@[toc]
首先输入图像:224*224*3(L*W*C)
Patch Partition 是将图像打成patch,然后再拉平。 patch size为4,即分成4*4大小的块,图像长和宽变为H/4,通道数变为48(4*4*3)。 Linear Embedding即使用一个线性变换将图像变换到任意维度C,不同的模型的C有所不同。 在网络结构中,只有第一个stage是 Linear Embedding,其余都是Patch Merging。在实现代码中 Patch Partition 与 Linear Embedding是在一起的。
每个stage都重复偶数次swin transformer Block。
注意:第一个和第二个swin transformer Block有所不同 如图。
两次swin transformer Block结构如下图所示:
首先第一个block经过一个线性层,然后经过窗口的多头自注意力机制模块(W-MSA),然后再做一次残差连接,之后经过线性层,然后经过多层神经网络进入第二个block,第二个block与第一个block唯一不同的就是 线性层之后变成了 移动窗口的多头自注意力机制模块(SW-MSA)。
使用W-MSA的目的是什么呢?减少计算量!
要了解W-MSA首先要了解首先是多头自注意力机制(Multi-head Self-Attention)。
多头自注意力机制要对每一个patch求Q,K,V;然后对于每一个patch所求的Q会对特征图中的每一个K进行匹配。所有每一个patch会和特征图中所有的patch进行沟通。
而下面W-MSA即为将一个特征图分为多个window,然后每个window进行MSA,而各个window之间不进行信息交互。所以就有了W-MSA的缺点,各个窗口之间无法进行信息交互,感受野太小。
引入SW-MSA的目的即为解决W-MSA的缺点,各个window中无法进行信息交互。
以上图图为例,特征图向右向下移动了两个patch,划分成了9个window,然后在每个window中计算时,就会融合之间不同window的信息。
将上图简化为下图为:
例如:计算编号为2的window的MSA则会融合两个window的信息,计算编号为5的window的MSA则会融合四个window的信息。
但是这样需要计算9个window的MSA 效率并不太高,所以作者继续改进,如下图所示,将最上面的编号1 2 3移到下面,然后再将最左边的4 7 1移到最右边就形成了下图,然后将6,4合成一块,8,2合成一块,9 ,7 ,3 ,1合成一块,而5不动。然后对这4个合成window进行WSA,但是不同的分开的区域(如 6,4)他们本身不在一块,所以不能简单粗暴的计算,需要分开计算。
如计算编号为6的MSA的时候,将区域4的权重设为很小的数,然后进行softmax这些数就变为了0,所以进行计算编号6的MSA时,其实就与编号4无关了。
假设特征矩阵4*4,单通道。2*2大小为一个窗口。颜色相同表示patch的位置是相同的。 然后将相同位置中的像素提取出来即为图中第二步。 然后经过Concat在深度方向进行拼接。 然后在深度方向进行LayerNorm 归一化处理 然后经过全连接层将每一个像素线性方向进行映射。 经过Patch Merging图像尺寸变为原来的一半,通道数翻倍
SwinTransformer(
(patch_embed): PatchEmbed(
(proj): Conv2d(3, 96, kernel_size=(4, 4), stride=(4, 4))
(norm): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
)
(pos_drop): Dropout(p=0.0, inplace=False)
(layers): ModuleList(
(0): 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()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=384, out_features=96, bias=True)
(drop2): 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()
(norm2): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=96, out_features=384, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=384, out_features=96, bias=True)
(drop2): 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)
)
)
(1): BasicLayer(
(blocks): ModuleList(
(0): SwinTransformerBlock(
(norm1): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=192, out_features=576, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=192, out_features=192, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=192, out_features=768, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=768, out_features=192, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(1): SwinTransformerBlock(
(norm1): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=192, out_features=576, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=192, out_features=192, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=192, out_features=768, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=768, out_features=192, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
)
(downsample): PatchMerging(
(reduction): Linear(in_features=768, out_features=384, bias=False)
(norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
)
)
(2): BasicLayer(
(blocks): ModuleList(
(0): SwinTransformerBlock(
(norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=384, out_features=1152, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=384, out_features=384, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=384, out_features=1536, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=1536, out_features=384, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(1): SwinTransformerBlock(
(norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=384, out_features=1152, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=384, out_features=384, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=384, out_features=1536, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=1536, out_features=384, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(2): SwinTransformerBlock(
(norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=384, out_features=1152, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=384, out_features=384, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=384, out_features=1536, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=1536, out_features=384, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(3): SwinTransformerBlock(
(norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=384, out_features=1152, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=384, out_features=384, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=384, out_features=1536, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=1536, out_features=384, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(4): SwinTransformerBlock(
(norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=384, out_features=1152, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=384, out_features=384, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=384, out_features=1536, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=1536, out_features=384, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(5): SwinTransformerBlock(
(norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=384, out_features=1152, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=384, out_features=384, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=384, out_features=1536, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=1536, out_features=384, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
)
(downsample): PatchMerging(
(reduction): Linear(in_features=1536, out_features=768, bias=False)
(norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
)
)
(3): BasicLayer(
(blocks): ModuleList(
(0): SwinTransformerBlock(
(norm1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=768, out_features=2304, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=768, out_features=768, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=768, out_features=3072, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=3072, out_features=768, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
(1): SwinTransformerBlock(
(norm1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
(attn): WindowAttention(
(qkv): Linear(in_features=768, out_features=2304, bias=True)
(attn_drop): Dropout(p=0.0, inplace=False)
(proj): Linear(in_features=768, out_features=768, bias=True)
(proj_drop): Dropout(p=0.0, inplace=False)
(softmax): Softmax(dim=-1)
)
(drop_path): DropPath()
(norm2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
(mlp): Mlp(
(fc1): Linear(in_features=768, out_features=3072, bias=True)
(act): GELU()
(drop1): Dropout(p=0.0, inplace=False)
(fc2): Linear(in_features=3072, out_features=768, bias=True)
(drop2): Dropout(p=0.0, inplace=False)
)
)
)
)
)
(norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
(avgpool): AdaptiveAvgPool1d(output_size=1)
(head): Linear(in_features=768, out_features=10, bias=True)
)
cnn的灵感主要来源于resnet残差网络。
Cnn(
(init_classifier): Sequential(
(0): Conv2d(3, 64, kernel_size=(3, 3), stride=(2, 2), padding=(2, 2))
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU(inplace=True)
(3): MaxPool2d(kernel_size=3, stride=2, padding=0, dilation=1, ceil_mode=False)
)
(cnnBlock64d): Sequential(
(0): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU(inplace=True)
(3): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
)
(cnnBlock_128d): Sequential(
(0): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1))
(1): ReLU(inplace=True)
(2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
)
(cnnBlock_u128d): Sequential(
(0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2))
)
(layer1): Sequential(
(0): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1))
(1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), padding=(1, 1))
(3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(4): Conv2d(512, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(5): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(6): ReLU(inplace=True)
)
(cnnBlock_u2048d): Sequential(
(0): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))
(1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(4): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1))
(5): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(6): ReLU(inplace=True)
)
(cnnBlock1d): Sequential(
(0): Conv2d(1024, 2048, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(2): ReLU(inplace=True)
)
(cnnBlock_l768): Sequential(
(0): Conv2d(2048, 256, kernel_size=(1, 1), stride=(1, 1))
(1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(cnnBlock_r768): Sequential(
(0): Conv2d(2048, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(cnnBlock): Conv2d(256, 768, kernel_size=(1, 1), stride=(1, 1), padding=(1, 1))
(avgpool): AvgPool2d(kernel_size=3, stride=2, padding=0)
(relu): ReLU(inplace=True)
)
图像经过cnn,最终变为batch_size*7*7*768。
而swin transformer经过第三个stage后,图像变为了 batch_size*768*49, 只有三维的原因是:图像经过swin transformer的第一层即 patch partition和linear embedding 时,图像进行了展平,在第2维进行展平,图像从 batch_size*3*224*224变成了 batch_size*3*(224*224),然后交换第1维和第2维,变成了 batch_size*(224*224)*3. 代码如下: pytorch实现
flatten: [B, C, H, W] -> [B, C, HW]
transpose: [B, C, HW] -> [B, HW, C]
x = x.flatten(2).transpose(1, 2)
两者的维度和尺寸相同,所以才能进行相加,然后进行一系列归一、池化、全连接等操作进行特征输出。
""" Swin Transformer
A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows`
- https://arxiv.org/pdf/2103.14030
Code/weights from https://github.com/microsoft/Swin-Transformer
"""
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.checkpoint as checkpoint
import numpy as np
from typing import Optional
from cnn import Cnn
def drop_path_f(x, drop_prob: float = 0., training: bool = False):
"""Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
This is the same as the DropConnect impl I created for EfficientNet, etc networks, however,
the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for
changing the layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use
'survival rate' as the argument.
"""
if drop_prob == 0. or not training:
return x
keep_prob = 1 - drop_prob
shape = (x.shape[0],) + (1,) * (x.ndim - 1) # work with diff dim tensors, not just 2D ConvNets
random_tensor = keep_prob + torch.rand(shape, dtype=x.dtype, device=x.device)
random_tensor.floor_() # binarize
output = x.div(keep_prob) * random_tensor
return output
class DropPath(nn.Module):
"""Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
"""
def __init__(self, drop_prob=None):
super(DropPath, self).__init__()
self.drop_prob = drop_prob
def forward(self, x):
return drop_path_f(x, self.drop_prob, self.training)
def window_partition(x, window_size: int):
"""
Args:
x: (B, H, W, C)
window_size (int): window size(M)
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)
# permute: [B, H//Mh, Mh, W//Mw, Mw, C] -> [B, H//Mh, W//Mh, Mw, Mw, C]
# view: [B, H//Mh, W//Mw, Mh, Mw, C] -> [B*num_windows, Mh, Mw, 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: int, H: int, W: int):
"""
Args:
windows: (num_windows*B, window_size, window_size, C)
window_size (int): Window size(M)
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))
# view: [B*num_windows, Mh, Mw, C] -> [B, H//Mh, W//Mw, Mh, Mw, C]
x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)
# permute: [B, H//Mh, W//Mw, Mh, Mw, C] -> [B, H//Mh, Mh, W//Mw, Mw, C]
# view: [B, H//Mh, Mh, W//Mw, Mw, C] -> [B, H, W, C]
x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
return x
class PatchEmbed(nn.Module):
"""
2D Image to Patch Embedding
"""
def __init__(self, patch_size=4, in_c=3, embed_dim=96, norm_layer=None):
super().__init__()
patch_size = (patch_size, patch_size)
self.patch_size = patch_size
self.in_chans = in_c
self.embed_dim = embed_dim
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):
_, _, H, W = x.shape
# padding
pad_input = (H % self.patch_size[0] != 0) or (W % self.patch_size[1] != 0)
if pad_input:
# to pad the last 3 dimensions,
# (W_left, W_right, H_top,H_bottom, C_front, C_back)
x = F.pad(x, (0, self.patch_size[1] - W % self.patch_size[1],
0, self.patch_size[0] - H % self.patch_size[0],
0, 0))
x = self.proj(x)
_, _, H, W = x.shape
# flatten: [B, C, H, W] -> [B, C, HW]
# transpose: [B, C, HW] -> [B, HW, C]
x = x.flatten(2).transpose(1, 2)
x = self.norm(x)
return x, H, W
class PatchMerging(nn.Module):
r""" 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):
"""
x: B, H*W, C
"""
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:
# to pad the last 3 dimensions, starting from the last dimension and moving forward.
# (C_front, C_back, W_left, W_right, H_top, H_bottom)
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) # [B, H/2*W/2, 2*C]
return x
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.drop1 = nn.Dropout(drop)
self.fc2 = nn.Linear(hidden_features, out_features)
self.drop2 = nn.Dropout(drop)
def forward(self, x):
x = self.fc1(x)
x = self.act(x)
x = self.drop1(x)
x = self.fc2(x)
x = self.drop2(x)
return x
class WindowAttention(nn.Module):
r""" Window based multi-head self attention (W-MSA) module with relative position bias.
It supports both of shifted and non-shifted window.
Args:
dim (int): Number of input channels.
window_size (tuple[int]): The height and width of the window.
num_heads (int): Number of attention heads.
qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0
proj_drop (float, optional): Dropout ratio of output. Default: 0.0
"""
def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0.):
super().__init__()
self.dim = dim
self.window_size = window_size # [Mh, Mw]
self.num_heads = num_heads
head_dim = dim // num_heads
self.scale = 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*Mh-1 * 2*Mw-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], indexing="ij")) # [2, Mh, Mw]
coords_flatten = torch.flatten(coords, 1) # [2, Mh*Mw]
# [2, Mh*Mw, 1] - [2, 1, Mh*Mw]
relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # [2, Mh*Mw, Mh*Mw]
relative_coords = relative_coords.permute(1, 2, 0).contiguous() # [Mh*Mw, Mh*Mw, 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) # [Mh*Mw, Mh*Mw]
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)
nn.init.trunc_normal_(self.relative_position_bias_table, std=.02)
self.softmax = nn.Softmax(dim=-1)
def forward(self, x, mask: Optional[torch.Tensor] = None):
"""
Args:
x: input features with shape of (num_windows*B, Mh*Mw, C)
mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None
"""
# [batch_size*num_windows, Mh*Mw, total_embed_dim]
B_, N, C = x.shape
# qkv(): -> [batch_size*num_windows, Mh*Mw, 3 * total_embed_dim]
# reshape: -> [batch_size*num_windows, Mh*Mw, 3, num_heads, embed_dim_per_head]
# permute: -> [3, batch_size*num_windows, num_heads, Mh*Mw, 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_windows, num_heads, Mh*Mw, embed_dim_per_head]
q, k, v = qkv.unbind(0) # make torchscript happy (cannot use tensor as tuple)
# transpose: -> [batch_size*num_windows, num_heads, embed_dim_per_head, Mh*Mw]
# @: multiply -> [batch_size*num_windows, num_heads, Mh*Mw, Mh*Mw]
q = q * self.scale
attn = (q @ k.transpose(-2, -1))
# relative_position_bias_table.view: [Mh*Mw*Mh*Mw,nH] -> [Mh*Mw,Mh*Mw,nH]
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)
relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # [nH, Mh*Mw, Mh*Mw]
attn = attn + relative_position_bias.unsqueeze(0)
if mask is not None:
# mask: [nW, Mh*Mw, Mh*Mw]
nW = mask.shape[0] # num_windows
# attn.view: [batch_size, num_windows, num_heads, Mh*Mw, Mh*Mw]
# mask.unsqueeze: [1, nW, 1, Mh*Mw, Mh*Mw]
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)
# @: multiply -> [batch_size*num_windows, num_heads, Mh*Mw, embed_dim_per_head]
# transpose: -> [batch_size*num_windows, Mh*Mw, num_heads, embed_dim_per_head]
# reshape: -> [batch_size*num_windows, Mh*Mw, total_embed_dim]
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):
r""" 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
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, 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=(self.window_size, self.window_size), num_heads=num_heads, qkv_bias=qkv_bias,
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)
def forward(self, x, attn_mask):
H, W = self.H, self.W
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)
# 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))
else:
shifted_x = x
attn_mask = None
# partition windows
x_windows = window_partition(shifted_x, self.window_size) # [nW*B, Mh, Mw, C]
x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # [nW*B, Mh*Mw, C]
# W-MSA/SW-MSA
attn_windows = self.attn(x_windows, mask=attn_mask) # [nW*B, Mh*Mw, C]
# merge windows
attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) # [nW*B, Mh, Mw, 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 BasicLayer(nn.Module):
"""
A basic Swin Transformer layer for one stage.
Args:
dim (int): Number of input channels.
depth (int): Number of blocks.
num_heads (int): Number of attention heads.
window_size (int): Local window size.
mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
drop (float, optional): Dropout rate. Default: 0.0
attn_drop (float, optional): Attention dropout rate. Default: 0.0
drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
"""
def __init__(self, dim, depth, num_heads, window_size,
mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0.,
drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False):
super().__init__()
self.dim = dim
self.depth = depth
self.window_size = window_size
self.use_checkpoint = use_checkpoint
self.shift_size = window_size // 2
# 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 self.shift_size,
mlp_ratio=mlp_ratio,
qkv_bias=qkv_bias,
drop=drop,
attn_drop=attn_drop,
drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path,
norm_layer=norm_layer)
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 create_mask(self, x, H, W):
# 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, Mh, Mw, 1]
mask_windows = mask_windows.view(-1, self.window_size * self.window_size) # [nW, Mh*Mw]
attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) # [nW, 1, Mh*Mw] - [nW, Mh*Mw, 1]
# [nW, Mh*Mw, Mh*Mw]
attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))
return attn_mask
def forward(self, x, H, W):
attn_mask = self.create_mask(x, H, W) # [nW, Mh*Mw, Mh*Mw]
for blk in self.blocks:
blk.H, blk.W = H, W
if not torch.jit.is_scripting() and self.use_checkpoint:
x = checkpoint.checkpoint(blk, x, attn_mask)
else:
x = blk(x, attn_mask)
if self.downsample is not None:
x = self.downsample(x, H, W)
H, W = (H + 1) // 2, (W + 1) // 2
return x, H, W
class SwinTransformer(nn.Module):
r""" Swin Transformer
A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` -
https://arxiv.org/pdf/2103.14030
Args:
patch_size (int | tuple(int)): Patch size. Default: 4
in_chans (int): Number of input image channels. Default: 3
num_classes (int): Number of classes for classification head. Default: 1000
embed_dim (int): Patch embedding dimension. Default: 96 输出通道 后面的翻倍即可(*2)
depths (tuple(int)): Depth of each Swin Transformer layer. 每个block重复次数
num_heads (tuple(int)): Number of attention heads in different layers.
window_size (int): Window size. Default: 7
mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4
qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True
drop_rate (float): Dropout rate. Default: 0
attn_drop_rate (float): Attention dropout rate. Default: 0
drop_path_rate (float): Stochastic depth rate. Default: 0.1
norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.
patch_norm (bool): If True, add normalization after patch embedding. Default: True
use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False
"""
def __init__(self, patch_size=4, in_chans=3, num_classes=1000,
embed_dim=96, depths=(2, 2, 6, 2), num_heads=(3, 6, 12, 24),
window_size=7, mlp_ratio=4., qkv_bias=True,
drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1,
norm_layer=nn.LayerNorm, patch_norm=True,
use_checkpoint=False, **kwargs):
super().__init__()
self.num_classes = num_classes
self.num_layers = len(depths)
self.embed_dim = embed_dim
self.patch_norm = patch_norm
self.num_features = int(embed_dim * 2 ** (self.num_layers - 1))
self.mlp_ratio = mlp_ratio
# split image into non-overlapping patches
self.patch_embed = PatchEmbed(
patch_size=patch_size, in_c=in_chans, embed_dim=embed_dim,
norm_layer=norm_layer if self.patch_norm else None)
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):
layers = BasicLayer(dim=int(embed_dim * 2 ** i_layer),
depth=depths[i_layer],
num_heads=num_heads[i_layer],
window_size=window_size,
mlp_ratio=self.mlp_ratio,
qkv_bias=qkv_bias,
drop=drop_rate,
attn_drop=attn_drop_rate,
drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])],
norm_layer=norm_layer,
downsample=PatchMerging if (i_layer < self.num_layers - 1) else None,
use_checkpoint=use_checkpoint)
self.layers.append(layers)
self.norm = norm_layer(self.num_features)
self.avgpool = nn.AdaptiveAvgPool1d(1)
self.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity()
self.apply(self._init_weights)
# cnn
self.cnn_layer = Cnn(in_chans)
def _init_weights(self, m):
if isinstance(m, nn.Linear):
nn.init.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)
def forward(self, x):
# cnn
x1 = self.cnn_layer(x)
# x: [B, L, C]
x, H, W = self.patch_embed(x)
x = self.pos_drop(x)
for layer in self.layers:
x, H, W = layer(x, H, W)
# print(x.shape)
# print(x1.shape)
x = x + x1
x = self.norm(x) # [B, L, C]
x = self.avgpool(x.transpose(1, 2)) # [B, C, 1]
x = torch.flatten(x, 1)
x = self.head(x)
return x
def swin_tiny_patch4_window7_224(num_classes: int = 1000, **kwargs):
# trained ImageNet-1K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=7,
embed_dim=96,
depths=(2, 2, 6, 2),
num_heads=(3, 6, 12, 24),
num_classes=num_classes,
**kwargs)
return model
def swin_small_patch4_window7_224(num_classes: int = 1000, **kwargs):
# trained ImageNet-1K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_small_patch4_window7_224.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=7,
embed_dim=96,
depths=(2, 2, 18, 2),
num_heads=(3, 6, 12, 24),
num_classes=num_classes,
**kwargs)
return model
def swin_base_patch4_window7_224(num_classes: int = 1000, **kwargs):
# trained ImageNet-1K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=7,
embed_dim=128,
depths=(2, 2, 18, 2),
num_heads=(4, 8, 16, 32),
num_classes=num_classes,
**kwargs)
return model
def swin_base_patch4_window12_384(num_classes: int = 1000, **kwargs):
# trained ImageNet-1K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=12,
embed_dim=128,
depths=(2, 2, 18, 2),
num_heads=(4, 8, 16, 32),
num_classes=num_classes,
**kwargs)
return model
def swin_base_patch4_window7_224_in22k(num_classes: int = 21841, **kwargs):
# trained ImageNet-22K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224_22k.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=7,
embed_dim=128,
depths=(2, 2, 18, 2),
num_heads=(4, 8, 16, 32),
num_classes=num_classes,
**kwargs)
return model
def swin_base_patch4_window12_384_in22k(num_classes: int = 21841, **kwargs):
# trained ImageNet-22K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384_22k.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=12,
embed_dim=128,
depths=(2, 2, 18, 2),
num_heads=(4, 8, 16, 32),
num_classes=num_classes,
**kwargs)
return model
def swin_large_patch4_window7_224_in22k(num_classes: int = 21841, **kwargs):
# trained ImageNet-22K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window7_224_22k.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=7,
embed_dim=192,
depths=(2, 2, 18, 2),
num_heads=(6, 12, 24, 48),
num_classes=num_classes,
**kwargs)
return model
def swin_large_patch4_window12_384_in22k(num_classes: int = 21841, **kwargs):
# trained ImageNet-22K
# https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window12_384_22k.pth
model = SwinTransformer(in_chans=3,
patch_size=4,
window_size=12,
embed_dim=192,
depths=(2, 2, 18, 2),
num_heads=(6, 12, 24, 48),
num_classes=num_classes,
**kwargs)
return model
class Cnn(nn.Module):
def __init__(self, in_chans=3):
super(Cnn, self).__init__()
self.init_classifier = nn.Sequential(
nn.Conv2d(in_chans, 64, kernel_size=3, stride=2, padding=2),
nn.BatchNorm2d(64),
nn.ReLU(True),
nn.MaxPool2d(kernel_size=3, stride=2)
)
self.cnnBlock64d = nn.Sequential(
nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=1),
nn.BatchNorm2d(64),
nn.ReLU(True),
nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=1),
)
self.cnnBlock_128d = nn.Sequential(
nn.Conv2d(64, 128, kernel_size=3, stride=2, padding=1),
nn.ReLU(True),
nn.Conv2d(128, 128, kernel_size=3, stride=1, padding=1),
)
self.cnnBlock_u128d = nn.Sequential(
nn.Conv2d(64, 128, kernel_size=1, stride=2)
)
self.layer1 = nn.Sequential(
nn.Conv2d(128, 256, kernel_size=3, stride=1),
nn.BatchNorm2d(256),
nn.Conv2d(256, 512, kernel_size=1, stride=2, padding=1),
nn.BatchNorm2d(512),
nn.Conv2d(512, 1024, kernel_size=3, stride=1, padding=1),
nn.BatchNorm2d(1024),
nn.ReLU(True)
)
self.cnnBlock_u2048d = nn.Sequential(
nn.Conv2d(1024, 512, kernel_size=1, stride=1),
nn.BatchNorm2d(512),
nn.Conv2d(512, 512, kernel_size=3, stride=1, padding=1),
nn.BatchNorm2d(512),
nn.Conv2d(512, 2048, kernel_size=1, stride=1),
nn.BatchNorm2d(2048),
nn.ReLU(True),
)
self.cnnBlock1d = nn.Sequential(
nn.Conv2d(1024, 2048, kernel_size=3, stride=1, padding=1),
nn.BatchNorm2d(2048),
nn.ReLU(True),
)
self.cnnBlock_l768 = nn.Sequential(
nn.Conv2d(2048, 256, kernel_size=1, stride=1),
nn.BatchNorm2d(256),
)
self.cnnBlock_r768 = nn.Sequential(
nn.Conv2d(2048, 256, kernel_size=3, stride=1, padding=1),
nn.BatchNorm2d(256),
)
self.cnnBlock = nn.Conv2d(256, 768, kernel_size=1, stride=1, padding=1)
self.avgpool = nn.AvgPool2d(3, 2)
self.relu = nn.ReLU(True)
def forward(self, x):
x = self.init_classifier(x)
x1 = self.cnnBlock64d(x)
x1 = x1 + x
x1 = self.relu(x1)
x2 = self.cnnBlock_128d(x1)
x3 = self.cnnBlock_u128d(x1)
x3 = x3 + x2
x3 = self.relu(x3)
x4 = self.layer1(x3)
x5 = self.cnnBlock_u2048d(x4)
x51 = self.cnnBlock1d(x4)
x5 = x5 + x51
x5 = self.relu(x5)
x5l = self.cnnBlock_l768(x5)
x5r = self.cnnBlock_r768(x5)
x6 = x5l + x5r
x6 = self.relu(x6)
x6 = self.cnnBlock(x6)
x6 = self.avgpool(x6)
# flatten: [B, C, H, W] -> [B, C, HW]
# transpose: [B, C, HW] -> [B, HW, C]
x6 = x6.flatten(2).transpose(1, 2)
return x6