Pytorch實現28個視覺Transformer,開源庫 timm 瞭解一下!附代碼

點擊上方AI算法與圖像處理”,選擇加"星標"或“置頂”

重磅乾貨,第一時間送達

作者丨科技猛獸
審稿丨鄧富城
編輯丨極市平臺

極市導讀

 

本文將介紹一個優秀的PyTorch開源庫——timm庫,並對其中的vision transformer.py代碼進行了詳細解讀。

Transformer 架構早已在自然語言處理任務中得到廣泛應用,但在計算機視覺領域中仍然受到限制。在計算機視覺領域,目前已有大量工作表明模型對 CNN 的依賴不是必需的,當直接應用於圖像塊序列時,Transformer 也能很好地執行圖像分類任務。
本文將簡要介紹了優秀的 PyTorch Image Model 庫:timm庫。與此同時,將會爲大家詳細介紹其中的視覺Transformer代碼以及 一個優秀的視覺Transformer 的PyTorch實現,以幫助大家更快地開展相關實 驗。

什麼是timm庫?

PyTorchImageModels,簡稱timm,是一個巨大的PyTorch代碼集合,包括了一系列:

  • image models
  • layers
  • utilities
  • optimizers
  • schedulers
  • data-loaders / augmentations
  • training / validation scripts

旨在將各種SOTA模型整合在一起,並具有復現ImageNet訓練結果的能力。

timm庫作者是來自加拿大溫哥華的Ross Wightman

作者github鏈接:

https://github.com/rwightman

timm庫鏈接:

https://github.com/rwightman/pytorch-image-models

所有的PyTorch模型及其對應arxiv鏈接如下:


  • Big Transfer ResNetV2 (BiT) - https://arxiv.org/abs/1912.11370

  • CspNet (Cross-Stage Partial Networks) - https://arxiv.org/abs/1911.11929

  • DeiT (Vision Transformer) - https://arxiv.org/abs/2012.12877

  • DenseNet - https://arxiv.org/abs/1608.06993

  • DLA - https://arxiv.org/abs/1707.06484

  • DPN (Dual-Path Network) - https://arxiv.org/abs/1707.01629

  • EfficientNet (MBConvNet Family)

  • EfficientNet NoisyStudent (B0-B7, L2) - https://arxiv.org/abs/1911.04252

  • EfficientNet AdvProp (B0-B8) - https://arxiv.org/abs/1911.09665

  • EfficientNet (B0-B7) - https://arxiv.org/abs/1905.11946

  • EfficientNet-EdgeTPU (S, M, L) - https://ai.googleblog.com/2019/08/efficientnet-edgetpu-creating.html

  • FBNet-C - https://arxiv.org/abs/1812.03443

  • MixNet - https://arxiv.org/abs/1907.09595

  • MNASNet B1, A1 (Squeeze-Excite), and Small - https://arxiv.org/abs/1807.11626

  • MobileNet-V2 - https://arxiv.org/abs/1801.04381

  • Single-Path NAS - https://arxiv.org/abs/1904.02877

  • GPU-Efficient Networks - https://arxiv.org/abs/2006.14090

  • HRNet - https://arxiv.org/abs/1908.07919

  • Inception-V3 - https://arxiv.org/abs/1512.00567

  • Inception-ResNet-V2 and Inception-V4 - https://arxiv.org/abs/1602.07261

  • MobileNet-V3 (MBConvNet w/ Efficient Head) - https://arxiv.org/abs/1905.02244

  • NASNet-A - https://arxiv.org/abs/1707.07012

  • NFNet-F - https://arxiv.org/abs/2102.06171

  • NF-RegNet / NF-ResNet - https://arxiv.org/abs/2101.08692

  • PNasNet - https://arxiv.org/abs/1712.00559

  • RegNet - https://arxiv.org/abs/2003.13678

  • RepVGG - https://arxiv.org/abs/2101.03697

  • ResNet/ResNeXt

  • ResNet (v1b/v1.5) - https://arxiv.org/abs/1512.03385

  • ResNeXt - https://arxiv.org/abs/1611.05431

  • 'Bag of Tricks' / Gluon C, D, E, S variations - https://arxiv.org/abs/1812.01187

  • Weakly-supervised (WSL) Instagram pretrained / ImageNet tuned ResNeXt101 - https://arxiv.org/abs/1805.00932

  • Semi-supervised (SSL) / Semi-weakly Supervised (SWSL) ResNet/ResNeXts - https://arxiv.org/abs/1905.00546

  • ECA-Net (ECAResNet) - https://arxiv.org/abs/1910.03151v4

  • Squeeze-and-Excitation Networks (SEResNet) - https://arxiv.org/abs/1709.01507

  • Res2Net - https://arxiv.org/abs/1904.01169

  • ResNeSt - https://arxiv.org/abs/2004.08955

  • ReXNet - https://arxiv.org/abs/2007.00992

  • SelecSLS - https://arxiv.org/abs/1907.00837

  • Selective Kernel Networks - https://arxiv.org/abs/1903.06586

  • TResNet - https://arxiv.org/abs/2003.13630

  • Vision Transformer - https://arxiv.org/abs/2010.11929

  • VovNet V2 and V1 - https://arxiv.org/abs/1911.06667

  • Xception - https://arxiv.org/abs/1610.02357

  • Xception (Modified Aligned, Gluon) - https://arxiv.org/abs/1802.02611

  • Xception (Modified Aligned, TF) - https://arxiv.org/abs/1802.02611


timm庫特點

所有的模型都有默認的API:

  • accessing/changing the classifier -  get_classifier and  reset_classifier
  • 只對features做前向傳播 -  forward_features

所有模型都支持多尺度特徵提取 (feature pyramids) (通過create_model函數):

  • create_model(name, features_only=True, out_indices=..., output_stride=...)

out_indices 指定返回哪個feature maps to return, 從0開始,out_indices[i]對應着 C(i + 1) feature level。

output_stride 通過dilated convolutions控制網絡的output stride。大多數網絡默認 stride 32 。

所有的模型都有一致的pretrained weight loader,adapts last linear if necessary。

訓練方式支持:

  • NVIDIA DDP w/ a single GPU per process, multiple processes with APEX present (AMP mixed-precision optional)
  • PyTorch DistributedDataParallel w/ multi-gpu, single process (AMP disabled as it crashes when enabled)
  • PyTorch w/ single GPU single process (AMP optional)

動態的全局池化方式可以選擇: average pooling, max pooling, average + max, or concat([average, max]),默認是adaptive average。

Schedulers:

Schedulers 包括step,cosinew/ restarts,tanhw/ restarts,plateau 。

Optimizer:

  • rmsprop_tf adapted from PyTorch RMSProp by myself. Reproduces much improved Tensorflow RMSProp behaviour.
  • radam by Liyuan Liu (https://arxiv.org/abs/1908.03265)
  • novograd by Masashi Kimura (https://arxiv.org/abs/1905.11286)
  • lookahead adapted from impl by Liam (https://arxiv.org/abs/1907.08610)
  • fused<name> optimizers by name with NVIDIA Apex installed
  • adamp and  sgdp by Naver ClovAI (https://arxiv.org/abs/2006.08217)
  • adafactor adapted from FAIRSeq impl (https://arxiv.org/abs/1804.04235)
  • adahessian by David Samuel (https://arxiv.org/abs/2006.00719)

timm庫 vision_transformer.py代碼解讀

代碼來自:

https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py

對應的論文是ViT,是除了官方開源的代碼之外的又一個優秀的PyTorch implement。

An Image Is Worth 16 x 16 Words: Transformers for Image Recognition at Scale

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

https://arxiv.org/abs/2010.11929

另一篇工作DeiT也大量借鑑了timm庫這份代碼的實現:

Training data-efficient image transformers & distillation through attention

Training data-efficient image transformers & distillation through attention

https://arxiv.org/abs/2012.12877

vision_transformer.py:

代碼中定義的變量的含義如下:

img_size:tuple 類型,裏面是int類型,代表輸入的圖片大小,默認是 224
patch_size:tuple 類型,裏面是int類型,代表Patch的大小,默認是 16
in_chans:int 類型,代表輸入圖片的channel數,默認是3
num_classes:int 類型classification head的分類數,比如CIFAR100就是100,默認是 1000
embed_dim:int 類型Transformer的embedding dimension,默認是 768
depth:int  類型,Transformer的Block的數量,默認是 12
num_heads:int 類型,attention heads的數量,默認是12
mlp_ratio:int 類型,mlp hidden dim/embedding dim的值,默認是 4
qkv_bias:bool 類型,attention模塊計算qkv時需要bias嗎,默認是 True
qk_scale: 一般設置成 None 就行。
drop_rate:float 類型,dropout rate,默認是 0
attn_drop_rate:float 類型,attention模塊的dropout rate,默認是 0
drop_path_rate:float 類型,默認是 0
hybrid_backbone:nn.Module 類型,在把圖片轉換成Patch之前,需要先通過一個Backbone嗎?默認是 None
如果是None,就直接把圖片轉化成Patch。
如果不是None,就先通過這個Backbone,再轉化成Patch。
norm_layer:nn.Module 類型,歸一化層類型,默認是 None

1. 導入必要的庫和模型

import mathimport loggingfrom functools import partialfrom collections import OrderedDict
import torchimport torch.nn as nnimport torch.nn.functional as F
from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom .helpers import load_pretrainedfrom .layers import StdConv2dSame, DropPath, to_2tuple, trunc_normal_from .resnet import resnet26d, resnet50dfrom .resnetv2 import ResNetV2from .registry import register_model
2. 定義一個字典,代表標準的模型,如果需要更改模型超參數只需要改變_cfg
的傳入的參數即可。
def _cfg(url='', **kwargs):    return {        'url': url,        'num_classes': 1000, 'input_size': (3, 224, 224), 'pool_size': None,        'crop_pct': .9, 'interpolation': 'bicubic',        'mean': IMAGENET_DEFAULT_MEAN, 'std': IMAGENET_DEFAULT_STD,        'first_conv': 'patch_embed.proj', 'classifier': 'head',        **kwargs    }

3. default_cfgs代表支持的所有模型,也定義成字典的形式:

vit_small_patch16_224裏面的small代表小模型。
ViT的第一步要把圖片分成一個個patch,然後把這些patch組合在一起作爲對圖像的序列化操作,比如一張224 × 224的圖片分成大小爲16 × 16的patch,那一共可以分成196個。所以這個圖片就序列化成了(196, 256)的tensor。所以這裏的:
16: 就代表patch的大小。
224: 就代表輸入圖片的大小。
按照這個命名方式,支持的模型有:vit_base_patch16_224,vit_base_patch16_384等等。

後面的vit_deit_base_patch16_224等等模型代表DeiT這篇論文的模型。

   
   
   
default_cfgs = {    # patch models (my experiments)    'vit_small_patch16_224': _cfg(        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/vit_small_p16_224-15ec54c9.pth',    ),
# patch models (weights ported from official Google JAX impl) 'vit_base_patch16_224': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_p16_224-80ecf9dd.pth', mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), ), 'vit_base_patch32_224': _cfg( url='', # no official model weights for this combo, only for in21k mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_base_patch16_384': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_p16_384-83fb41ba.pth', input_size=(3, 384, 384), mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), crop_pct=1.0), 'vit_base_patch32_384': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_p32_384-830016f5.pth', input_size=(3, 384, 384), mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), crop_pct=1.0), 'vit_large_patch16_224': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_p16_224-4ee7a4dc.pth', mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_large_patch32_224': _cfg( url='', # no official model weights for this combo, only for in21k mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_large_patch16_384': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_p16_384-b3be5167.pth', input_size=(3, 384, 384), mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), crop_pct=1.0), 'vit_large_patch32_384': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_p32_384-9b920ba8.pth', input_size=(3, 384, 384), mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), crop_pct=1.0),
# patch models, imagenet21k (weights ported from official Google JAX impl) 'vit_base_patch16_224_in21k': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_patch16_224_in21k-e5005f0a.pth', num_classes=21843, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_base_patch32_224_in21k': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_patch32_224_in21k-8db57226.pth', num_classes=21843, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_large_patch16_224_in21k': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_patch16_224_in21k-606da67d.pth', num_classes=21843, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_large_patch32_224_in21k': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_patch32_224_in21k-9046d2e7.pth', num_classes=21843, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)), 'vit_huge_patch14_224_in21k': _cfg( url='', # FIXME I have weights for this but > 2GB limit for github release binaries num_classes=21843, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)),
# hybrid models (weights ported from official Google JAX impl) 'vit_base_resnet50_224_in21k': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_resnet50_224_in21k-6f7c7740.pth', num_classes=21843, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), crop_pct=0.9, first_conv='patch_embed.backbone.stem.conv'), 'vit_base_resnet50_384': _cfg( url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_resnet50_384-9fd3c705.pth', input_size=(3, 384, 384), mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), crop_pct=1.0, first_conv='patch_embed.backbone.stem.conv'),
# hybrid models (my experiments) 'vit_small_resnet26d_224': _cfg(), 'vit_small_resnet50d_s3_224': _cfg(), 'vit_base_resnet26d_224': _cfg(), 'vit_base_resnet50d_224': _cfg(),
# deit models (FB weights) 'vit_deit_tiny_patch16_224': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_tiny_patch16_224-a1311bcf.pth'), 'vit_deit_small_patch16_224': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_small_patch16_224-cd65a155.pth'), 'vit_deit_base_patch16_224': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_base_patch16_224-b5f2ef4d.pth',), 'vit_deit_base_patch16_384': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_base_patch16_384-8de9b5d1.pth', input_size=(3, 384, 384), crop_pct=1.0), 'vit_deit_tiny_distilled_patch16_224': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_tiny_distilled_patch16_224-b40b3cf7.pth'), 'vit_deit_small_distilled_patch16_224': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_small_distilled_patch16_224-649709d9.pth'), 'vit_deit_base_distilled_patch16_224': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_base_distilled_patch16_224-df68dfff.pth', ), 'vit_deit_base_distilled_patch16_384': _cfg( url='https://dl.fbaipublicfiles.com/deit/deit_base_distilled_patch16_384-d0272ac0.pth', input_size=(3, 384, 384), crop_pct=1.0),}

4. FFN實現:

   
   
   
class Mlp(nn.Module):    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

5. Attention實現:

在python 3.5以後,@是一個操作符,表示矩陣-向量乘法
[email protected] 就是矩陣-向量乘法A*x: np.dot(A, x)。

   
   
   
class Attention(nn.Module):    def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0.):        super().__init__()        self.num_heads = num_heads        head_dim = dim // num_heads        # NOTE scale factor was wrong in my original version, can set manually to be compat with prev weights        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) self.proj = nn.Linear(dim, dim) self.proj_drop = nn.Dropout(proj_drop)
def forward(self, x): 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)
attn = (q @ k.transpose(-2, -1)) * self.scale attn = attn.softmax(dim=-1) attn = self.attn_drop(attn)
x = (attn @ v).transpose(1, 2).reshape(B, N, C) x = self.proj(x) x = self.proj_drop(x)
# x: (B, N, C) return x

6. 包含Attention和Add & Norm的Block實現:

圖1:Block類對應結構

不同之處是:
先進行Norm,再Attention;先進行Norm,再通過FFN (MLP)。

   
   
   
class Block(nn.Module):    def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0.,                 drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm):        super().__init__()        self.norm1 = norm_layer(dim)        self.attn = Attention(            dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop)        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here        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): x = x + self.drop_path(self.attn(self.norm1(x))) x = x + self.drop_path(self.mlp(self.norm2(x))) return x

7. 接下來要把圖片轉換成Patch,一種做法是直接把Image轉化成Patch,另一種做法是把Backbone輸出的特徵轉化成Patch。

1) 直接把Image轉化成Patch:

輸入的x的維度是:(B, C, H, W)
輸出的PatchEmbedding的維度是:(B, 14*14, 768),768表示embed_dim,14*14表示一共有196個Patches。

   
   
   
class PatchEmbed(nn.Module):    """ Image to Patch Embedding    """    def __init__(self, img_size=224, patch_size=16, in_chans=3, embed_dim=768):        super().__init__()        img_size = to_2tuple(img_size)        patch_size = to_2tuple(patch_size)        num_patches = (img_size[1] // patch_size[1]) * (img_size[0] // patch_size[0])        self.img_size = img_size        self.patch_size = patch_size        self.num_patches = num_patches
self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)
def forward(self, x): B, C, H, W = x.shape # FIXME look at relaxing size constraints 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]})." x = self.proj(x).flatten(2).transpose(1, 2)
# x: (B, 14*14, 768)        return x

2) 把Backbone輸出的特徵轉化成Patch:

輸入的x的維度是:(B, C, H, W)
得到Backbone輸出的維度是:(B, feature_size, feature_size, feature_dim)
輸出的PatchEmbedding的維度是:(B, feature_size, feature_size, embed_dim),一共有feature_size * feature_size個Patches。

   
   
   
class HybridEmbed(nn.Module):    """ CNN Feature Map Embedding    Extract feature map from CNN, flatten, project to embedding dim.    """    def __init__(self, backbone, img_size=224, feature_size=None, in_chans=3, embed_dim=768):        super().__init__()        assert isinstance(backbone, nn.Module)        img_size = to_2tuple(img_size)        self.img_size = img_size        self.backbone = backbone        if feature_size is None:            with torch.no_grad():                # FIXME this is hacky, but most reliable way of determining the exact dim of the output feature                # map for all networks, the feature metadata has reliable channel and stride info, but using                # stride to calc feature dim requires info about padding of each stage that isn't captured.                training = backbone.training                if training:                    backbone.eval()                o = self.backbone(torch.zeros(1, in_chans, img_size[0], img_size[1]))                if isinstance(o, (list, tuple)):                    o = o[-1]  # last feature if backbone outputs list/tuple of features                feature_size = o.shape[-2:]                feature_dim = o.shape[1]                backbone.train(training)        else:            feature_size = to_2tuple(feature_size)            if hasattr(self.backbone, 'feature_info'):                feature_dim = self.backbone.feature_info.channels()[-1]            else:                feature_dim = self.backbone.num_features        self.num_patches = feature_size[0] * feature_size[1]        self.proj = nn.Conv2d(feature_dim, embed_dim, 1)
def forward(self, x): x = self.backbone(x) if isinstance(x, (list, tuple)): x = x[-1] # last feature if backbone outputs list/tuple of features x = self.proj(x).flatten(2).transpose(1, 2) return x

8. 以上是ViT所需的所有模塊的定義,下面是VisionTransformer 這個類的實現:

8.1 使用這個類時需要傳入的變量,其含義已經在本小節一開始介紹。

class VisionTransformer(nn.Module):    """ Vision Transformer
A PyTorch impl of : `An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale` - https://arxiv.org/abs/2010.11929 """ def __init__(self, img_size=224, patch_size=16, in_chans=3, num_classes=1000, embed_dim=768, depth=12, num_heads=12, mlp_ratio=4., qkv_bias=True, qk_scale=None, representation_size=None,                 drop_rate=0., attn_drop_rate=0., drop_path_rate=0., hybrid_backbone=None, norm_layer=None):

8.2 得到分塊後的Patch的數量:

super().__init__()self.num_classes = num_classesself.num_features = self.embed_dim = embed_dim  # num_features for consistency with other modelsnorm_layer = norm_layer or partial(nn.LayerNorm, eps=1e-6)
if hybrid_backbone is not None: self.patch_embed = HybridEmbed( hybrid_backbone, img_size=img_size, in_chans=in_chans, embed_dim=embed_dim)else: self.patch_embed = PatchEmbed( img_size=img_size, patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim)num_patches = self.patch_embed.num_patches

8.3 class token:

一開始定義成(1, 1, 768),之後再變成(B, 1, 768)。

   
   
   
self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))

8.4 定義位置編碼:

self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + 1, embed_dim))

8.5 把12個Block連接起來:

   
   
   
self.pos_drop = nn.Dropout(p=drop_rate)
dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)] # stochastic depth decay ruleself.blocks = nn.ModuleList([ Block( dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale, drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer) for i in range(depth)])self.norm = norm_layer(embed_dim)

8.6 表示層和分類頭:

表示層輸出維度是representation_size,分類頭輸出維度是num_classes。

# Representation layerif representation_size:    self.num_features = representation_size    self.pre_logits = nn.Sequential(OrderedDict([        ('fc', nn.Linear(embed_dim, representation_size)),        ('act', nn.Tanh())    ]))else:    self.pre_logits = nn.Identity()
# Classifier headself.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity()

8.7 初始化各個模塊:

函數trunc_normal_(tensor, mean=0., std=1., a=-2., b=2.)的目的是用截斷的正態分佈繪製的值填充輸入張量,我們只需要輸入均值mean,標準差std,下界a,上界b即可。

self.apply(self._init_weights)表示對各個模塊的權重進行初始化。apply函數的代碼是:

   
   
   
        for module in self.children():            module.apply(fn)        fn(self)        return self

遞歸地將fn應用於每個子模塊,相當於在遞歸調用fn,即_init_weights這個函數。
也就是把模型的所有子模塊的nn.Linear和nn.LayerNorm層都初始化掉。

   
   
   
trunc_normal_(self.pos_embed, std=.02)trunc_normal_(self.cls_token, std=.02)self.apply(self._init_weights)
def _init_weights(self, 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)

8.8 最後就是整個ViT模型的forward實現:

   
   
   
def forward_features(self, x):    B = x.shape[0]    x = self.patch_embed(x)
cls_tokens = self.cls_token.expand(B, -1, -1) # stole cls_tokens impl from Phil Wang, thanks x = torch.cat((cls_tokens, x), dim=1) x = x + self.pos_embed x = self.pos_drop(x)
for blk in self.blocks: x = blk(x)
x = self.norm(x)[:, 0] x = self.pre_logits(x) return x
def forward(self, x): x = self.forward_features(x) x = self.head(x)    return x

9. 下面是Training data-efficient image transformers & distillation through attention這篇論文的DeiT這個類的實現:

整體結構與ViT相似,繼承了上面的VisionTransformer類。

   
   
   
class DistilledVisionTransformer(VisionTransformer):

再額外定義以下3個變量:

  • distillation token:dist_token
  • 新的位置編碼:pos_embed
  • 蒸餾分類頭:head_dist

DeiT相關介紹可以參考:Vision Transformer 超詳細解讀 (原理分析+代碼解讀) (三)。

self.dist_token = nn.Parameter(torch.zeros(1, 1, self.embed_dim))num_patches = self.patch_embed.num_patchesself.pos_embed = nn.Parameter(torch.zeros(1, num_patches + 2, self.embed_dim))self.head_dist = nn.Linear(self.embed_dim, self.num_classes) if self.num_classes > 0 else nn.Identity()

初始化新定義的變量:

   
   
   
trunc_normal_(self.dist_token, std=.02)trunc_normal_(self.pos_embed, std=.02)self.head_dist.apply(self._init_weights)

前向函數:

   
   
   
def forward_features(self, x):    B = x.shape[0]    x = self.patch_embed(x)
cls_tokens = self.cls_token.expand(B, -1, -1) # stole cls_tokens impl from Phil Wang, thanks dist_token = self.dist_token.expand(B, -1, -1) x = torch.cat((cls_tokens, dist_token, x), dim=1)
x = x + self.pos_embed x = self.pos_drop(x)
for blk in self.blocks: x = blk(x)
x = self.norm(x) return x[:, 0], x[:, 1]
def forward(self, x): x, x_dist = self.forward_features(x) x = self.head(x) x_dist = self.head_dist(x_dist) if self.training: return x, x_dist else: # during inference, return the average of both classifier predictions return (x + x_dist) / 2

10. 對位置編碼進行插值:

posemb代表未插值的位置編碼權值,posemb_tok爲位置編碼的token部分,posemb_grid爲位置編碼的插值部分。
首先把要插值部分posemb_grid給reshape成(1, gs_old, gs_old, -1)的形式,再插值成(1, gs_new, gs_new, -1)的形式,最後與token部分在第1維度拼接在一起,得到插值後的位置編碼posemb。

def resize_pos_embed(posemb, posemb_new):    # Rescale the grid of position embeddings when loading from state_dict. Adapted from    # https://github.com/google-research/vision_transformer/blob/00883dd691c63a6830751563748663526e811cee/vit_jax/checkpoint.py#L224    _logger.info('Resized position embedding: %s to %s', posemb.shape, posemb_new.shape)    ntok_new = posemb_new.shape[1]    if True:        posemb_tok, posemb_grid = posemb[:, :1], posemb[0, 1:]        ntok_new -= 1    else:        posemb_tok, posemb_grid = posemb[:, :0], posemb[0]    gs_old = int(math.sqrt(len(posemb_grid)))    gs_new = int(math.sqrt(ntok_new))    _logger.info('Position embedding grid-size from %s to %s', gs_old, gs_new)    posemb_grid = posemb_grid.reshape(1, gs_old, gs_old, -1).permute(0, 3, 1, 2)    posemb_grid = F.interpolate(posemb_grid, size=(gs_new, gs_new), mode='bilinear')    posemb_grid = posemb_grid.permute(0, 2, 3, 1).reshape(1, gs_new * gs_new, -1)    posemb = torch.cat([posemb_tok, posemb_grid], dim=1)    return posemb

11. _create_vision_transformer函數用於創建vision transformer:

checkpoint_filter_fn的作用是加載預訓練權重。

   
   
   
def checkpoint_filter_fn(state_dict, model):    """ convert patch embedding weight from manual patchify + linear proj to conv"""    out_dict = {}    if 'model' in state_dict:        # For deit models        state_dict = state_dict['model']    for k, v in state_dict.items():        if 'patch_embed.proj.weight' in k and len(v.shape) < 4:            # For old models that I trained prior to conv based patchification            O, I, H, W = model.patch_embed.proj.weight.shape            v = v.reshape(O, -1, H, W)        elif k == 'pos_embed' and v.shape != model.pos_embed.shape:            # To resize pos embedding when using model at different size from pretrained weights            v = resize_pos_embed(v, model.pos_embed)        out_dict[k] = v    return out_dict

def _create_vision_transformer(variant, pretrained=False, distilled=False, **kwargs): default_cfg = default_cfgs[variant] default_num_classes = default_cfg['num_classes'] default_img_size = default_cfg['input_size'][-1]
num_classes = kwargs.pop('num_classes', default_num_classes) img_size = kwargs.pop('img_size', default_img_size) repr_size = kwargs.pop('representation_size', None) if repr_size is not None and num_classes != default_num_classes: # Remove representation layer if fine-tuning. This may not always be the desired action, # but I feel better than doing nothing by default for fine-tuning. Perhaps a better interface? _logger.warning("Removing representation layer for fine-tuning.") repr_size = None
model_cls = DistilledVisionTransformer if distilled else VisionTransformer model = model_cls(img_size=img_size, num_classes=num_classes, representation_size=repr_size, **kwargs) model.default_cfg = default_cfg
if pretrained: load_pretrained( model, num_classes=num_classes, in_chans=kwargs.get('in_chans', 3), filter_fn=partial(checkpoint_filter_fn, model=model)) return model

12. 定義和註冊vision transformer模型:

@ 指裝飾器。
@register_model代表註冊器,註冊這個新定義的模型。
model_kwargs是一個存有模型所有超參數的字典。
最後使用上面定義的_create_vision_transformer函數創建模型。

   
   
   
@register_modeldef vit_base_patch16_224(pretrained=False, **kwargs):    """ ViT-Base (ViT-B/16) from original paper (https://arxiv.org/abs/2010.11929).    ImageNet-1k weights fine-tuned from in21k @ 224x224, source https://github.com/google-research/vision_transformer.    """    model_kwargs = dict(patch_size=16, embed_dim=768, depth=12, num_heads=12, **kwargs)    model = _create_vision_transformer('vit_base_patch16_224', pretrained=pretrained, **model_kwargs)    return model

一共可以選擇的模型包括:

ViT系列:
vit_small_patch16_224
vit_base_patch16_224
vit_base_patch32_224
vit_base_patch16_384
vit_base_patch32_384
vit_large_patch16_224
vit_large_patch32_224
vit_large_patch16_384
vit_large_patch32_384
vit_base_patch16_224_in21k
vit_base_patch32_224_in21k
vit_large_patch16_224_in21k
vit_large_patch32_224_in21k
vit_huge_patch14_224_in21k
vit_base_resnet50_224_in21k
vit_base_resnet50_384
vit_small_resnet26d_224
vit_small_resnet50d_s3_224
vit_base_resnet26d_224
vit_base_resnet50d_224

DeiT系列:
vit_deit_tiny_patch16_224
vit_deit_small_patch16_224
vit_deit_base_patch16_224
vit_deit_base_patch16_384
vit_deit_tiny_distilled_patch16_224
vit_deit_small_distilled_patch16_224
vit_deit_base_distilled_patch16_224
vit_deit_base_distilled_patch16_384

以上就是對timm庫 vision_transformer.py代碼的分析。

如何使用timm庫以及 vision_transformer.py代碼搭建自己的模型?

在搭建我們自己的視覺Transformer模型時,我們可以按照下面的步驟操作:首先

  • 繼承timm庫的 VisionTransformer這個類。
  • 添加上自己模型 獨有的一些變量
  • 重寫 forward函數。
  • 通過timm庫的 註冊器註冊新模型。

我們以ViT模型的改進版DeiT爲例:

首先,DeiT的所有模型列表如下:

__all__ = [    'deit_tiny_patch16_224', 'deit_small_patch16_224', 'deit_base_patch16_224',    'deit_tiny_distilled_patch16_224', 'deit_small_distilled_patch16_224',    'deit_base_distilled_patch16_224', 'deit_base_patch16_384',    'deit_base_distilled_patch16_384',]

導入VisionTransformer這個類,註冊器register_model,以及初始化函數trunc_normal_:

from timm.models.vision_transformer import VisionTransformer, _cfgfrom timm.models.registry import register_modelfrom timm.models.layers import trunc_normal_
DeiT的class名稱是DistilledVisionTransformer,它直接繼承了VisionTransformer這個類:
   
   
   
class DistilledVisionTransformer(VisionTransformer):

添加上自己模型獨有的一些變量:

   
   
   
def __init__(self, *args, **kwargs):    super().__init__(*args, **kwargs)    self.dist_token = nn.Parameter(torch.zeros(1, 1, self.embed_dim))    num_patches = self.patch_embed.num_patches    # 位置編碼不是ViT中的(b, N, 256), 而變成了(b, N+2, 256), 原因是還有class token和distillation token.    self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + 2, self.embed_dim))    self.head_dist = nn.Linear(self.embed_dim, self.num_classes) if self.num_classes > 0 else nn.Identity()
trunc_normal_(self.dist_token, std=.02) trunc_normal_(self.pos_embed, std=.02) self.head_dist.apply(self._init_weights)

重寫forward函數:

   
   
   
def forward_features(self, x):    # taken from https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py    # with slight modifications to add the dist_token    B = x.shape[0]
x = self.patch_embed(x)
cls_tokens = self.cls_token.expand(B, -1, -1) # stole cls_tokens impl from Phil Wang, thanks dist_token = self.dist_token.expand(B, -1, -1)
x = torch.cat((cls_tokens, dist_token, x), dim=1)
x = x + self.pos_embed x = self.pos_drop(x)
for blk in self.blocks: x = blk(x)
x = self.norm(x)
return x[:, 0], x[:, 1]
def forward(self, x): x, x_dist = self.forward_features(x) x = self.head(x) x_dist = self.head_dist(x_dist) if self.training: return x, x_dist else: # during inference, return the average of both classifier predictions return (x + x_dist) / 2

通過timm庫的註冊器註冊新模型:

   
   
   
@register_modeldef deit_base_patch16_224(pretrained=False, **kwargs):    model = VisionTransformer(        patch_size=16, embed_dim=768, depth=12, num_heads=12, mlp_ratio=4, qkv_bias=True,        norm_layer=partial(nn.LayerNorm, eps=1e-6), **kwargs)    model.default_cfg = _cfg()    if pretrained:        checkpoint = torch.hub.load_state_dict_from_url(            url="https://dl.fbaipublicfiles.com/deit/deit_base_patch16_224-b5f2ef4d.pth",            map_location="cpu", check_hash=True        )        model.load_state_dict(checkpoint["model"])    return model
   
      
      
      
個人微信(如果沒有備註不拉羣!
請註明: 地區+學校/企業+研究方向+暱稱



下載1:何愷明頂會分享


AI算法與圖像處理」公衆號後臺回覆:何愷明,即可下載。總共有6份PDF,涉及 ResNet、Mask RCNN等經典工作的總結分析


下載2:終身受益的編程指南:Google編程風格指南


AI算法與圖像處理」公衆號後臺回覆:c++,即可下載。歷經十年考驗,最權威的編程規範!



 
    
    
    
下載3 CVPR2020

AI算法與圖像處公衆號後臺回覆: CVPR2020 即可下載1467篇CVPR 2020論文


覺得不錯就點亮在看吧



本文分享自微信公衆號 - AI算法與圖像處理(AI_study)。
如有侵權,請聯繫 [email protected] 刪除。
本文參與“OSC源創計劃”,歡迎正在閱讀的你也加入,一起分享。

發表評論
所有評論
還沒有人評論,想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章