Transformer实战-系列教程10:SwinTransformer 源码解读3(SwinTransformerBlock类)

Transformer实战-系列教程总目录

有任何问题欢迎在下面留言
本篇文章的代码运行界面均在Pycharm中进行
本篇文章配套的代码资源已经上传
点我下载源码

SwinTransformer 算法原理
SwinTransformer 源码解读1(项目配置/SwinTransformer类)
SwinTransformer 源码解读2(PatchEmbed类/BasicLayer类)
SwinTransformer 源码解读3(SwinTransformerBlock类)
SwinTransformer 源码解读4(WindowAttention类)
SwinTransformer 源码解读5(Mlp类/PatchMerging类)

5、SwinTransformerBlock类

class SwinTransformerBlock(nn.Module):
    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}"

5.1 构造函数

SwinTransformerBlock 是 Swin Transformer 模型中的一个基本构建块。它结合了自注意力机制和多层感知机(MLP),并通过窗口划分和可选的窗口位移来实现局部注意力

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:
            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:
            H, W = self.input_resolution
            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)
            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))
        else:
            attn_mask = None
        self.register_buffer("attn_mask", attn_mask)
  1. dim:输入特征的通道数。
  2. input_resolution:输入特征的分辨率(高度和宽度)
  3. num_heads:自注意力头的数量
  4. window_size:窗口大小,决定了注意力机制的局部范围
  5. shift_size:窗口位移的大小,用于实现错位窗口多头自注意力(SW-MSA)
  6. mlp_ratio:MLP隐层大小与输入通道数的比率
  7. qkv_bias:QKV的偏置
  8. qk_scale:QKV的缩放因子
  9. drop:丢弃率
  10. drop_path:分别控制QKV的偏差、缩放因子、丢弃率、注意力丢弃率和随机深度率
  11. norm_layer:激活层和标准化层,默认分别为 GELU 和 LayerNorm
  12. WindowAttention:窗口注意力模块
  13. Mlp:一个包含全连接层、激活函数、Dropout的模块
  14. img_mask :图像掩码,用于生成错位窗口自注意力
  15. h_slicesw_slices:水平和垂直方向上的切片,用于划分图像掩码
  16. cnt :计数器,标记不同的窗口
  17. mask_windows :图像掩码划分为窗口,并将每个窗口的掩码重塑为一维向量
  18. window_partition
  19. attn_mask :注意力掩码,用于在自注意力计算中排除窗口外的位置
  20. register_buffer:注意力掩码注册为一个模型的缓冲区

5.2 前向传播

def forward(self, x):
        H, W = self.input_resolution
        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)
        if self.shift_size > 0:
            shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))
        else:
            shifted_x = x
        x_windows = window_partition(shifted_x, self.window_size)
        x_windows = x_windows.view(-1, self.window_size * self.window_size, C)
        attn_windows = self.attn(x_windows, mask=self.attn_mask)  # nW*B, window_size*window_size, C
        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
        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(x)
        x = x + self.drop_path(self.mlp(self.norm2(x)))
        return x
  1. 原始输入: torch.Size([4, 3136, 96]),输入的是一个长度为3136的序列,每个向量的维度为96,在
    被多次调用的时候,维度也发生了变化原始输入: torch.Size([4, 784, 192])、torch.Size([4, 196, 384])、torch.Size([4, 49, 768])
  2. H,W=[ 56,56],输入分辨率中的高度和宽度
  3. B, L, C=[ 4,3136,96],当前输入的维度,批次大小、序列长度和向量的维度
  4. shortcut是用于残差
  5. norm1(x): torch.Size([4, 3136, 96]),经过一个层归一化,维度不变
  6. x.view(B, H, W, C): torch.Size([4, 56, 56, 96]),将序列重塑为(Batch_size,Height,Width,Channel)的形状
  7. shift_size用来判断是否做偏移,最开始的时候shift_size为0,window_partition用来做windows的划分,特征图为[56,56,96],默认窗口为77的大小,因为可以划分出88个窗口
  8. shifted_x: torch.Size([4, 56, 56, 96]),位移操作后的x
  9. x_windows: torch.Size([256, 7, 7, 96]),使用窗口划分函数,划分256个窗口,每个窗口7*7个特征,每个特征96维向量
  10. x_windows: torch.Size([256, 49, 96]),将窗口重塑为一维向量,以便进行自注意力计算
  11. attn_windows: torch.Size([256, 7, 7, 96]),WindowAttention处理x,对每个窗口应用窗口注意力机制,考虑到可能的注意力掩码
  12. shifted_x: torch.Size([4, 56, 56, 96]),注意力操作后的窗口重塑回原始形状,并将它们合并回完整的特征图
  13. torch.Size([4, 56, 56, 96]),如果进行了循环位移,则执行逆向循环位移操作,以恢复原始特征图的位置
  14. torch.Size([4, 3136, 96]),特征图重塑回原始的[B, L, C]形状
  15. torch.Size([4, 3136, 96]),应用残差连接,并通过随机深度(如果设置了的话)
  16. torch.Size([4, 3136, 96]),应用第二个标准化层,然后是MLP,并再次应用随机深度,完成残差连接的最后一步。

5.3 窗口划分函数window_partition()

def window_partition(x, window_size):
    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
  1. B, H, W, C =torch.Size([4, 56, 56, 96]),这是函数的原始输入,56*56等于原来的3136
  2. x = torch.Size([4, 8, 7, 8, 7, 96])
  3. windows = torch.Size([256, 7, 7, 96]),256的计算方法为原始数据为5656,要分出77大小为一个窗口,所以窗口数量为(56/7)(56/7)=88=64,而64需要乘上batch=4,所以就是256,向量维度96不变

而Windows Attention就是在这7*7里面的窗口进行计算

SwinTransformer 算法原理
SwinTransformer 源码解读1(项目配置/SwinTransformer类)
SwinTransformer 源码解读2(PatchEmbed类/BasicLayer类)
SwinTransformer 源码解读3(SwinTransformerBlock类)
SwinTransformer 源码解读4(WindowAttention类)
SwinTransformer 源码解读5(Mlp类/PatchMerging类)

你可能感兴趣的:(Transformer实战,transformer,深度学习,人工智能,计算机视觉,self-Attention,pytroch,SwinTransformer)