Vision Transformer理论＋PyTorch代码全解析（附图解）

一、前言

        Transformer最初提出是针对NLP领域的，并且在NLP领域大获成功。这篇论文也是受到其启发，尝试将Transformer应用到CV领域。关于Transformer的部分理论之前的博文中有讲，链接，这里不在赘述。通过这篇文章的实验，给出的最佳模型在ImageNet1K上能够达到88.55%的准确率（先在Google自家的JFT数据集上进行了预训练），说明Transformer在CV领域确实是有效的，而且效果还挺惊人。（下图为原论文中的效果图）

二、模型详解 

1、ViT的整体框架图

        Vision Transformer的模型结构相比于Transformer来说更简单，在Transformer模型中，主要包含Encoder和Decoder结构，而ViT(Vision Transformer)仅借鉴了Encoder结构。

简单而言，模型由三个模块组成：

        1）Linear Projection of Flattened Patches(Embedding层)
        2）Transformer Encoder(图右侧有给出更加详细的结构)
        3）MLP Head（最终用于分类的层结构)

2、Embedding层讲解 

        对于标准的Transformer模块，要求输入的是token（向量）序列，即二维矩阵[num_token, token_dim]，如下图，token0-9对应的都是向量，以ViT-B/16为例，每个token向量长度为768。

         对于图像数据而言，其数据格式为[H, W, C]是三维矩阵明显不是Transformer想要的。所以需要先通过一个Embedding层来对数据做个变换。如下图所示，首先将一张图片按给定大小分成一堆Patches。以ViT-B/16为例，将输入图片(224x224)按照16x16大小的Patch进行划分，划分后会得到 (224/16)^2=196个Patches。接着通过线性映射将每个Patch映射到一维向量中，以ViT-B/16为例，每个Patche数据shape为[16, 16, 3]通过映射得到一个长度为768的向量（后面都直接称为token）。

        在代码实现中，直接通过一个卷积层来实现。 以ViT-B/16为例，直接使用一个卷积核大小为16x16，步距为16，卷积核个数为768的卷积来实现。通过卷积[224, 224, 3] -> [14, 14, 768]，然后把H以及W两个维度展平即可[14, 14, 768] -> [196, 768]，此时正好变成了一个二维矩阵，正是Transformer想要的。下面是卷积层计算公式：

输入为[224,244,3]，经过卷积层变成[14,14,768]，再映射为[196,768]。这样，就完成了从图片到Token之间的转换。

        在上面的结构图中可以看到，输入Encoder的最左侧部分添加了一个0*这个Token，这个就是额外添加的一个[class]token，单独用来处理类别信息，经过Encoder之后，需要单独将这个Token再提取出来，输入到MLP Head之中再输出分类结果。这也是为什么结构图中MLP Head的位置是和这个[class]token对齐。

        在Transformer中，位置编码的作用是为了记忆输入的语序信息。ViT中，同样需要位置编码来记录各图像块之间的位置信息。关于Position Embedding就是之前Transformer中讲到的Positional Encoding，这里的Position Embedding采用的是一个可训练的参数（1D Pos. Emb.），是直接叠加在tokens上的（add），所以shape要一样。以ViT-B/16为例，刚刚拼接[class]token后shape是[197, 768]，那么这里的Position Embedding的shape也是[197, 768]。

3、 Transformer Encoder

        ViT虽然采用的是Transformer Encoder的结构，但是和Transformer原始的Encoder还是有所区别，我将两者的结构进行对比，如下图所示，左侧为Transformer原始的Encoder结构。

        Transformer Encoder其实就是重复堆叠Encoder Block L次，下图是引用博主太阳花的小绿豆[1]绘制的Encoder Block，主要由以下几部分组成：

        1）Layer Normalization，这种Normalization方法主要是针对NLP领域提出的，这里是对每个token进行Norm处理，在Transformer论文中有很详细的讲解
        2）Multi-Head Attention，这个结构在Transformer论文中有很详细的讲解。

        3）Dropout/DropPath，在原论文的代码中是直接使用的Dropout层，在但rwightman实现的代码中使用的是DropPath（stochastic depth），可能后者会更好一点。
        4）MLP Block，如下图右侧所示，就是全连接+GELU激活函数+Dropout组成也非常简单，需要注意的是第一个全连接层会把输入节点个数翻4倍[197, 768] -> [197, 3072]，第二个全连接层会还原回原节点个数[197, 3072] -> [197, 768]。

4、MIP Head详解

在Transformer Encoder输出结果之后，需要再将第一个添加的Class Token提取出来，然后输入到MLP Head进行分类。在论文中，作者先是在ImageNet21K上进行预训练，MLP Head结构由Linear+tanh激活函数+Linear组成，但是迁移到其它数据集训练时，只需要用一个一个Linear即可。

5、基于ViT-B/16的Vision Transformer网络结构图

三、实践测试

        下面就来尝试使用ViT做一个简单的分类任务。官方使用的是Tensorflow2框架，下面我使用的是别人利用Pytorch实现的ViT。

代码来源于(包含数据集和权重下载地址)deep-learning-for-image-processing:deep learning for image processing including classification and object-detection etc. - GitCode

实验采用的是花蕊数据集，共5个类别，约4000多个样本。

1、图像预处理

包括训练集上的随机裁剪、水平翻转；验证集上的中心裁剪

data_transform = {
  "train": transforms.Compose([transforms.RandomResizedCrop(224),
                               transforms.RandomHorizontalFlip(),
                               transforms.ToTensor(),
                               transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])]),
  "val": transforms.Compose([transforms.Resize(256),
                             transforms.CenterCrop(224),
                             transforms.ToTensor(),
                             transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])])}

 2、图像分割

通过768个16*16步长为16的卷积核将图像转换成一个个Patch

class PatchEmbed(nn.Module):
    """
    2D Image to Patch Embedding
    """
    def __init__(self, img_size=224, patch_size=16, in_c=3, embed_dim=768, norm_layer=None):
        super().__init__()
        img_size = (img_size, img_size)
        patch_size = (patch_size, patch_size)
        self.img_size = img_size
        self.patch_size = patch_size
        self.grid_size = (img_size[0] // patch_size[0], img_size[1] // patch_size[1])
        self.num_patches = self.grid_size[0] * self.grid_size[1]

        self.proj = nn.Conv2d(in_c, embed_dim, kernel_size=patch_size, stride=patch_size)
        self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()

    def forward(self, x):
        B, C, H, W = x.shape
        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]})."

        # flatten: [B, C, H, W] -> [B, C, HW]
        # transpose: [B, C, HW] -> [B, HW, C]
        x = self.proj(x).flatten(2).transpose(1, 2)
        x = self.norm(x)
        return x

3、 多头注意力

多头注意力模块实现时注意是通过一个线性层来同时计算qkv三个矩阵，这样可以提升计算效率。

class Attention(nn.Module):
    def __init__(self,
                 dim,   # 输入token的dim
                 num_heads=8,
                 qkv_bias=False,
                 qk_scale=None,
                 attn_drop_ratio=0.,
                 proj_drop_ratio=0.):
        super(Attention, self).__init__()
        self.num_heads = num_heads
        head_dim = dim // num_heads
        self.scale = qk_scale or head_dim ** -0.5
        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop_ratio)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop_ratio)

    def forward(self, x):
        # [batch_size, num_patches + 1, total_embed_dim]
        B, N, C = x.shape

        # qkv(): -> [batch_size, num_patches + 1, 3 * total_embed_dim]
        # reshape: -> [batch_size, num_patches + 1, 3, num_heads, embed_dim_per_head]
        # permute: -> [3, batch_size, num_heads, num_patches + 1, embed_dim_per_head]
        qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
        # [batch_size, num_heads, num_patches + 1, embed_dim_per_head]
        q, k, v = qkv[0], qkv[1], qkv[2]  # make torchscript happy (cannot use tensor as tuple)

        # transpose: -> [batch_size, num_heads, embed_dim_per_head, num_patches + 1]
        # @: multiply -> [batch_size, num_heads, num_patches + 1, num_patches + 1]
        attn = (q @ k.transpose(-2, -1)) * self.scale
        attn = attn.softmax(dim=-1)
        attn = self.attn_drop(attn)

        # @: multiply -> [batch_size, num_heads, num_patches + 1, embed_dim_per_head]
        # transpose: -> [batch_size, num_patches + 1, num_heads, embed_dim_per_head]
        # reshape: -> [batch_size, num_patches + 1, total_embed_dim]
        x = (attn @ v).transpose(1, 2).reshape(B, N, C)
        x = self.proj(x)
        x = self.proj_drop(x)
        return x

 4、MLP Block 

class Mlp(nn.Module):
    """
    MLP as used in Vision Transformer, MLP-Mixer and related networks
    """
    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


class Block(nn.Module):
    def __init__(self,
                 dim,
                 num_heads,
                 mlp_ratio=4.,
                 qkv_bias=False,
                 qk_scale=None,
                 drop_ratio=0.,
                 attn_drop_ratio=0.,
                 drop_path_ratio=0.,
                 act_layer=nn.GELU,
                 norm_layer=nn.LayerNorm):
        super(Block, self).__init__()
        self.norm1 = norm_layer(dim)
        self.attn = Attention(dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale,
                              attn_drop_ratio=attn_drop_ratio, proj_drop_ratio=drop_ratio)
        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
        self.drop_path = DropPath(drop_path_ratio) if drop_path_ratio > 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_ratio)

    def forward(self, x):
        x = x + self.drop_path(self.attn(self.norm1(x)))
        x = x + self.drop_path(self.mlp(self.norm2(x)))
        return x

5、Vision Transformer

将上面各部分结合起来，得到VisionTransformer整体结构，可以和上面的结构图对比

class VisionTransformer(nn.Module):
    def __init__(self, img_size=224, patch_size=16, in_c=3, num_classes=1000,
                 embed_dim=768, depth=12, num_heads=12, mlp_ratio=4.0, qkv_bias=True,
                 qk_scale=None, representation_size=None, distilled=False, drop_ratio=0.,
                 attn_drop_ratio=0., drop_path_ratio=0., embed_layer=PatchEmbed, norm_layer=None,
                 act_layer=None):
        """
        Args:
            img_size (int, tuple): input image size
            patch_size (int, tuple): patch size
            in_c (int): number of input channels
            num_classes (int): number of classes for classification head
            embed_dim (int): embedding dimension
            depth (int): depth of transformer
            num_heads (int): number of attention heads
            mlp_ratio (int): ratio of mlp hidden dim to embedding dim
            qkv_bias (bool): enable bias for qkv if True
            qk_scale (float): override default qk scale of head_dim ** -0.5 if set
            representation_size (Optional[int]): enable and set representation layer (pre-logits) to this value if set
            distilled (bool): model includes a distillation token and head as in DeiT models
            drop_ratio (float): dropout rate
            attn_drop_ratio (float): attention dropout rate
            drop_path_ratio (float): stochastic depth rate
            embed_layer (nn.Module): patch embedding layer
            norm_layer: (nn.Module): normalization layer
        """
        super(VisionTransformer, self).__init__()
        self.num_classes = num_classes
        self.num_features = self.embed_dim = embed_dim  # num_features for consistency with other models
        self.num_tokens = 2 if distilled else 1
        norm_layer = norm_layer or partial(nn.LayerNorm, eps=1e-6)
        act_layer = act_layer or nn.GELU

        self.patch_embed = embed_layer(img_size=img_size, patch_size=patch_size, in_c=in_c, embed_dim=embed_dim)
        num_patches = self.patch_embed.num_patches

        self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
        self.dist_token = nn.Parameter(torch.zeros(1, 1, embed_dim)) if distilled else None
        self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + self.num_tokens, embed_dim))
        self.pos_drop = nn.Dropout(p=drop_ratio)

        dpr = [x.item() for x in torch.linspace(0, drop_path_ratio, depth)]  # stochastic depth decay rule
        self.blocks = nn.Sequential(*[
            Block(dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale,
                  drop_ratio=drop_ratio, attn_drop_ratio=attn_drop_ratio, drop_path_ratio=dpr[i],
                  norm_layer=norm_layer, act_layer=act_layer)
            for i in range(depth)
        ])
        self.norm = norm_layer(embed_dim)

        # Representation layer
        if representation_size and not distilled:
            self.has_logits = True
            self.num_features = representation_size
            self.pre_logits = nn.Sequential(OrderedDict([
                ("fc", nn.Linear(embed_dim, representation_size)),
                ("act", nn.Tanh())
            ]))
        else:
            self.has_logits = False
            self.pre_logits = nn.Identity()

        # Classifier head(s)
        self.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity()
        self.head_dist = None
        if distilled:
            self.head_dist = nn.Linear(self.embed_dim, self.num_classes) if num_classes > 0 else nn.Identity()

        # Weight init
        nn.init.trunc_normal_(self.pos_embed, std=0.02)
        if self.dist_token is not None:
            nn.init.trunc_normal_(self.dist_token, std=0.02)

        nn.init.trunc_normal_(self.cls_token, std=0.02)
        self.apply(_init_vit_weights)

    def forward_features(self, x):
        # [B, C, H, W] -> [B, num_patches, embed_dim]
        x = self.patch_embed(x)  # [B, 196, 768]
        # [1, 1, 768] -> [B, 1, 768]
        cls_token = self.cls_token.expand(x.shape[0], -1, -1)
        if self.dist_token is None:
            x = torch.cat((cls_token, x), dim=1)  # [B, 197, 768]
        else:
            x = torch.cat((cls_token, self.dist_token.expand(x.shape[0], -1, -1), x), dim=1)

        x = self.pos_drop(x + self.pos_embed)
        x = self.blocks(x)
        x = self.norm(x)
        if self.dist_token is None:
            return self.pre_logits(x[:, 0])
        else:
            return x[:, 0], x[:, 1]

    def forward(self, x):
        x = self.forward_features(x)
        if self.head_dist is not None:
            x, x_dist = self.head(x[0]), self.head_dist(x[1])
            if self.training and not torch.jit.is_scripting():
                # during inference, return the average of both classifier predictions
                return x, x_dist
            else:
                return (x + x_dist) / 2
        else:
            x = self.head(x)
        return x

 我使用vit_base_patch16_224_in21k这个模型，训练了10个epoch，验证集上准确率达到了98.5%。
整体模型还是比较大的，预训练权重大小为393MB，但是训练速度还是挺快的，因为在代码中有个冻结权重的操作，主干部分全部冻结，仅训练分类头。

6、完整代码＋权重

vision_transformer_免费高速下载|百度网盘-分享无限制 (baidu.com)

四、References

[1]https://blog.csdn.net/qq_37541097/article/details/118242600