diff --git a/ccrestoration/arch/__init__.py b/ccrestoration/arch/__init__.py index b55e0a6..be57604 100644 --- a/ccrestoration/arch/__init__.py +++ b/ccrestoration/arch/__init__.py @@ -15,3 +15,4 @@ from ccrestoration.arch.scunet_arch import SCUNet # noqa from ccrestoration.arch.dat_arch import DAT # noqa from ccrestoration.arch.srcnn_arch import SRCNN # noqa +from ccrestoration.arch.hat_arch import HAT # noqa diff --git a/ccrestoration/arch/hat_arch.py b/ccrestoration/arch/hat_arch.py new file mode 100644 index 0000000..19355dd --- /dev/null +++ b/ccrestoration/arch/hat_arch.py @@ -0,0 +1,1063 @@ +# type: ignore +import math + +import torch +from einops import rearrange +from torch import nn +from torch.nn.init import trunc_normal_ + +from ccrestoration.arch import ARCH_REGISTRY +from ccrestoration.arch.arch_util import DropPath, to_2tuple +from ccrestoration.type import ArchType + + +@ARCH_REGISTRY.register(name=ArchType.HAT) +class HAT(nn.Module): + r"""Hybrid Attention Transformer + A PyTorch implementation of : `Activating More Pixels in Image Super-Resolution Transformer`. + Some codes are based on SwinIR. + + 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 + depth (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 + 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, + depth=(6, 6, 6, 6), + num_heads=(6, 6, 6, 6), + window_size=7, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + mlp_ratio=4.0, + qkv_bias=True, + qk_scale=None, + drop_rate=0.0, + attn_drop_rate=0.0, + drop_path_rate=0.1, + norm_layer=nn.LayerNorm, + ape=False, + patch_norm=True, + upscale=2, + img_range=1.0, + upsampler="", + resi_connection="1conv", + **kwargs, + ): + super(HAT, self).__init__() + + self.window_size = window_size + self.shift_size = window_size // 2 + self.overlap_ratio = overlap_ratio + + num_in_ch = in_chans + num_out_ch = in_chans + num_feat = 64 + self.img_range = img_range + if in_chans == 3: + rgb_mean = (0.5, 0.5, 0.5) + 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 + + # relative position index + relative_position_index_SA = self.calculate_rpi_sa() + relative_position_index_OCA = self.calculate_rpi_oca() + self.register_buffer("relative_position_index_SA", relative_position_index_SA) + self.register_buffer("relative_position_index_OCA", relative_position_index_OCA) + + # ------------------------- 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(depth) + 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 is True: + self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim)) + trunc_normal_(self.absolute_pos_embed, std=0.02) + + self.pos_drop = nn.Dropout(p=drop_rate) + + # stochastic depth + dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depth))] # stochastic depth decay rule + + # build Residual Hybrid Attention Groups (RHAG) + self.layers = nn.ModuleList() + for i_layer in range(self.num_layers): + layer = RHAG( + dim=embed_dim, + input_resolution=(patches_resolution[0], patches_resolution[1]), + depth=depth[i_layer], + num_heads=num_heads[i_layer], + window_size=window_size, + compress_ratio=compress_ratio, + squeeze_factor=squeeze_factor, + conv_scale=conv_scale, + overlap_ratio=overlap_ratio, + 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(depth[:i_layer]) : sum(depth[: i_layer + 1])], # no impact on SR results + norm_layer=norm_layer, + downsample=None, + 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 == "identity": + self.conv_after_body = nn.Identity() + + # ------------------------- 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) + + self.apply(self._init_weights) + + def _init_weights(self, m): + if isinstance(m, nn.Linear): + trunc_normal_(m.weight, std=0.02) + 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) + + def calculate_rpi_sa(self): + # calculate relative position index for SA + coords_h = torch.arange(self.window_size) + coords_w = torch.arange(self.window_size) + coords = torch.stack(torch.meshgrid([coords_h, coords_w], indexing="ij")) # 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 - 1 # shift to start from 0 + relative_coords[:, :, 1] += self.window_size - 1 + relative_coords[:, :, 0] *= 2 * self.window_size - 1 + relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww + return relative_position_index + + def calculate_rpi_oca(self): + # calculate relative position index for OCA + window_size_ori = self.window_size + window_size_ext = self.window_size + int(self.overlap_ratio * self.window_size) + + coords_h = torch.arange(window_size_ori) + coords_w = torch.arange(window_size_ori) + coords_ori = torch.stack(torch.meshgrid([coords_h, coords_w], indexing="ij")) # 2, ws, ws + coords_ori_flatten = torch.flatten(coords_ori, 1) # 2, ws*ws + + coords_h = torch.arange(window_size_ext) + coords_w = torch.arange(window_size_ext) + coords_ext = torch.stack(torch.meshgrid([coords_h, coords_w], indexing="ij")) # 2, wse, wse + coords_ext_flatten = torch.flatten(coords_ext, 1) # 2, wse*wse + + relative_coords = coords_ext_flatten[:, None, :] - coords_ori_flatten[:, :, None] # 2, ws*ws, wse*wse + + relative_coords = relative_coords.permute(1, 2, 0).contiguous() # ws*ws, wse*wse, 2 + relative_coords[:, :, 0] += window_size_ori - window_size_ext + 1 # shift to start from 0 + relative_coords[:, :, 1] += window_size_ori - window_size_ext + 1 + + relative_coords[:, :, 0] *= window_size_ori + window_size_ext - 1 + relative_position_index = relative_coords.sum(-1) + return relative_position_index + + 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, -100.0).masked_fill(attn_mask == 0, 0.0) + + return attn_mask + + @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 forward_features(self, x): + x_size = (x.shape[2], x.shape[3]) + + # Calculate attention mask and relative position index in advance to speed up inference. + # The original code is very time-cosuming for large window size. + attn_mask = self.calculate_mask(x_size).to(x.device) + params = { + "attn_mask": attn_mask, + "rpi_sa": self.relative_position_index_SA, + "rpi_oca": self.relative_position_index_OCA, + } + + x = self.patch_embed(x) + if self.ape is True: + x = x + self.absolute_pos_embed + x = self.pos_drop(x) + + for layer in self.layers: + x = layer(x, x_size, params) + + x = self.norm(x) # b seq_len c + x = self.patch_unembed(x, x_size) + + return x + + def forward(self, 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)) + + x = x / self.img_range + self.mean + + return x + + +class ChannelAttention(nn.Module): + """Channel attention used in RCAN. + + Args: + ---- + num_feat (int): Channel number of intermediate features. + squeeze_factor (int): Channel squeeze factor. Default: 16. + + """ + + def __init__(self, num_feat, squeeze_factor=16): + super(ChannelAttention, self).__init__() + self.attention = nn.Sequential( + nn.AdaptiveAvgPool2d(1), + nn.Conv2d(num_feat, num_feat // squeeze_factor, 1, padding=0), + nn.ReLU(inplace=True), + nn.Conv2d(num_feat // squeeze_factor, num_feat, 1, padding=0), + nn.Sigmoid(), + ) + + def forward(self, x): + y = self.attention(x) + return x * y + + +class CAB(nn.Module): + def __init__(self, num_feat, compress_ratio=3, squeeze_factor=30): + super(CAB, self).__init__() + + self.cab = nn.Sequential( + nn.Conv2d(num_feat, num_feat // compress_ratio, 3, 1, 1), + nn.GELU(), + nn.Conv2d(num_feat // compress_ratio, num_feat, 3, 1, 1), + ChannelAttention(num_feat, squeeze_factor), + ) + + def forward(self, x): + return self.cab(x) + + +class Mlp(nn.Module): + def __init__( + self, + in_features, + hidden_features=None, + out_features=None, + act_layer=nn.GELU, + drop=0.0, + ): + 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=0.0, + proj_drop=0.0, + ): + 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 + + 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=0.02) + self.softmax = nn.Softmax(dim=-1) + + def forward(self, x, rpi, 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[rpi.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 + + +class HAB(nn.Module): + r"""Hybrid Attention 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, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + mlp_ratio=4.0, + qkv_bias=True, + qk_scale=None, + drop=0.0, + attn_drop=0.0, + drop_path=0.0, + 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.conv_scale = conv_scale + self.conv_block = CAB(num_feat=dim, compress_ratio=compress_ratio, squeeze_factor=squeeze_factor) + + self.drop_path = DropPath(drop_path) if drop_path > 0.0 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, + ) + + def forward(self, x, x_size, rpi_sa, attn_mask): + h, w = x_size + b, _, c = x.shape + # assert seq_len == h * w, "input feature has wrong size" + + shortcut = x + x = self.norm1(x) + x = x.view(b, h, w, c) + + # Conv_X + conv_x = self.conv_block(x.permute(0, 3, 1, 2)) + conv_x = conv_x.permute(0, 2, 3, 1).contiguous().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)) + attn_mask = attn_mask + else: + shifted_x = x + attn_mask = None + + # 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 + attn_windows = self.attn(x_windows, rpi=rpi_sa, mask=attn_mask) + + # 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: + attn_x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)) + else: + attn_x = shifted_x + attn_x = attn_x.view(b, h * w, c) + + # FFN + x = shortcut + self.drop_path(attn_x) + conv_x * self.conv_scale + x = x + self.drop_path(self.mlp(self.norm2(x))) + + return x + + +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, seq_len, c = x.shape + assert seq_len == 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 + + +class OCAB(nn.Module): + # overlapping cross-attention block + + def __init__( + self, + dim, + input_resolution, + window_size, + overlap_ratio, + num_heads, + qkv_bias=True, + qk_scale=None, + mlp_ratio=2, + norm_layer=nn.LayerNorm, + ): + super().__init__() + self.dim = dim + self.input_resolution = input_resolution + self.window_size = window_size + self.num_heads = num_heads + head_dim = dim // num_heads + self.scale = qk_scale or head_dim**-0.5 + self.overlap_win_size = int(window_size * overlap_ratio) + window_size + + self.norm1 = norm_layer(dim) + self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) + self.unfold = nn.Unfold( + kernel_size=(self.overlap_win_size, self.overlap_win_size), + stride=window_size, + padding=(self.overlap_win_size - window_size) // 2, + ) + + # define a parameter table of relative position bias + self.relative_position_bias_table = nn.Parameter( + torch.zeros( + (window_size + self.overlap_win_size - 1) * (window_size + self.overlap_win_size - 1), + num_heads, + ) + ) # 2*Wh-1 * 2*Ww-1, nH + + trunc_normal_(self.relative_position_bias_table, std=0.02) + self.softmax = nn.Softmax(dim=-1) + + self.proj = nn.Linear(dim, dim) + + 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=nn.GELU) + + def forward(self, x, x_size, rpi): + h, w = x_size + b, _, c = x.shape + + shortcut = x + x = self.norm1(x) + x = x.view(b, h, w, c) + + qkv = self.qkv(x).reshape(b, h, w, 3, c).permute(3, 0, 4, 1, 2) # 3, b, c, h, w + q = qkv[0].permute(0, 2, 3, 1) # b, h, w, c + kv = torch.cat((qkv[1], qkv[2]), dim=1) # b, 2*c, h, w + + # partition windows + q_windows = window_partition(q, self.window_size) # nw*b, window_size, window_size, c + q_windows = q_windows.view(-1, self.window_size * self.window_size, c) # nw*b, window_size*window_size, c + + kv_windows = self.unfold(kv) # b, c*w*w, nw + kv_windows = rearrange( + kv_windows, + "b (nc ch owh oww) nw -> nc (b nw) (owh oww) ch", + nc=2, + ch=c, + owh=self.overlap_win_size, + oww=self.overlap_win_size, + ).contiguous() # 2, nw*b, ow*ow, c + k_windows, v_windows = kv_windows[0], kv_windows[1] # nw*b, ow*ow, c + + b_, nq, _ = q_windows.shape + _, n, _ = k_windows.shape + d = self.dim // self.num_heads + q = q_windows.reshape(b_, nq, self.num_heads, d).permute(0, 2, 1, 3) # nw*b, nH, nq, d + k = k_windows.reshape(b_, n, self.num_heads, d).permute(0, 2, 1, 3) # nw*b, nH, n, d + v = v_windows.reshape(b_, n, self.num_heads, d).permute(0, 2, 1, 3) # nw*b, nH, n, d + + q = q * self.scale + attn = q @ k.transpose(-2, -1) + + relative_position_bias = self.relative_position_bias_table[rpi.view(-1)].view( + self.window_size * self.window_size, + self.overlap_win_size * self.overlap_win_size, + -1, + ) # ws*ws, wse*wse, nH + relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # nH, ws*ws, wse*wse + attn = attn + relative_position_bias.unsqueeze(0) + + attn = self.softmax(attn) + attn_windows = (attn @ v).transpose(1, 2).reshape(b_, nq, self.dim) + + # merge windows + attn_windows = attn_windows.view(-1, self.window_size, self.window_size, self.dim) + x = window_reverse(attn_windows, self.window_size, h, w) # b h w c + x = x.view(b, h * w, self.dim) + + x = self.proj(x) + shortcut + + x = x + self.mlp(self.norm2(x)) + return x + + +class AttenBlocks(nn.Module): + """A series of attention blocks for one RHAG. + + 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 + + """ + + def __init__( + self, + dim, + input_resolution, + depth, + num_heads, + window_size, + compress_ratio, + squeeze_factor, + conv_scale, + overlap_ratio, + mlp_ratio=4.0, + qkv_bias=True, + qk_scale=None, + drop=0.0, + attn_drop=0.0, + drop_path=0.0, + norm_layer=nn.LayerNorm, + downsample=None, + ): + super().__init__() + self.dim = dim + self.input_resolution = input_resolution + self.depth = depth + + # build blocks + self.blocks = nn.ModuleList( + [ + HAB( + 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, + compress_ratio=compress_ratio, + squeeze_factor=squeeze_factor, + conv_scale=conv_scale, + 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) + ] + ) + + # OCAB + self.overlap_attn = OCAB( + dim=dim, + input_resolution=input_resolution, + window_size=window_size, + overlap_ratio=overlap_ratio, + num_heads=num_heads, + qkv_bias=qkv_bias, + qk_scale=qk_scale, + mlp_ratio=mlp_ratio, + norm_layer=norm_layer, + ) + + # 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, params): + for blk in self.blocks: + x = blk(x, x_size, params["rpi_sa"], params["attn_mask"]) + + x = self.overlap_attn(x, x_size, params["rpi_oca"]) + + if self.downsample is not None: + x = self.downsample(x) + return x + + +class RHAG(nn.Module): + """Residual Hybrid Attention Group (RHAG). + + 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 + 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, + compress_ratio, + squeeze_factor, + conv_scale, + overlap_ratio, + mlp_ratio=4.0, + qkv_bias=True, + qk_scale=None, + drop=0.0, + attn_drop=0.0, + drop_path=0.0, + norm_layer=nn.LayerNorm, + downsample=None, + img_size=224, + patch_size=4, + resi_connection="1conv", + ): + super(RHAG, self).__init__() + + self.dim = dim + self.input_resolution = input_resolution + + self.residual_group = AttenBlocks( + dim=dim, + input_resolution=input_resolution, + depth=depth, + num_heads=num_heads, + window_size=window_size, + compress_ratio=compress_ratio, + squeeze_factor=squeeze_factor, + conv_scale=conv_scale, + overlap_ratio=overlap_ratio, + 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, + ) + + if resi_connection == "1conv": + self.conv = nn.Conv2d(dim, dim, 3, 1, 1) + elif resi_connection == "identity": + self.conv = nn.Identity() + + 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, params): + return self.patch_embed(self.conv(self.patch_unembed(self.residual_group(x, x_size, params), x_size))) + x + + +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 + + +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): + x = x.transpose(1, 2).contiguous().view(x.shape[0], self.embed_dim, x_size[0], x_size[1]) # b Ph*Pw c + return x + + +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.log2(scale))): + 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) diff --git a/ccrestoration/config/__init__.py b/ccrestoration/config/__init__.py index ef39a0a..151a778 100644 --- a/ccrestoration/config/__init__.py +++ b/ccrestoration/config/__init__.py @@ -14,3 +14,4 @@ from ccrestoration.config.scunet_config import SCUNetConfig # noqa from ccrestoration.config.dat_config import DATConfig # noqa from ccrestoration.config.srcnn_config import SRCNNConfig # noqa +from ccrestoration.config.hat_config import HATConfig # noqa diff --git a/ccrestoration/config/hat_config.py b/ccrestoration/config/hat_config.py new file mode 100644 index 0000000..cfd5dfa --- /dev/null +++ b/ccrestoration/config/hat_config.py @@ -0,0 +1,334 @@ +from typing import Any, List, Optional, Tuple, Union + +from pydantic import field_validator +from torch import nn + +from ccrestoration.config import CONFIG_REGISTRY +from ccrestoration.type import ArchType, BaseConfig, ConfigType, ModelType + + +class HATConfig(BaseConfig): + arch: Union[ArchType, str] = ArchType.HAT + model: Union[ModelType, str] = ModelType.HAT + scale: int = 2 + patch_size: int = 1 + in_chans: int = 3 + img_size: Union[int, Tuple[int, ...]] = 64 + img_range: float = 1.0 + depth: Union[List[int], Tuple[int, ...]] = (6, 6, 6, 6) + embed_dim: int = 96 + num_heads: Union[List[int], Tuple[int, ...]] = (6, 6, 6, 6) + window_size: int = 7 + compress_ratio: int = 3 + squeeze_factor: int = 30 + conv_scale: float = 0.01 + overlap_ratio: float = 0.5 + mlp_ratio: float = 4.0 + resi_connection: str = "1conv" + qkv_bias: bool = True + qk_scale: Optional[float] = None + drop_rate: float = 0.0 + attn_drop_rate: float = 0.0 + drop_path_rate: float = 0.1 + ape: bool = False + patch_norm: bool = True + act_layer: Any = nn.GELU + norm_layer: Any = nn.LayerNorm + upsampler: str = "pixelshuffle" + + @field_validator("scale") + def scale_match(cls, v: int) -> int: + if v not in [1, 2, 3, 4, 8]: + raise ValueError("scale factor must be one of [1, 2, 3, 4, 8]") + return v + + @field_validator("upsampler") + def upsampler_match(cls, v: str) -> str: + if v not in ["pixelshuffle", "pixelshuffledirect", "nearest+conv", ""]: + raise ValueError("Upsampler must be one of ['pixelshuffle','pixelshuffledirect','nearest+conv', '']") + return v + + @field_validator("resi_connection") + def resi_connection_match(cls, v: str) -> str: + if v not in ["1conv", "3conv"]: + raise ValueError("Residual connection must be one of ['1conv', '3conv']") + return v + + +HATConfigs = [ + # official models + # hat_s + HATConfig( + name=ConfigType.HAT_S_2x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_S_2x.pth", + hash="3e249a901c2aed6b82548875e21847ef6c015a40c814237a7a0abb10c69d5ddf", + scale=2, + in_chans=3, + window_size=16, + compress_ratio=24, + squeeze_factor=24, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=144, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_S_3x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_S_3x.pth", + hash="3733345caeb2b6cb0a7514d634b143f318e2441faf32316195a47e4cda67669e", + scale=3, + in_chans=3, + window_size=16, + compress_ratio=24, + squeeze_factor=24, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=144, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_S_4x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_S_4x.pth", + hash="a92f81bd2c0c1aaa371a6e4d6cac69e749fde2e36196885ee47a4a3667542c9a", + scale=4, + in_chans=3, + window_size=16, + compress_ratio=24, + squeeze_factor=24, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=144, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + # hat_m + HATConfig( + name=ConfigType.HAT_2x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_2x.pth", + hash="35e9409849016ab2bfc0c549639ea4f1d4d5b2bdd75f856a925e124dd05670d0", + scale=2, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_3x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_3x.pth", + hash="453c90b94d494d041c8bff2d222987af311f07b4a09b4dafbd9ab6492f15206c", + scale=3, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_4x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_4x.pth", + hash="02dabea478aa5902a7170ad89350124e691bd89c91356f24b3267022622dc030", + scale=4, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + # hat_l + HATConfig( + name=ConfigType.HAT_L_ImageNet_pretrain_2x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_L_ImageNet_pretrain_2x.pth", + hash="2818c7ca8d72ec4cc5f31c93203d55252a662dd35cda34ce1a69661f97dcd38f", + scale=2, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_L_ImageNet_pretrain_3x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_L_ImageNet_pretrain_3x.pth", + hash="78af181bedf1e805fd7517d3738bc5824f7ebdc477fc9f757b708bcf49ad4e3d", + scale=3, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_L_ImageNet_pretrain_4x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_L_ImageNet_pretrain_4x.pth", + hash="5992bd38522f2b8faf11ea4bd8ee08de92465bb66892166576999afc36d60043", + scale=4, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + # hat_image_net + HATConfig( + name=ConfigType.HAT_ImageNet_pretrain_2x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_ImageNet_pretrain_2x.pth", + hash="82ebd911263bcc886fbef46b30cf97b92a932a27a3cba30163d4577afb09b9d7", + scale=2, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_ImageNet_pretrain_3x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_ImageNet_pretrain_3x.pth", + hash="8469db903d464497f94419587054c9e83dff994edc409d563ab8f5a503767be8", + scale=3, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_ImageNet_pretrain_4x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_ImageNet_pretrain_4x.pth", + hash="4ee053c42461187846dc0e93aa5abd34591c0725a8e044a59000e92ee215e833", + scale=4, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + # hat_real_gan + HATConfig( + name=ConfigType.HAT_Real_GAN_4x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_Real_GAN_4x.pth", + hash="f5b1e3bbbb05147ca2beefcc715279cb647d7976cbda67d62ea7e6e20d5ffcc7", + scale=4, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), + HATConfig( + name=ConfigType.HAT_Real_GAN_sharper_4x, + url="https://github.com/TensoRaws/ccrestoration/releases/download/model_zoo/HAT_Real_GAN_sharper_4x.pth", + hash="5800b67136006eb8cab3b4ed7c8d73b6a195bb18e6cc709b674f9aa069c00271", + scale=4, + in_chans=3, + window_size=16, + compress_ratio=3, + squeeze_factor=30, + conv_scale=0.01, + overlap_ratio=0.5, + img_range=1.0, + depth=[6, 6, 6, 6, 6, 6], + embed_dim=180, + num_heads=[6, 6, 6, 6, 6, 6], + mlp_ratio=2, + upsampler="pixelshuffle", + resi_connection="1conv", + ), +] + +for cfg in HATConfigs: + CONFIG_REGISTRY.register(cfg) diff --git a/ccrestoration/model/__init__.py b/ccrestoration/model/__init__.py index f715542..0a4b940 100644 --- a/ccrestoration/model/__init__.py +++ b/ccrestoration/model/__init__.py @@ -18,3 +18,4 @@ from ccrestoration.model.scunet_model import SCUNetModel # noqa from ccrestoration.model.dat_model import DATModel # noqa from ccrestoration.model.srcnn_model import SRCNNModel # noqa +from ccrestoration.model.hat_model import HATModel # noqa diff --git a/ccrestoration/model/hat_model.py b/ccrestoration/model/hat_model.py new file mode 100644 index 0000000..7c8ef6d --- /dev/null +++ b/ccrestoration/model/hat_model.py @@ -0,0 +1,53 @@ +from typing import Any + +from ccrestoration.arch import HAT +from ccrestoration.config import HATConfig +from ccrestoration.model import MODEL_REGISTRY +from ccrestoration.model.sr_base_model import SRBaseModel +from ccrestoration.type import ModelType + + +@MODEL_REGISTRY.register(name=ModelType.HAT) +class HATModel(SRBaseModel): + def load_model(self) -> Any: + cfg: HATConfig = self.config + state_dict = self.get_state_dict() + + if "params_ema" in state_dict: + state_dict = state_dict["params_ema"] + elif "params" in state_dict: + state_dict = state_dict["params"] + elif "model_state_dict" in state_dict: + # For APISR's model + state_dict = state_dict["model_state_dict"] + + model = HAT( + img_size=cfg.img_size, + patch_size=cfg.patch_size, + in_chans=cfg.in_chans, + embed_dim=cfg.embed_dim, + depth=cfg.depth, + num_heads=cfg.num_heads, + window_size=cfg.window_size, + compress_ratio=cfg.compress_ratio, + squeeze_factor=cfg.squeeze_factor, + conv_scale=cfg.conv_scale, + overlap_ratio=cfg.overlap_ratio, + mlp_ratio=cfg.mlp_ratio, + qkv_bias=cfg.qkv_bias, + qk_scale=cfg.qk_scale, + drop_rate=cfg.drop_rate, + attn_drop_rate=cfg.attn_drop_rate, + drop_path_rate=cfg.drop_path_rate, + norm_layer=cfg.norm_layer, + ape=cfg.ape, + patch_norm=cfg.patch_norm, + upscale=cfg.scale, + img_range=cfg.img_range, + upsampler=cfg.upsampler, + resi_connection=cfg.resi_connection, + ) + + model.load_state_dict(state_dict) + model.eval().to(self.device) + return model diff --git a/ccrestoration/type/arch.py b/ccrestoration/type/arch.py index aa178ef..733ce8b 100644 --- a/ccrestoration/type/arch.py +++ b/ccrestoration/type/arch.py @@ -13,6 +13,7 @@ class ArchType(str, Enum): SCUNET = "SCUNET" DAT = "DAT" SRCNN = "SRCNN" + HAT = "HAT" # ------------------------------------- Auxiliary Network ---------------------------------------------------------- diff --git a/ccrestoration/type/config.py b/ccrestoration/type/config.py index cda314c..003e303 100644 --- a/ccrestoration/type/config.py +++ b/ccrestoration/type/config.py @@ -77,6 +77,22 @@ class ConfigType(str, Enum): SRCNN_3x = "SRCNN_3x.pth" SRCNN_4x = "SRCNN_4x.pth" + # HAT + HAT_S_2x = "HAT_S_2x.pth" + HAT_S_3x = "HAT_S_3x.pth" + HAT_S_4x = "HAT_S_4x.pth" + HAT_2x = "HAT_2x.pth" + HAT_3x = "HAT_3x.pth" + HAT_4x = "HAT_4x.pth" + HAT_Real_GAN_sharper_4x = "HAT_Real_GAN_sharper_4x.pth" + HAT_Real_GAN_4x = "HAT_Real_GAN_4x.pth" + HAT_ImageNet_pretrain_2x = "HAT_ImageNet_pretrain_2x.pth" + HAT_ImageNet_pretrain_3x = "HAT_ImageNet_pretrain_3x.pth" + HAT_ImageNet_pretrain_4x = "HAT_ImageNet_pretrain_4x.pth" + HAT_L_ImageNet_pretrain_2x = "HAT_L_ImageNet_pretrain_2x.pth" + HAT_L_ImageNet_pretrain_3x = "HAT_L_ImageNet_pretrain_3x.pth" + HAT_L_ImageNet_pretrain_4x = "HAT_L_ImageNet_pretrain_4x.pth" + # ------------------------------------- Auxiliary Network ---------------------------------------------------------- # SpyNet diff --git a/ccrestoration/type/model.py b/ccrestoration/type/model.py index 387e4e4..d92cc96 100644 --- a/ccrestoration/type/model.py +++ b/ccrestoration/type/model.py @@ -12,6 +12,7 @@ class ModelType(str, Enum): SCUNet = "SCUNet" DAT = "DAT" SRCNN = "SRCNN" + HAT = "HAT" # ------------------------------------- Auxiliary Network ---------------------------------------------------------- diff --git a/tests/test_hat.py b/tests/test_hat.py new file mode 100644 index 0000000..c939728 --- /dev/null +++ b/tests/test_hat.py @@ -0,0 +1,76 @@ +import os + +import cv2 +import pytest + +from ccrestoration import AutoConfig, AutoModel, BaseConfig, ConfigType +from ccrestoration.model import SRBaseModel + +from .util import ASSETS_PATH, calculate_image_similarity, compare_image_size, get_device, load_image + + +class Test_HAT: + def test_official_light(self) -> None: + img1 = load_image() + + for k in [ + ConfigType.HAT_S_2x, + ConfigType.HAT_S_3x, + ConfigType.HAT_S_4x, + ]: + print(f"Testing {k}") + cfg: BaseConfig = AutoConfig.from_pretrained(k) + model: SRBaseModel = AutoModel.from_config(config=cfg, fp16=False, device=get_device()) + print(model.device) + + img2 = model.inference_image(img1) + cv2.imwrite(str(ASSETS_PATH / f"test_{k}_out.jpg"), img2) + + assert calculate_image_similarity(img1, img2) + assert compare_image_size(img1, img2, cfg.scale) + + @pytest.mark.skipif(os.environ.get("GITHUB_ACTIONS") == "true", reason="Skip on CI test") + def test_official(self) -> None: + img1 = load_image() + + for k in [ + ConfigType.HAT_S_2x, + ConfigType.HAT_S_3x, + ConfigType.HAT_S_4x, + ConfigType.HAT_2x, + ConfigType.HAT_3x, + ConfigType.HAT_4x, + ConfigType.HAT_ImageNet_pretrain_2x, + ConfigType.HAT_ImageNet_pretrain_3x, + ConfigType.HAT_ImageNet_pretrain_4x, + ConfigType.HAT_L_ImageNet_pretrain_2x, + ConfigType.HAT_L_ImageNet_pretrain_3x, + ConfigType.HAT_L_ImageNet_pretrain_4x, + ConfigType.HAT_Real_GAN_sharper_4x, + ConfigType.HAT_Real_GAN_4x, + ]: + print(f"Testing {k}") + cfg: BaseConfig = AutoConfig.from_pretrained(k) + model: SRBaseModel = AutoModel.from_config(config=cfg, fp16=False, device=get_device()) + print(model.device) + + img2 = model.inference_image(img1) + cv2.imwrite(str(ASSETS_PATH / f"test_{k}_out.jpg"), img2) + + assert calculate_image_similarity(img1, img2) + assert compare_image_size(img1, img2, cfg.scale) + + def test_custom(self) -> None: + img1 = load_image() + + for k in [ConfigType.HAT_Real_GAN_sharper_4x]: + print(f"Testing {k}") + cfg: BaseConfig = AutoConfig.from_pretrained(k) + model: SRBaseModel = AutoModel.from_config(config=cfg, fp16=False, device=get_device()) + print(model.device) + + img2 = model.inference_image(img1) + cv2.imwrite(str(ASSETS_PATH / f"test_{k}_out.jpg"), img2) + + assert calculate_image_similarity(img1, img2) + assert compare_image_size(img1, img2, cfg.scale)