Swin Transformer结构详解与代码运行
最近一直再看感知相关算法，如LSS、pointnet、pointpillar等相关论文与代码，而当看到多相机检测方法bevfomer论文时候，发现其结构使用了self.attention与cross-attention的transformer方法。
介于此，我将原来沉浸几月的swin-tranformer结构回忆一遍，也想通过这次会议记录个人总结，希望对读者有所帮助。
transformer来源NLP结构，可参考我的另一篇博客，了解transformer结构，链接:https://www.cnblogs.com/tangjunjun/p/15617342.html
查看NLP链接博客，肯定对self-attention及cross-attention有所了解，下文将继续介绍相关swin-transformer的内容：
一.动机
与NLP相比，Transformer应用在视觉领域有两个难点：
- 物体尺度变化大。比如在目标检测和语义分割中，物体大小可能在5%-95%甚至更大的范围内，这时候多尺度分层特征图是至关重要的，检测中的FPN、分割中unet的skip connection都是用分层特征图来处理多尺度问题的。
- 图像像素点多，尤其是密集预测的任务。如果把像素点作为token，序列长度会非常大，计算量与图片尺寸的平方成正比。之前的方法要么把特征图作为输入，要么像ViT用patch作为输入，但ViT的特征比较粗糙且没有考虑视觉信号的特点，无法做精细的任务。
Swin Transformer的主要思想是把建模能力很强的transformer和视觉信号的先验联系起来，这些先验具有层次性、局部性和平移不变性，具体做法是用shifted window来建立分层特征图，有了分层特征图就可以用FPN/Unet等结构去做密集预测的任务，而且计算量与图片尺寸成正比。
作为Swin Transformer的重要设计，shifted window有两个含义，一是不重叠的窗口中进行self attention，可以减少计算量，并且引入了局部性先验。此外，不同于传统的滑动窗口，不重叠窗口的设计对硬件实现更加友好，从而具有更快的实际运行速度。Swin Transformer 使用的不重叠窗口中，统一窗口内的点将采用相同的邻域来进行计算，对速度更友好。实际测试表明，非重叠窗口方法的速度比滑动窗口方法快了2倍左右。另一个含义是在两个连续的层中做了移位的操作，在 L 层中，窗口分区从图像的左上角开始，在 L+1 层中，窗口划分则往右下移动了半个窗口，这样的设计保证了不重叠的窗口间可以有信息的交换。
引用:https://zhuanlan.zhihu.com/p/495246390
而本文仅介绍下图红框部分的结构:
二.红框结构介绍
我将根据数据流走向介绍红框的各个模块，现假设输入为[batch,channel,height,width]=[2,3,800,800]
1.Pathch Partition结构：
Each patch is treated as a “token” and its feature is set as a concatenation of the raw pixel RGB values. In our implementation, we use a patch size of 4 × 4 and thus the feature dimension of each patch is 4 × 4 × 3 = 48. A linear embedding layer is applied on this raw-valued feature to project it to an arbitrary dimension (denoted as C). Several Transformer blocks with modified self-attention computation (Swin Transformer blocks) are applied on these patch tokens. The Transformer blocks maintain the number of tokens (H/4 × W/4 ), and together with the linear embedding are referred to as “Stage 1”.
patch embedding模块实现的是patch partition和linear embedding功能，用来切patch并将patch特征嵌入到指定维度。直接用一个kernel_size=4和stride等于patch_size的卷积来实现。模型默认patch_size=4.
x = self.patch_embed(x)
代码如下:
class PatchEmbed(nn.Module): """ Image to Patch Embedding Args: 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, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): super().__init__() patch_size = to_2tuple(patch_size) self.patch_size = patch_size 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): """Forward function.""" # padding _, _, H, W = x.size() if W % self.patch_size[1] != 0: x = F.pad(x, (0, self.patch_size[1] - W % self.patch_size[1])) if H % self.patch_size[0] != 0: x = F.pad(x, (0, 0, 0, self.patch_size[0] - H % self.patch_size[0])) x = self.proj(x) # B C Wh Ww if self.norm is not None: Wh, Ww = x.size(2), x.size(3) x = x.flatten(2).transpose(1, 2) x = self.norm(x) x = x.transpose(1, 2).view(-1, self.embed_dim, Wh, Ww) return x
输出为:[2,96,200,200]
之后将其展平，通过
x = x.flatten(2).transpose(1, 2)
输出为:[2,40000,96]
接下来将介绍如下结构:
BasicLayer( (blocks): ModuleList( (0): SwinTransformerBlock( (norm1): LayerNorm((96,), eps=1e-05, elementwise_affine=True) (attn): WindowAttention( (qkv): Linear(in_features=96, out_features=288, bias=True) (attn_drop): Dropout(p=0.0, inplace=False) (proj): Linear(in_features=96, out_features=96, bias=True) (proj_drop): Dropout(p=0.0, inplace=False) (softmax): Softmax(dim=-1) ) (drop_path): Identity() (norm2): LayerNorm((96,), eps=1e-05, elementwise_affine=True) (mlp): Mlp( (fc1): Linear(in_features=96, out_features=384, bias=True) (act): GELU() (fc2): Linear(in_features=384, out_features=96, bias=True) (drop): Dropout(p=0.0, inplace=False) ) ) (1): SwinTransformerBlock( (norm1): LayerNorm((96,), eps=1e-05, elementwise_affine=True) (attn): WindowAttention( (qkv): Linear(in_features=96, out_features=288, bias=True) (attn_drop): Dropout(p=0.0, inplace=False) (proj): Linear(in_features=96, out_features=96, bias=True) (proj_drop): Dropout(p=0.0, inplace=False) (softmax): Softmax(dim=-1) ) (drop_path): DropPath(drop_prob=0.018) (norm2): LayerNorm((96,), eps=1e-05, elementwise_affine=True) (mlp): Mlp( (fc1): Linear(in_features=96, out_features=384, bias=True) (act): GELU() (fc2): Linear(in_features=384, out_features=96, bias=True) (drop): Dropout(p=0.0, inplace=False) ) ) ) (downsample): PatchMerging( (reduction): Linear(in_features=384, out_features=192, bias=False) (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True) ) )
2.SwinTransformerBlock结构
(1).img_mask将生成方法
每一个block模块均会生成img_mask方法,如下：
随后会对生成的img_mask做如下变化:
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))
有关img_mask与attn_mask的方法，参考:https://blog.csdn.net/weixin_52185313/article/details/125127850
顺道说一个如下结构，后面也会用到，该结构实际是划分窗口。
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
输出:attn_mask=[841,49,49] 来源img_mask=[1,203,203,1]==>[1,29,7,29,7]==>[1,29,29,7,7]==>[841,49]
输出：x=[2,40000,96]还未发生改变
2.SwinTransformerBlock，包含MSA和SW-MSA模块(非常重要)
将以上输出x与attn_mask输入以下结构中:
class SwinTransformerBlock(nn.Module): """ Swin Transformer Block. Args: dim (int): Number of input channels. 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, 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.num_heads = num_heads self.window_size = window_size self.shift_size = shift_size self.mlp_ratio = mlp_ratio 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) self.H = None self.W = None def forward(self, x, mask_matrix): """ Forward function. Args: x: Input feature, tensor size (B, H*W, C). H, W: Spatial resolution of the input feature. mask_matrix: Attention mask for cyclic shift. """ B, L, C = x.shape H, W = self.H, self.W assert L == H * W, "input feature has wrong size" shortcut = x x = self.norm1(x) x = x.view(B, H, W, C) # pad feature maps to multiples of window size pad_l = pad_t = 0 pad_r = (self.window_size - W % self.window_size) % self.window_size pad_b = (self.window_size - H % self.window_size) % self.window_size x = F.pad(x, (0, 0, pad_l, pad_r, pad_t, pad_b)) _, Hp, Wp, _ = x.shape # cyclic shift if self.shift_size > 0: shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) attn_mask = mask_matrix 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 attn_windows = self.attn(x_windows, mask=attn_mask) # nW*B, window_size*window_size, C # merge windows attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) shifted_x = window_reverse(attn_windows, self.window_size, Hp, Wp) # 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 if pad_r > 0 or pad_b > 0: x = x[:, :H, :W, :].contiguous() 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
以下，我将介绍SwinTransformerBlock结构中的分支结构:
# cyclic shift if self.shift_size > 0: shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) attn_mask = mask_matrix else: shifted_x = x attn_mask = None
以上通过self.shift_size控制是否将输入x执行shift操作，若执行shift操作，后面注意力机制将满足论文所说的shift-windows-transformer。
输出:shifted_x=x=[2,40000,96]
以下将通过window_partition将其分成self.window_size的结构，
# partition windows x_windows = window_partition(shifted_x, self.window_size) # nW*B, window_size, window_size, C =[1682,49,96] x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # nW*B, window_size*window_size, C =[]
输出:x_windows=[1682,49,96]
核心重点内容：
将x_windows与attn_mask输入以下代码，而我将不在解释此内容，因为看了我attention的博客将会明白
# W-MSA/SW-MSA attn_windows = self.attn(x_windows, mask=attn_mask) # nW*B, window_size*window_size, C
self.attn代码结构:
class WindowAttention(nn.Module): """ 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): """ Forward function. 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
输出:attn_windows=[1682,49,96]
将attn_windows输入到以下代码,进行reverse:
3.patch_merging结构
# merge windows attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) #[1682,7,7,96] shifted_x = window_reverse(attn_windows, self.window_size, Hp, Wp) # B H' W' C
window_reverse恢复代码:
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
输出:shifted_x=[2,203,203,96]
将shifted_x输入以下代码，进行恢复
# 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 if pad_r > 0 or pad_b > 0: x = x[:, :H, :W, :].contiguous() x = x.view(B, H * W, C)
输出：x=[2,40000,96]
4.FFN操作
# FFN x = shortcut + self.drop_path(x) x = x + self.drop_path(self.mlp(self.norm2(x)))
输出：x=[2,40000,96]
拷贝可执行结构代码：
# -------------------------------------------------------- # Swin Transformer # Copyright (c) 2021 Microsoft # Licensed under The MIT License [see LICENSE for details] # Written by Ze Liu, Yutong Lin, Yixuan Wei # modified by tangjunjun # -------------------------------------------------------- import torch import torch.nn as nn import torch.nn.functional as F import torch.utils.checkpoint as checkpoint import numpy as np from timm.models.layers import DropPath, to_2tuple, trunc_normal_ # swin_transformer分类模块 class swin_res(nn.Module): def __init__(self, inplanes, planes, stride=2): super(swin_res, self).__init__() self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, stride=stride, bias=False) self.bn1 = nn.BatchNorm2d(planes) self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False) self.bn2 = nn.BatchNorm2d(planes) self.conv3 = nn.Conv2d(planes, planes, kernel_size=1, bias=False) self.bn3 = nn.BatchNorm2d(planes) self.relu = nn.ReLU(inplace=True) self.downsample = nn.Sequential( nn.Conv2d(inplanes, planes, kernel_size=1, stride=1, bias=False), nn.BatchNorm2d(planes), ) self.stride = stride def forward(self, x): residual = x out = self.conv1(x) out = self.bn1(out) out = self.relu(out) out = self.conv2(out) out = self.bn2(out) out = self.relu(out) out = self.conv3(out) out = self.bn3(out) if self.downsample is not None: residual = self.downsample(x) out += residual out = self.relu(out) return out class Mlp(nn.Module): """ Multilayer perceptron.""" 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): """ 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): """ Forward function. 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 class SwinTransformerBlock(nn.Module): """ Swin Transformer Block. Args: dim (int): Number of input channels. 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, 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.num_heads = num_heads self.window_size = window_size self.shift_size = shift_size self.mlp_ratio = mlp_ratio 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) self.H = None self.W = None def forward(self, x, mask_matrix): """ Forward function. Args: x: Input feature, tensor size (B, H*W, C). H, W: Spatial resolution of the input feature. mask_matrix: Attention mask for cyclic shift. """ B, L, C = x.shape H, W = self.H, self.W assert L == H * W, "input feature has wrong size" shortcut = x x = self.norm1(x) x = x.view(B, H, W, C) # pad feature maps to multiples of window size pad_l = pad_t = 0 pad_r = (self.window_size - W % self.window_size) % self.window_size pad_b = (self.window_size - H % self.window_size) % self.window_size x = F.pad(x, (0, 0, pad_l, pad_r, pad_t, pad_b)) _, Hp, Wp, _ = x.shape # cyclic shift if self.shift_size > 0: shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) attn_mask = mask_matrix 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 attn_windows = self.attn(x_windows, mask=attn_mask) # nW*B, window_size*window_size, C # merge windows attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) shifted_x = window_reverse(attn_windows, self.window_size, Hp, Wp) # 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 if pad_r > 0 or pad_b > 0: x = x[:, :H, :W, :].contiguous() 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 class PatchMerging(nn.Module): """ Patch Merging Layer Args: dim (int): Number of input channels. norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, dim, norm_layer=nn.LayerNorm): super().__init__() self.dim = dim self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False) self.norm = norm_layer(4 * dim) def forward(self, x, H, W): """ Forward function. Args: x: Input feature, tensor size (B, H*W, C). H, W: Spatial resolution of the input feature. """ B, L, C = x.shape assert L == H * W, "input feature has wrong size" x = x.view(B, H, W, C) # padding pad_input = (H % 2 == 1) or (W % 2 == 1) if pad_input: x = F.pad(x, (0, 0, 0, W % 2, 0, H % 2)) 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 BasicLayer(nn.Module): """ A basic Swin Transformer layer for one stage. Args: dim (int): Number of feature channels depth (int): Depths of this stage. num_heads (int): Number of attention head. window_size (int): Local window size. Default: 7. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4. 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, depth, num_heads, window_size=7, 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.window_size = window_size self.shift_size = window_size // 2 self.depth = depth self.use_checkpoint = use_checkpoint # build blocks self.blocks = nn.ModuleList([ SwinTransformerBlock( dim=dim, 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(dim=dim, norm_layer=norm_layer) else: self.downsample = None def forward(self, x, H, W): """ Forward function. Args: x: Input feature, tensor size (B, H*W, C). H, W: Spatial resolution of the input feature. """ # calculate attention mask for SW-MSA Hp = int(np.ceil(H / self.window_size)) * self.window_size Wp = int(np.ceil(W / self.window_size)) * self.window_size img_mask = torch.zeros((1, Hp, Wp, 1), device=x.device) # 1 Hp Wp 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)) for blk in self.blocks: blk.H, blk.W = H, W if self.use_checkpoint: x = checkpoint.checkpoint(blk, x, attn_mask) else: x = blk(x, attn_mask) if self.downsample is not None: x_down = self.downsample(x, H, W) Wh, Ww = (H + 1) // 2, (W + 1) // 2 return x, H, W, x_down, Wh, Ww else: return x, H, W, x, H, W class PatchEmbed(nn.Module): """ Image to Patch Embedding Args: 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, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): super().__init__() patch_size = to_2tuple(patch_size) self.patch_size = patch_size 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): """Forward function.""" # padding _, _, H, W = x.size() if W % self.patch_size[1] != 0: x = F.pad(x, (0, self.patch_size[1] - W % self.patch_size[1])) if H % self.patch_size[0] != 0: x = F.pad(x, (0, 0, 0, self.patch_size[0] - H % self.patch_size[0])) x = self.proj(x) # B C Wh Ww if self.norm is not None: Wh, Ww = x.size(2), x.size(3) x = x.flatten(2).transpose(1, 2) x = self.norm(x) x = x.transpose(1, 2).view(-1, self.embed_dim, Wh, Ww) return x class SwinTransformer(nn.Module): """ Swin Transformer backbone. A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` - https://arxiv.org/pdf/2103.14030 Args: pretrain_img_size (int): Input image size for training the pretrained model, used in absolute postion embedding. Default 224. patch_size (int | tuple(int)): Patch size. Default: 4. in_chans (int): Number of input image channels. Default: 3. embed_dim (int): Number of linear projection output channels. Default: 96. depths (tuple[int]): Depths of each Swin Transformer stage. num_heads (tuple[int]): Number of attention head of each stage. 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. drop_rate (float): Dropout rate. attn_drop_rate (float): Attention dropout rate. Default: 0. drop_path_rate (float): Stochastic depth rate. Default: 0.2. 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. out_indices (Sequence[int]): Output from which stages. frozen_stages (int): Stages to be frozen (stop grad and set eval mode). -1 means not freezing any parameters. use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. """ def __init__(self, pretrain_img_size=224, patch_size=4, in_chans=3, embed_dim=96, depths=[2, 2, 6, 2], num_heads=[3, 6, 12, 24], window_size=7, mlp_ratio=4., qkv_bias=True, qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0.2, norm_layer=nn.LayerNorm, ape=False, patch_norm=True, out_indices=(0, 1, 2, 3), frozen_stages=-1, use_checkpoint=False): super().__init__() self.pretrain_img_size = pretrain_img_size self.num_layers = len(depths) self.embed_dim = embed_dim self.ape = ape self.patch_norm = patch_norm self.out_indices = out_indices self.frozen_stages = frozen_stages # split image into non-overlapping patches self.patch_embed = PatchEmbed( patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim, norm_layer=norm_layer if self.patch_norm else None) # absolute position embedding if self.ape: pretrain_img_size = to_2tuple(pretrain_img_size) patch_size = to_2tuple(patch_size) patches_resolution = [pretrain_img_size[0] // patch_size[0], pretrain_img_size[1] // patch_size[1]] self.absolute_pos_embed = nn.Parameter(torch.zeros(1, embed_dim, patches_resolution[0], patches_resolution[1])) 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 layers self.layers = nn.ModuleList() for i_layer in range(self.num_layers): layer = BasicLayer( dim=int(embed_dim * 2 ** i_layer), depth=depths[i_layer], num_heads=num_heads[i_layer], window_size=window_size, mlp_ratio=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])], norm_layer=norm_layer, downsample=PatchMerging if (i_layer < self.num_layers - 1) else None, use_checkpoint=use_checkpoint) self.layers.append(layer) num_features = [int(embed_dim * 2 ** i) for i in range(self.num_layers)] self.num_features = num_features self.out_shape = {'C3_size': num_features[-3], 'C4_size': num_features[-2], 'C5_size': num_features[-1]} # add a norm layer for each output for i_layer in out_indices: layer = norm_layer(num_features[i_layer]) layer_name = f'norm{i_layer}' self.add_module(layer_name, layer) self._freeze_stages() self.upsample = nn.Upsample(scale_factor=2,mode='nearest') self.new_block = swin_res(inplanes=384,planes=1024,stride=1) def _freeze_stages(self): if self.frozen_stages >= 0: self.patch_embed.eval() for param in self.patch_embed.parameters(): param.requires_grad = False if self.frozen_stages >= 1 and self.ape: self.absolute_pos_embed.requires_grad = False if self.frozen_stages >= 2: self.pos_drop.eval() for i in range(0, self.frozen_stages - 1): m = self.layers[i] m.eval() for param in m.parameters(): param.requires_grad = False def init_weights(self, pretrained=None): """Initialize the weights in backbone. Args: pretrained (str, optional): Path to pre-trained weights. Defaults to None. """ def _init_weights(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) if isinstance(pretrained, str): self.apply(_init_weights) elif pretrained is None: self.apply(_init_weights) else: raise TypeError('pretrained must be a str or None') def forward(self, x): """Forward function.""" x = self.patch_embed(x) Wh, Ww = x.size(2), x.size(3) if self.ape: # interpolate the position embedding to the corresponding size absolute_pos_embed = F.interpolate(self.absolute_pos_embed, size=(Wh, Ww), mode='bicubic') x = (x + absolute_pos_embed).flatten(2).transpose(1, 2) # B Wh*Ww C else: x = x.flatten(2).transpose(1, 2) x = self.pos_drop(x) outs = [] for i in range(self.num_layers-1): layer = self.layers[i] x_out, H, W, x, Wh, Ww = layer(x, Wh, Ww) norm_layer = getattr(self, f'norm{i}') x_out = norm_layer(x_out) out = x_out.view(-1, H, W, self.num_features[i]).permute(0, 3, 1, 2).contiguous() outs.append(out) out = self.new_block(out) return out # if i in self.out_indices: # norm_layer = getattr(self, f'norm{i}') # x_out = norm_layer(x_out) # out = x_out.view(-1, H, W, self.num_features[i]).permute(0, 3, 1, 2).contiguous() # outs.append(out) # return outs[1:] def train(self, mode=True): """Convert the model into training mode while keep layers freezed.""" super(SwinTransformer, self).train(mode) self._freeze_stages() # def Swin_T(): # model = SwinTransformer() # # classifier = list([model.layer4, model.avgpool]) # # features = nn.Sequential(*features) # classifier = nn.Sequential(*classifier) # return features, classifier if __name__ == '__main__': input = torch.randn(2, 3, 800, 800) M = SwinTransformer() output=M(input) print(output)
