一、backbone
1.ResNet:
from __future__ import absolute_import
from torch import nn
from torch.nn import functional as F
from torch.nn import init
import torchvision
import torch
__all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50', 'resnet101',
'resnet152']
class LayerNorm(nn.LayerNorm):
"""Subclass torch's LayerNorm to handle fp16."""
def forward(self, x: torch.Tensor):
orig_type = x.dtype
ret = super().forward(x.type(torch.float32))
return ret.type(orig_type)
class ResNet(nn.Module):
__factory = {
18: torchvision.models.resnet18,
34: torchvision.models.resnet34,
50: torchvision.models.resnet50,
101: torchvision.models.resnet101,
152: torchvision.models.resnet152,
}
def __init__(self, depth, pretrained=True, cut_at_pooling=False,
num_features=0, norm=False, dropout=0, num_classes=0):
super(ResNet, self).__init__()
self.pretrained = pretrained
self.depth = depth
self.cut_at_pooling = cut_at_pooling
# Construct base (pretrained) resnet
if depth not in ResNet.__factory:
raise KeyError("Unsupported depth:", depth)
resnet = ResNet.__factory[depth](pretrained=pretrained)
resnet.layer4[0].conv2.stride = (1, 1)
resnet.layer4[0].downsample[0].stride = (1, 1)
self.base = nn.Sequential(
resnet.conv1, resnet.bn1, resnet.relu, resnet.maxpool,
resnet.layer1, resnet.layer2, resnet.layer3, resnet.layer4)
self.gap = nn.AdaptiveAvgPool2d(1)
self.conv2d = nn.Conv2d(2048, 768, kernel_size=1, stride=1)
if not self.cut_at_pooling:
self.num_features = num_features
self.norm = norm
self.dropout = dropout
self.has_embedding = num_features > 0
self.num_classes = num_classes
out_planes = resnet.fc.in_features
# Append new layers
if self.has_embedding:
self.feat = nn.Linear(out_planes, self.num_features)
self.feat_bn = nn.BatchNorm1d(self.num_features)
init.kaiming_normal_(self.feat.weight, mode='fan_out')
init.constant_(self.feat.bias, 0)
else:
# Change the num_features to CNN output channels
self.num_features = out_planes
self.feat_bn = nn.BatchNorm1d(self.num_features)
self.feat_bn.bias.requires_grad_(False)
if self.dropout > 0:
self.drop = nn.Dropout(self.dropout)
if self.num_classes > 0:
self.classifier = nn.Linear(self.num_features, self.num_classes, bias=False)
init.normal_(self.classifier.weight, std=0.001)
init.constant_(self.feat_bn.weight, 1)
init.constant_(self.feat_bn.bias, 0)
scale = 768 ** -0.5
self.ln_post = LayerNorm(768)
self.proj = nn.Parameter(scale * torch.randn(768, 512))
if not pretrained:
self.reset_params()
def forward(self, x):
bs = x.size(0)
x = self.base(x)
# x = self.conv2d(x)
# # x = x.reshape(x.shape[0], x.shape[1], -1)
# # x = x.permute(0, 2, 1)
# x = self.gap(x)
# # x = x.reshape(x.shape[0], x.shape[1], -1)
# x = x.view(x.size(0), -1)
# ln_proj = self.ln_post(x)
# if self.proj is not None:
# proj = ln_proj @ self.proj
#
# return x.unsqueeze(1),x.unsqueeze(1),proj.unsqueeze(1)
x = x.view(x.size(0), -1)
if self.cut_at_pooling:
return x
if self.has_embedding:
bn_x = self.feat_bn(self.feat(x))
else:
bn_x = self.feat_bn(x)
if (self.training is False):
bn_x = F.normalize(bn_x)
return bn_x
if self.norm:
bn_x = F.normalize(bn_x)
elif self.has_embedding:
bn_x = F.relu(bn_x)
if self.dropout > 0:
bn_x = self.drop(bn_x)
if self.num_classes > 0:
prob = self.classifier(bn_x)
else:
return bn_x
return prob
def reset_params(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
init.kaiming_normal_(m.weight, mode='fan_out')
if m.bias is not None:
init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm2d):
init.constant_(m.weight, 1)
init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm1d):
init.constant_(m.weight, 1)
init.constant_(m.bias, 0)
elif isinstance(m, nn.Linear):
init.normal_(m.weight, std=0.001)
if m.bias is not None:
init.constant_(m.bias, 0)
def resnet18(**kwargs):
return ResNet(18, **kwargs)
def resnet34(**kwargs):
return ResNet(34, **kwargs)
def resnet50(**kwargs):
return ResNet(50, **kwargs)
def resnet101(**kwargs):
return ResNet(101, **kwargs)
def resnet152(**kwargs):
return ResNet(152, **kwargs)
2.IBN-ResNet:
- resnet_ibn_a.py (基层):
import torch
import torch.nn as nn
import math
import torch.utils.model_zoo as model_zoo
__all__ = ['ResNet', 'resnet50_ibn_a', 'resnet101_ibn_a']
model_urls = {
'ibn_resnet50a': './model/backbone/pretrained_models/resnet50_ibn_a.pth.tar',
'ibn_resnet101a': './logs/pretrained/resnet101_ibn_a.pth.tar',
}
def conv3x3(in_planes, out_planes, stride=1):
"3x3 convolution with padding"
return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
padding=1, bias=False)
class BasicBlock(nn.Module):
expansion = 1
def __init__(self, inplanes, planes, stride=1, downsample=None):
super(BasicBlock, self).__init__()
self.conv1 = conv3x3(inplanes, planes, stride)
self.bn1 = nn.BatchNorm2d(planes)
self.relu = nn.ReLU(inplace=True)
self.conv2 = conv3x3(planes, planes)
self.bn2 = nn.BatchNorm2d(planes)
self.downsample = downsample
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)
if self.downsample is not None:
residual = self.downsample(x)
out += residual
out = self.relu(out)
return out
class IBN(nn.Module):
def __init__(self, planes):
super(IBN, self).__init__()
half1 = int(planes/2)
self.half = half1
half2 = planes - half1
self.IN = nn.InstanceNorm2d(half1, affine=True)
self.BN = nn.BatchNorm2d(half2)
def forward(self, x):
split = torch.split(x, self.half, 1)
out1 = self.IN(split[0].contiguous())
out2 = self.BN(split[1].contiguous())
out = torch.cat((out1, out2), 1)
return out
class Bottleneck(nn.Module):
expansion = 4
def __init__(self, inplanes, planes, ibn=False, stride=1, downsample=None):
super(Bottleneck, self).__init__()
self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
if ibn:
self.bn1 = IBN(planes)
else:
self.bn1 = nn.BatchNorm2d(planes)
self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride,
padding=1, bias=False)
self.bn2 = nn.BatchNorm2d(planes)
self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1, bias=False)
self.bn3 = nn.BatchNorm2d(planes * self.expansion)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
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 ResNet(nn.Module):
def __init__(self, block, layers, num_classes=1000):
scale = 64
self.inplanes = scale
super(ResNet, self).__init__()
self.conv1 = nn.Conv2d(3, scale, kernel_size=7, stride=2, padding=3,
bias=False)
self.bn1 = nn.BatchNorm2d(scale)
self.relu = nn.ReLU(inplace=True)
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.layer1 = self._make_layer(block, scale, layers[0])
self.layer2 = self._make_layer(block, scale*2, layers[1], stride=2)
self.layer3 = self._make_layer(block, scale*4, layers[2], stride=2)
self.layer4 = self._make_layer(block, scale*8, layers[3], stride=2)
self.avgpool = nn.AvgPool2d(7)
self.fc = nn.Linear(scale * 8 * block.expansion, num_classes)
for m in self.modules():
if isinstance(m, nn.Conv2d):
n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
m.weight.data.normal_(0, math.sqrt(2. / n))
elif isinstance(m, nn.BatchNorm2d):
m.weight.data.fill_(1)
m.bias.data.zero_()
elif isinstance(m, nn.InstanceNorm2d):
m.weight.data.fill_(1)
m.bias.data.zero_()
def _make_layer(self, block, planes, blocks, stride=1):
downsample = None
if stride != 1 or self.inplanes != planes * block.expansion:
downsample = nn.Sequential(
nn.Conv2d(self.inplanes, planes * block.expansion,
kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(planes * block.expansion),
)
layers = []
ibn = True
if planes == 512:
ibn = False
layers.append(block(self.inplanes, planes, ibn, stride, downsample))
self.inplanes = planes * block.expansion
for i in range(1, blocks):
layers.append(block(self.inplanes, planes, ibn))
return nn.Sequential(*layers)
def forward(self, x):
x = self.conv1(x)
x = self.bn1(x)
x = self.relu(x)
x = self.maxpool(x)
x = self.layer1(x)
x = self.layer2(x)
x = self.layer3(x)
x = self.layer4(x)
x = self.avgpool(x)
x = x.view(x.size(0), -1)
x = self.fc(x)
return x
def resnet50_ibn_a(pretrained=False, **kwargs):
"""Constructs a ResNet-50 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)
if pretrained:
print("pretrained true")
state_dict = torch.load(model_urls['ibn_resnet50a'], map_location=torch.device('cpu'))['state_dict']
state_dict = remove_module_key(state_dict)
model.load_state_dict(state_dict)
return model
def resnet101_ibn_a(pretrained=False, **kwargs):
"""Constructs a ResNet-101 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNelayer1t
"""
model = ResNet(Bottleneck, [3, 4, 23, 3], **kwargs)
if pretrained:
state_dict = torch.load(model_urls['ibn_resnet101a'], map_location=torch.device('cpu'))['state_dict']
state_dict = remove_module_key(state_dict)
model.load_state_dict(state_dict)
return model
def remove_module_key(state_dict):
for key in list(state_dict.keys()):
if 'module' in key:
state_dict[key.replace('module.','')] = state_dict.pop(key)
return state_dict
- resnet_ibn.py (代码修改过,forward对着resnet重新写):
from __future__ import absolute_import
import os
from torch import nn
from torch.nn import functional as F
from torch.nn import init
import torchvision
import torch
from .resnet_ibn_a import resnet50_ibn_a, resnet101_ibn_a
__all__ = ['ResNetIBN', 'resnet_ibn50a', 'resnet_ibn101a']
class ResNetIBN(nn.Module):
__factory = {
'50a': resnet50_ibn_a,
'101a': resnet101_ibn_a
}
def __init__(self, depth, pretrained=True, cut_at_pooling=False,
num_features=0, norm=False, dropout=0, num_classes=0):
super(ResNetIBN, self).__init__()
self.depth = depth
self.pretrained = pretrained
self.cut_at_pooling = cut_at_pooling
print(os.getcwd())
resnet = ResNetIBN.__factory[depth](pretrained=pretrained)
resnet.layer4[0].conv2.stride = (1,1)
resnet.layer4[0].downsample[0].stride = (1,1)
self.base = nn.Sequential(
resnet.conv1, resnet.bn1, resnet.relu, resnet.maxpool)
self.stage1 = resnet.layer1
self.stage2 = resnet.layer2
self.stage3 = resnet.layer3
self.stage4 = resnet.layer4
self.gap = nn.AdaptiveAvgPool2d(1)
if not self.cut_at_pooling:
self.num_features = num_features
self.norm = norm
self.dropout = dropout
self.has_embedding = num_features > 0
self.num_classes = num_classes
out_planes = resnet.fc.in_features
# Append new layers
if self.has_embedding:
self.feat = nn.Linear(out_planes, self.num_features)
self.feat_bn = nn.BatchNorm1d(self.num_features)
init.kaiming_normal_(self.feat.weight, mode='fan_out')
init.constant_(self.feat.bias, 0)
else:
# Change the num_features to CNN output channels
self.num_features = out_planes
self.feat_bn = nn.BatchNorm1d(self.num_features)
self.feat_bn.bias.requires_grad_(False)
if self.dropout > 0:
self.drop = nn.Dropout(self.dropout)
if self.num_classes > 0:
self.classifier = nn.Linear(self.num_features, self.num_classes, bias=False)
init.normal_(self.classifier.weight, std=0.001)
init.constant_(self.feat_bn.weight, 1)
init.constant_(self.feat_bn.bias, 0)
if not pretrained:
print("not pretrained")
self.reset_params()
def forward(self, x, feature_withbn=False):
bs = x.size(0)
# calculate base_features
x = self.base(x)
x1 = self.stage1(x)
x2 = self.stage2(x1)
x3 = self.stage3(x2)
x4 = self.stage4(x3)
# reverse model
f1 = self.f1(x1)
f2 = self.f2(x2)
f3 = self.f3(x3)
f_g = self.gap(x4)
f1 = f1.view(f1.shape[0], f1.shape[1], -1).permute(0, 2, 1)
f2 = f2.view(f2.shape[0], f2.shape[1], -1).permute(0, 2, 1)
f3 = f3.view(f3.shape[0], f3.shape[1], -1).permute(0, 2, 1)
x4 = f_g.view(f_g.shape[0], f_g.shape[1], -1).permute(0, 2, 1)
# print("f:", f1.shape, f2.shape, f3.shape, x4.shape)
avg_f = torch.mean(torch.stack([f1, f2, f3, x4], dim=0), dim=0)
sfa1 = self.attention(f1, f2, avg_f)
sfa2 = self.attention(f3, x4, avg_f)
sfa = self.attention(sfa1, sfa2, avg_f)
# print(sfa.shape, f_g.shape)
x = f_g.view(x.size(0), -1)
sfa = sfa.view(sfa.size(0), -1)
sfa += x
if self.cut_at_pooling:
return x
if self.has_embedding:
bn_x = self.feat_bn(self.feat(x))
else:
bn_x = self.feat_bn(x)
sfa = self.bnneck_sfa(sfa)
if (self.training is False):
bn_x = F.normalize(bn_x)
sfa = F.normalize(sfa)
return bn_x, sfa
if self.norm:
bn_x = F.normalize(bn_x)
sfa = F.normalize(sfa)
elif self.has_embedding:
bn_x = F.relu(bn_x)
if self.dropout > 0:
bn_x = self.drop(bn_x)
if self.num_classes > 0:
prob = self.classifier(bn_x)
sfa_prob = self.classifier_sfa(sfa)
else:
return bn_x, sfa
return bn_x, prob, sfa, sfa_prob
def reset_params(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
init.kaiming_normal_(m.weight, mode='fan_out')
if m.bias is not None:
init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm2d):
init.constant_(m.weight, 1)
init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm1d):
init.constant_(m.weight, 1)
init.constant_(m.bias, 0)
elif isinstance(m, nn.Linear):
init.normal_(m.weight, std=0.001)
if m.bias is not None:
init.constant_(m.bias, 0)
resnet = ResNetIBN.__factory[self.depth](pretrained=self.pretrained)
self.base[0].load_state_dict(resnet.conv1.state_dict())
self.base[1].load_state_dict(resnet.bn1.state_dict())
self.stage1.load_state_dict(resnet.layer1.state_dict())
self.stage2.load_state_dict(resnet.layer2.state_dict())
self.stage3.load_state_dict(resnet.layer3.state_dict())
self.stage4.load_state_dict(resnet.layer4.state_dict())
def resnet_ibn50a(**kwargs):
return ResNetIBN('50a', **kwargs)
def resnet_ibn101a(**kwargs):
return ResNetIBN('101a', **kwargs)
if __name__ == '__main__':
net = resnet_ibn50a()
print(net)
3.ViT(B/16,B/32,L/16,L/32):
"""
original code from rwightman:
https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py
"""
from functools import partial
from collections import OrderedDict
import torch
import torch.nn as nn
def drop_path(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(x, self.drop_prob, self.training)
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
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
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
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
def _init_vit_weights(m):
"""
ViT weight initialization
:param m: module
"""
if isinstance(m, nn.Linear):
nn.init.trunc_normal_(m.weight, std=.01)
if m.bias is not None:
nn.init.zeros_(m.bias)
elif isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(m.weight, mode="fan_out")
if m.bias is not None:
nn.init.zeros_(m.bias)
elif isinstance(m, nn.LayerNorm):
nn.init.zeros_(m.bias)
nn.init.ones_(m.weight)
def vit_base_patch16_224(num_classes: int = 1000):
"""
ViT-Base model (ViT-B/16) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-1k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
链接: https://pan.baidu.com/s/1zqb08naP0RPqqfSXfkB2EA 密码: eu9f
"""
model = VisionTransformer(img_size=224,
patch_size=16,
embed_dim=768,
depth=12,
num_heads=12,
representation_size=None,
num_classes=num_classes)
return model
def vit_base_patch16_224_in21k(num_classes: int = 21843, has_logits: bool = True):
"""
ViT-Base model (ViT-B/16) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-21k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_patch16_224_in21k-e5005f0a.pth
"""
model = VisionTransformer(img_size=224,
patch_size=16,
embed_dim=768,
depth=12,
num_heads=12,
representation_size=768 if has_logits else None,
num_classes=num_classes)
return model
def vit_base_patch32_224(num_classes: int = 1000):
"""
ViT-Base model (ViT-B/32) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-1k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
链接: https://pan.baidu.com/s/1hCv0U8pQomwAtHBYc4hmZg 密码: s5hl
"""
model = VisionTransformer(img_size=224,
patch_size=32,
embed_dim=768,
depth=12,
num_heads=12,
representation_size=None,
num_classes=num_classes)
return model
def vit_base_patch32_224_in21k(num_classes: int = 21843, has_logits: bool = True):
"""
ViT-Base model (ViT-B/32) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-21k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_patch32_224_in21k-8db57226.pth
"""
model = VisionTransformer(img_size=224,
patch_size=32,
embed_dim=768,
depth=12,
num_heads=12,
representation_size=768 if has_logits else None,
num_classes=num_classes)
return model
def vit_large_patch16_224(num_classes: int = 1000):
"""
ViT-Large model (ViT-L/16) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-1k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
链接: https://pan.baidu.com/s/1cxBgZJJ6qUWPSBNcE4TdRQ 密码: qqt8
"""
model = VisionTransformer(img_size=224,
patch_size=16,
embed_dim=1024,
depth=24,
num_heads=16,
representation_size=None,
num_classes=num_classes)
return model
def vit_large_patch16_224_in21k(num_classes: int = 21843, has_logits: bool = True):
"""
ViT-Large model (ViT-L/16) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-21k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_patch16_224_in21k-606da67d.pth
"""
model = VisionTransformer(img_size=224,
patch_size=16,
embed_dim=1024,
depth=24,
num_heads=16,
representation_size=1024 if has_logits else None,
num_classes=num_classes)
return model
def vit_large_patch32_224_in21k(num_classes: int = 21843, has_logits: bool = True):
"""
ViT-Large model (ViT-L/32) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-21k weights @ 224x224, source https://github.com/google-research/vision_transformer.
weights ported from official Google JAX impl:
https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_patch32_224_in21k-9046d2e7.pth
"""
model = VisionTransformer(img_size=224,
patch_size=32,
embed_dim=1024,
depth=24,
num_heads=16,
representation_size=1024 if has_logits else None,
num_classes=num_classes)
return model
def vit_huge_patch14_224_in21k(num_classes: int = 21843, has_logits: bool = True):
"""
ViT-Huge model (ViT-H/14) from original paper (https://arxiv.org/abs/2010.11929).
ImageNet-21k weights @ 224x224, source https://github.com/google-research/vision_transformer.
NOTE: converted weights not currently available, too large for github release hosting.
"""
model = VisionTransformer(img_size=224,
patch_size=14,
embed_dim=1280,
depth=32,
num_heads=16,
representation_size=1280 if has_logits else None,
num_classes=num_classes)
return model
4. EffienctNet:
总结
个人学习使用,主要总结主流backbone用于消融实验。使用方法很简单,把网络定义拿过去,forward也直接拿过去,根据最后输出的特征维度,写一个卷积之类的把维度进行统一。
后续还有其他的视觉backbone会根据自己科研实验再进行补充。