+ self.patch_embed = PatchEmbed(
+ img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim,
+ norm_layer=norm_layer if self.patch_norm else None)
+ num_patches = self.patch_embed.num_patches
+ patches_resolution = self.patch_embed.patches_resolution
+ self.patches_resolution = patches_resolution
+
+ # merge non-overlapping patches into image
+ self.patch_unembed = PatchUnEmbed(
+ img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim,
+ norm_layer=norm_layer if self.patch_norm else None)
+
+ # absolute position embedding
+ if self.ape:
+ self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim))
+ trunc_normal_(self.absolute_pos_embed, std=.02)
+
+ self.pos_drop = nn.Dropout(p=drop_rate)
+
+ # stochastic depth
+ dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule
+
+ # build Residual Swin Transformer blocks (RSTB)
+ self.layers = nn.ModuleList()
+ for i_layer in range(self.num_layers):
+ layer = RSTB(dim=embed_dim,
+ input_resolution=(patches_resolution[0],
+ patches_resolution[1]),
+ depth=depths[i_layer],
+ num_heads=num_heads[i_layer],
+ window_size=window_size,
+ mlp_ratio=self.mlp_ratio,
+ qkv_bias=qkv_bias,
+ drop=drop_rate, attn_drop=attn_drop_rate,
+ drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # no impact on SR results
+ norm_layer=norm_layer,
+ downsample=None,
+ use_checkpoint=use_checkpoint,
+ img_size=img_size,
+ patch_size=patch_size,
+ resi_connection=resi_connection
+
+ )
+ self.layers.append(layer)
+
+ if self.upsampler == 'pixelshuffle_hf':
+ self.layers_hf = nn.ModuleList()
+ for i_layer in range(self.num_layers):
+ layer = RSTB(dim=embed_dim,
+ input_resolution=(patches_resolution[0],
+ patches_resolution[1]),
+ depth=depths[i_layer],
+ num_heads=num_heads[i_layer],
+ window_size=window_size,
+ mlp_ratio=self.mlp_ratio,
+ qkv_bias=qkv_bias,
+ drop=drop_rate, attn_drop=attn_drop_rate,
+ drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # no impact on SR results
+ norm_layer=norm_layer,
+ downsample=None,
+ use_checkpoint=use_checkpoint,
+ img_size=img_size,
+ patch_size=patch_size,
+ resi_connection=resi_connection
+
+ )
+ self.layers_hf.append(layer)
+
+ self.norm = norm_layer(self.num_features)
+
+ # build the last conv layer in deep feature extraction
+ if resi_connection == '1conv':
+ self.conv_after_body = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1)
+ elif resi_connection == '3conv':
+ # to save parameters and memory
+ self.conv_after_body = nn.Sequential(nn.Conv2d(embed_dim, embed_dim // 4, 3, 1, 1),
+ nn.LeakyReLU(negative_slope=0.2, inplace=True),
+ nn.Conv2d(embed_dim // 4, embed_dim // 4, 1, 1, 0),
+ nn.LeakyReLU(negative_slope=0.2, inplace=True),
+ nn.Conv2d(embed_dim // 4, embed_dim, 3, 1, 1))
+
+ #####################################################################################################
+ ################################ 3, high quality image reconstruction ################################
+ if self.upsampler == 'pixelshuffle':
+ # for classical SR
+ self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.upsample = Upsample(upscale, num_feat)
+ self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+ elif self.upsampler == 'pixelshuffle_aux':
+ self.conv_bicubic = nn.Conv2d(num_in_ch, num_feat, 3, 1, 1)
+ self.conv_before_upsample = nn.Sequential(
+ nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.conv_aux = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+ self.conv_after_aux = nn.Sequential(
+ nn.Conv2d(3, num_feat, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.upsample = Upsample(upscale, num_feat)
+ self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+
+ elif self.upsampler == 'pixelshuffle_hf':
+ self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.upsample = Upsample(upscale, num_feat)
+ self.upsample_hf = Upsample_hf(upscale, num_feat)
+ self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+ self.conv_first_hf = nn.Sequential(nn.Conv2d(num_feat, embed_dim, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.conv_after_body_hf = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1)
+ self.conv_before_upsample_hf = nn.Sequential(
+ nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.conv_last_hf = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+
+ elif self.upsampler == 'pixelshuffledirect':
+ # for lightweight SR (to save parameters)
+ self.upsample = UpsampleOneStep(upscale, embed_dim, num_out_ch,
+ (patches_resolution[0], patches_resolution[1]))
+ elif self.upsampler == 'nearest+conv':
+ # for real-world SR (less artifacts)
+ assert self.upscale == 4, 'only support x4 now.'
+ self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+ nn.LeakyReLU(inplace=True))
+ self.conv_up1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
+ self.conv_up2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
+ self.conv_hr = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
+ self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+ self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True)
+ else:
+ # for image denoising and JPEG compression artifact reduction
+ self.conv_last = nn.Conv2d(embed_dim, num_out_ch, 3, 1, 1)
+
+ self.apply(self._init_weights)
+
+ def _init_weights(self, m):
+ if isinstance(m, nn.Linear):
+ trunc_normal_(m.weight, std=.02)
+ if isinstance(m, nn.Linear) and m.bias is not None:
+ nn.init.constant_(m.bias, 0)
+ elif isinstance(m, nn.LayerNorm):
+ nn.init.constant_(m.bias, 0)
+ nn.init.constant_(m.weight, 1.0)
+
+ @torch.jit.ignore
+ def no_weight_decay(self):
+ return {'absolute_pos_embed'}
+
+ @torch.jit.ignore
+ def no_weight_decay_keywords(self):
+ return {'relative_position_bias_table'}
+
+ def check_image_size(self, x):
+ _, _, h, w = x.size()
+ mod_pad_h = (self.window_size - h % self.window_size) % self.window_size
+ mod_pad_w = (self.window_size - w % self.window_size) % self.window_size
+ x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), 'reflect')
+ return x
+
+ def forward_features(self, x):
+ x_size = (x.shape[2], x.shape[3])
+ x = self.patch_embed(x)
+ if self.ape:
+ x = x + self.absolute_pos_embed
+ x = self.pos_drop(x)
+
+ for layer in self.layers:
+ x = layer(x, x_size)
+
+ x = self.norm(x) # B L C
+ x = self.patch_unembed(x, x_size)
+
+ return x
+
+ def forward_features_hf(self, x):
+ x_size = (x.shape[2], x.shape[3])
+ x = self.patch_embed(x)
+ if self.ape:
+ x = x + self.absolute_pos_embed
+ x = self.pos_drop(x)
+
+ for layer in self.layers_hf:
+ x = layer(x, x_size)
+
+ x = self.norm(x) # B L C
+ x = self.patch_unembed(x, x_size)
+
+ return x
+
+ def forward(self, x):
+ H, W = x.shape[2:]
+ x = self.check_image_size(x)
+
+ self.mean = self.mean.type_as(x)
+ x = (x - self.mean) * self.img_range
+
+ if self.upsampler == 'pixelshuffle':
+ # for classical SR
+ x = self.conv_first(x)
+ x = self.conv_after_body(self.forward_features(x)) + x
+ x = self.conv_before_upsample(x)
+ x = self.conv_last(self.upsample(x))
+ elif self.upsampler == 'pixelshuffle_aux':
+ bicubic = F.interpolate(x, size=(H * self.upscale, W * self.upscale), mode='bicubic', align_corners=False)
+ bicubic = self.conv_bicubic(bicubic)
+ x = self.conv_first(x)
+ x = self.conv_after_body(self.forward_features(x)) + x
+ x = self.conv_before_upsample(x)
+ aux = self.conv_aux(x) # b, 3, LR_H, LR_W
+ x = self.conv_after_aux(aux)
+ x = self.upsample(x)[:, :, :H * self.upscale, :W * self.upscale] + bicubic[:, :, :H * self.upscale, :W * self.upscale]
+ x = self.conv_last(x)
+ aux = aux / self.img_range + self.mean
+ elif self.upsampler == 'pixelshuffle_hf':
+ # for classical SR with HF
+ x = self.conv_first(x)
+ x = self.conv_after_body(self.forward_features(x)) + x
+ x_before = self.conv_before_upsample(x)
+ x_out = self.conv_last(self.upsample(x_before))
+
+ x_hf = self.conv_first_hf(x_before)
+ x_hf = self.conv_after_body_hf(self.forward_features_hf(x_hf)) + x_hf
+ x_hf = self.conv_before_upsample_hf(x_hf)
+ x_hf = self.conv_last_hf(self.upsample_hf(x_hf))
+ x = x_out + x_hf
+ x_hf = x_hf / self.img_range + self.mean
+
+ elif self.upsampler == 'pixelshuffledirect':
+ # for lightweight SR
+ x = self.conv_first(x)
+ x = self.conv_after_body(self.forward_features(x)) + x
+ x = self.upsample(x)
+ elif self.upsampler == 'nearest+conv':
+ # for real-world SR
+ x = self.conv_first(x)
+ x = self.conv_after_body(self.forward_features(x)) + x
+ x = self.conv_before_upsample(x)
+ x = self.lrelu(self.conv_up1(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest')))
+ x = self.lrelu(self.conv_up2(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest')))
+ x = self.conv_last(self.lrelu(self.conv_hr(x)))
+ else:
+ # for image denoising and JPEG compression artifact reduction
+ x_first = self.conv_first(x)
+ res = self.conv_after_body(self.forward_features(x_first)) + x_first
+ x = x + self.conv_last(res)
+
+ x = x / self.img_range + self.mean
+ if self.upsampler == "pixelshuffle_aux":
+ return x[:, :, :H*self.upscale, :W*self.upscale], aux
+
+ elif self.upsampler == "pixelshuffle_hf":
+ x_out = x_out / self.img_range + self.mean
+ return x_out[:, :, :H*self.upscale, :W*self.upscale], x[:, :, :H*self.upscale, :W*self.upscale], x_hf[:, :, :H*self.upscale, :W*self.upscale]
+
+ else:
+ return x[:, :, :H*self.upscale, :W*self.upscale]
+
+ def flops(self):
+ flops = 0
+ H, W = self.patches_resolution
+ flops += H * W * 3 * self.embed_dim * 9
+ flops += self.patch_embed.flops()
+ for i, layer in enumerate(self.layers):
+ flops += layer.flops()
+ flops += H * W * 3 * self.embed_dim * self.embed_dim
+ flops += self.upsample.flops()
+ return flops
+
+
+if __name__ == '__main__':
+ upscale = 4
+ window_size = 8
+ height = (1024 // upscale // window_size + 1) * window_size
+ width = (720 // upscale // window_size + 1) * window_size
+ model = Swin2SR(upscale=2, img_size=(height, width),
+ window_size=window_size, img_range=1., depths=[6, 6, 6, 6],
+ embed_dim=60, num_heads=[6, 6, 6, 6], mlp_ratio=2, upsampler='pixelshuffledirect')
+ print(model)
+ print(height, width, model.flops() / 1e9)
+
+ x = torch.randn((1, 3, height, width))
+ x = model(x)
+ print(x.shape)
\ No newline at end of file |