目标检测算法改进系列之Backbone替换为Swin Transformer

Swin Transformer简介

《Swin Transformer: Hierarchical Vision Transformer using Shifted Windows》作为2021 ICCV最佳论文，屠榜了各大CV任务，性能优于DeiT、ViT和EfficientNet等主干网络，已经替代经典的CNN架构，成为了计算机视觉领域通用的backbone。它基于了ViT模型的思想，创新性的引入了滑动窗口机制，让模型能够学习到跨窗口的信息，同时也。同时通过下采样层，使得模型能够处理超分辨率的图片，节省计算量以及能够关注全局和局部的信息。而本文将从原理和代码角度详细解析Swin Transformer的架构。

目前将 Transformer 从自然语言处理领域应用到计算机视觉领域主要有两大挑战：
（1）视觉实体的方差较大，例如同一个物体，拍摄角度不同，转化为二进制后的图片就会具有很大的差异。同时在不同场景下视觉 Transformer 性能未必很好。
（2）图像分辨率高，像素点多，如果采用ViT模型，自注意力的计算量会与像素的平方成正比。针对上述两个问题，论文中提出了一种基于滑动窗口机制，具有层级设计（下采样层） 的 Swin Transformer。

其中滑窗操作包括不重叠的 local window，和重叠的 cross-window。将注意力计算限制在一个窗口（window size固定）中，一方面能引入 CNN 卷积操作的局部性，另一方面能大幅度节省计算量，它只和窗口数量成线性关系。通过下采样的层级设计，能够逐渐增大感受野，从而使得注意力机制也能够注意到全局的特征。

Swin Transformer代码实现

# --------------------------------------------------------
# Swin Transformer
# Copyright (c) 2021 Microsoft
# Licensed under The MIT License [see LICENSE for details]
# Written by Ze Liu, Yutong Lin, Yixuan Wei
# --------------------------------------------------------import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.checkpoint as checkpoint
import numpy as np
from timm.models.layers import DropPath, to_2tuple, trunc_normal___all__ = ['SwinTransformer_Tiny']class Mlp(nn.Module):""" Multilayer perceptron."""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_featureshidden_features = hidden_features or in_featuresself.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 xdef window_partition(x, window_size):"""Args:x: (B, H, W, C)window_size (int): window sizeReturns:windows: (num_windows*B, window_size, window_size, C)"""B, H, W, C = x.shapex = x.view(B, H // window_size, window_size, W // window_size, window_size, C)windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)return windowsdef window_reverse(windows, window_size, H, W):"""Args:windows: (num_windows*B, window_size, window_size, C)window_size (int): Window sizeH (int): Height of imageW (int): Width of imageReturns:x: (B, H, W, C)"""B = int(windows.shape[0] / (H * W / window_size / window_size))x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)return xclass WindowAttention(nn.Module):""" 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: Trueqk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if setattn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0proj_drop (float, optional): Dropout ratio of output. Default: 0.0"""def __init__(self, dim, window_size, num_heads, qkv_bias=True, qk_scale=None, attn_drop=0., proj_drop=0.):super().__init__()self.dim = dimself.window_size = window_size  # Wh, Wwself.num_heads = num_headshead_dim = dim // num_headsself.scale = qk_scale or head_dim ** -0.5# define a parameter table of relative position biasself.relative_position_bias_table = nn.Parameter(torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads))  # 2*Wh-1 * 2*Ww-1, nH# get pair-wise relative position index for each token inside the windowcoords_h = torch.arange(self.window_size[0])coords_w = torch.arange(self.window_size[1])coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Wwcoords_flatten = torch.flatten(coords, 1)  # 2, Wh*Wwrelative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Wwrelative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0relative_coords[:, :, 1] += self.window_size[1] - 1relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Wwself.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)trunc_normal_(self.relative_position_bias_table, std=.02)self.softmax = nn.Softmax(dim=-1)def forward(self, x, mask=None):""" Forward function.Args:x: input features with shape of (num_windows*B, N, C)mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None"""B_, N, C = x.shapeqkv = 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)q = q * self.scaleattn = (q @ k.transpose(-2, -1))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)  # Wh*Ww,Wh*Ww,nHrelative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Wwattn = attn + relative_position_bias.unsqueeze(0)if mask is not None:nW = mask.shape[0]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)x = (attn @ v).transpose(1, 2).reshape(B_, N, C)x = self.proj(x)x = self.proj_drop(x)return xclass SwinTransformerBlock(nn.Module):""" 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: Trueqk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.drop (float, optional): Dropout rate. Default: 0.0attn_drop (float, optional): Attention dropout rate. Default: 0.0drop_path (float, optional): Stochastic depth rate. Default: 0.0act_layer (nn.Module, optional): Activation layer. Default: nn.GELUnorm_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, qk_scale=None, drop=0., attn_drop=0., drop_path=0.,act_layer=nn.GELU, norm_layer=nn.LayerNorm):super().__init__()self.dim = dimself.num_heads = num_headsself.window_size = window_sizeself.shift_size = shift_sizeself.mlp_ratio = mlp_ratioassert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size"self.norm1 = norm_layer(dim)self.attn = WindowAttention(dim, window_size=to_2tuple(self.window_size), num_heads=num_heads,qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop)self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()self.norm2 = norm_layer(dim)mlp_hidden_dim = int(dim * mlp_ratio)self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)self.H = Noneself.W = Nonedef forward(self, x, mask_matrix):""" Forward function.Args:x: Input feature, tensor size (B, H*W, C).H, W: Spatial resolution of the input feature.mask_matrix: Attention mask for cyclic shift."""B, L, C = x.shapeH, W = self.H, self.Wassert L == H * W, "input feature has wrong size"shortcut = xx = self.norm1(x)x = x.view(B, H, W, C)# pad feature maps to multiples of window sizepad_l = pad_t = 0pad_r = (self.window_size - W % self.window_size) % self.window_sizepad_b = (self.window_size - H % self.window_size) % self.window_sizex = F.pad(x, (0, 0, pad_l, pad_r, pad_t, pad_b))_, Hp, Wp, _ = x.shape# cyclic shiftif self.shift_size > 0:shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))attn_mask = mask_matrix.type(x.dtype)else:shifted_x = xattn_mask = None# partition windowsx_windows = window_partition(shifted_x, self.window_size)  # nW*B, window_size, window_size, Cx_windows = x_windows.view(-1, self.window_size * self.window_size, C)  # nW*B, window_size*window_size, C# W-MSA/SW-MSAattn_windows = self.attn(x_windows, mask=attn_mask)  # nW*B, window_size*window_size, C# merge windowsattn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)shifted_x = window_reverse(attn_windows, self.window_size, Hp, Wp)  # B H' W' C# reverse cyclic shiftif self.shift_size > 0:x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2))else:x = shifted_xif pad_r > 0 or pad_b > 0:x = x[:, :H, :W, :].contiguous()x = x.view(B, H * W, C)# FFNx = shortcut + self.drop_path(x)x = x + self.drop_path(self.mlp(self.norm2(x)))return xclass PatchMerging(nn.Module):""" Patch Merging LayerArgs: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 = dimself.reduction = nn.Linear(4 * dim, 2 * dim, bias=False)self.norm = norm_layer(4 * dim)def forward(self, x, H, W):""" Forward function.Args:x: Input feature, tensor size (B, H*W, C).H, W: Spatial resolution of the input feature."""B, L, C = x.shapeassert L == H * W, "input feature has wrong size"x = x.view(B, H, W, C)# paddingpad_input = (H % 2 == 1) or (W % 2 == 1)if pad_input:x = F.pad(x, (0, 0, 0, W % 2, 0, H % 2))x0 = x[:, 0::2, 0::2, :]  # B H/2 W/2 Cx1 = x[:, 1::2, 0::2, :]  # B H/2 W/2 Cx2 = x[:, 0::2, 1::2, :]  # B H/2 W/2 Cx3 = x[:, 1::2, 1::2, :]  # B H/2 W/2 Cx = torch.cat([x0, x1, x2, x3], -1)  # B H/2 W/2 4*Cx = x.view(B, -1, 4 * C)  # B H/2*W/2 4*Cx = self.norm(x)x = self.reduction(x)return xclass BasicLayer(nn.Module):""" A basic Swin Transformer layer for one stage.Args:dim (int): Number of feature channelsdepth (int): Depths of this stage.num_heads (int): Number of attention head.window_size (int): Local window size. Default: 7.mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4.qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: Trueqk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.drop (float, optional): Dropout rate. Default: 0.0attn_drop (float, optional): Attention dropout rate. Default: 0.0drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNormdownsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: Noneuse_checkpoint (bool): Whether to use checkpointing to save memory. Default: False."""def __init__(self,dim,depth,num_heads,window_size=7,mlp_ratio=4.,qkv_bias=True,qk_scale=None,drop=0.,attn_drop=0.,drop_path=0.,norm_layer=nn.LayerNorm,downsample=None,use_checkpoint=False):super().__init__()self.window_size = window_sizeself.shift_size = window_size // 2self.depth = depthself.use_checkpoint = use_checkpoint# build blocksself.blocks = nn.ModuleList([SwinTransformerBlock(dim=dim,num_heads=num_heads,window_size=window_size,shift_size=0 if (i % 2 == 0) else window_size // 2,mlp_ratio=mlp_ratio,qkv_bias=qkv_bias,qk_scale=qk_scale,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 layerif downsample is not None:self.downsample = downsample(dim=dim, norm_layer=norm_layer)else:self.downsample = Nonedef forward(self, x, H, W):""" Forward function.Args:x: Input feature, tensor size (B, H*W, C).H, W: Spatial resolution of the input feature."""# calculate attention mask for SW-MSAHp = int(np.ceil(H / self.window_size)) * self.window_sizeWp = int(np.ceil(W / self.window_size)) * self.window_sizeimg_mask = torch.zeros((1, Hp, Wp, 1), device=x.device)  # 1 Hp Wp 1h_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 = 0for h in h_slices:for w in w_slices:img_mask[:, h, w, :] = cntcnt += 1mask_windows = window_partition(img_mask, self.window_size)  # nW, window_size, window_size, 1mask_windows = mask_windows.view(-1, self.window_size * self.window_size)attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))for blk in self.blocks:blk.H, blk.W = H, Wif self.use_checkpoint:x = checkpoint.checkpoint(blk, x, attn_mask)else:x = blk(x, attn_mask)if self.downsample is not None:x_down = self.downsample(x, H, W)Wh, Ww = (H + 1) // 2, (W + 1) // 2return x, H, W, x_down, Wh, Wwelse:return x, H, W, x, H, Wclass PatchEmbed(nn.Module):""" Image to Patch EmbeddingArgs:patch_size (int): Patch token size. Default: 4.in_chans (int): Number of input image channels. Default: 3.embed_dim (int): Number of linear projection output channels. Default: 96.norm_layer (nn.Module, optional): Normalization layer. Default: None"""def __init__(self, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None):super().__init__()patch_size = to_2tuple(patch_size)self.patch_size = patch_sizeself.in_chans = in_chansself.embed_dim = embed_dimself.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)if norm_layer is not None:self.norm = norm_layer(embed_dim)else:self.norm = Nonedef forward(self, x):"""Forward function."""# padding_, _, H, W = x.size()if W % self.patch_size[1] != 0:x = F.pad(x, (0, self.patch_size[1] - W % self.patch_size[1]))if H % self.patch_size[0] != 0:x = F.pad(x, (0, 0, 0, self.patch_size[0] - H % self.patch_size[0]))x = self.proj(x)  # B C Wh Wwif self.norm is not None:Wh, Ww = x.size(2), x.size(3)x = x.flatten(2).transpose(1, 2)x = self.norm(x)x = x.transpose(1, 2).view(-1, self.embed_dim, Wh, Ww)return xclass SwinTransformer(nn.Module):""" Swin Transformer backbone.A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows`  -https://arxiv.org/pdf/2103.14030Args:pretrain_img_size (int): Input image size for training the pretrained model,used in absolute postion embedding. Default 224.patch_size (int | tuple(int)): Patch size. Default: 4.in_chans (int): Number of input image channels. Default: 3.embed_dim (int): Number of linear projection output channels. Default: 96.depths (tuple[int]): Depths of each Swin Transformer stage.num_heads (tuple[int]): Number of attention head of each stage.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: Trueqk_scale (float): Override default qk scale of head_dim ** -0.5 if set.drop_rate (float): Dropout rate.attn_drop_rate (float): Attention dropout rate. Default: 0.drop_path_rate (float): Stochastic depth rate. Default: 0.2.norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.ape (bool): If True, add absolute position embedding to the patch embedding. Default: False.patch_norm (bool): If True, add normalization after patch embedding. Default: True.out_indices (Sequence[int]): Output from which stages.frozen_stages (int): Stages to be frozen (stop grad and set eval mode).-1 means not freezing any parameters.use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False."""def __init__(self,pretrain_img_size=224,patch_size=4,in_chans=3,embed_dim=96,depths=[2, 2, 6, 2],num_heads=[3, 6, 12, 24],window_size=7,mlp_ratio=4.,qkv_bias=True,qk_scale=None,drop_rate=0.,attn_drop_rate=0.,drop_path_rate=0.2,norm_layer=nn.LayerNorm,ape=False,patch_norm=True,out_indices=(0, 1, 2, 3),frozen_stages=-1,use_checkpoint=False):super().__init__()self.pretrain_img_size = pretrain_img_sizeself.num_layers = len(depths)self.embed_dim = embed_dimself.ape = apeself.patch_norm = patch_normself.out_indices = out_indicesself.frozen_stages = frozen_stages# split image into non-overlapping patchesself.patch_embed = PatchEmbed(patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim,norm_layer=norm_layer if self.patch_norm else None)# absolute position embeddingif self.ape:pretrain_img_size = to_2tuple(pretrain_img_size)patch_size = to_2tuple(patch_size)patches_resolution = [pretrain_img_size[0] // patch_size[0], pretrain_img_size[1] // patch_size[1]]self.absolute_pos_embed = nn.Parameter(torch.zeros(1, embed_dim, patches_resolution[0], patches_resolution[1]))trunc_normal_(self.absolute_pos_embed, std=.02)self.pos_drop = nn.Dropout(p=drop_rate)# stochastic depthdpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]  # stochastic depth decay rule# build layersself.layers = nn.ModuleList()for i_layer in range(self.num_layers):layer = BasicLayer(dim=int(embed_dim * 2 ** i_layer),depth=depths[i_layer],num_heads=num_heads[i_layer],window_size=window_size,mlp_ratio=mlp_ratio,qkv_bias=qkv_bias,qk_scale=qk_scale,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(layer)num_features = [int(embed_dim * 2 ** i) for i in range(self.num_layers)]self.num_features = num_features# add a norm layer for each outputfor i_layer in out_indices:layer = norm_layer(num_features[i_layer])layer_name = f'norm{i_layer}'self.add_module(layer_name, layer)self.channel = [i.size(1) for i in self.forward(torch.randn(1, 3, 640, 640))]def forward(self, x):"""Forward function."""x = self.patch_embed(x)Wh, Ww = x.size(2), x.size(3)if self.ape:# interpolate the position embedding to the corresponding sizeabsolute_pos_embed = F.interpolate(self.absolute_pos_embed, size=(Wh, Ww), mode='bicubic')x = (x + absolute_pos_embed).flatten(2).transpose(1, 2)  # B Wh*Ww Celse:x = x.flatten(2).transpose(1, 2)x = self.pos_drop(x)outs = []for i in range(self.num_layers):layer = self.layers[i]x_out, H, W, x, Wh, Ww = layer(x, Wh, Ww)if i in self.out_indices:norm_layer = getattr(self, f'norm{i}')x_out = norm_layer(x_out)out = x_out.view(-1, H, W, self.num_features[i]).permute(0, 3, 1, 2).contiguous()outs.append(out)return outsdef update_weight(model_dict, weight_dict):idx, temp_dict = 0, {}for k, v in weight_dict.items():if k in model_dict.keys() and np.shape(model_dict[k]) == np.shape(v):temp_dict[k] = vidx += 1model_dict.update(temp_dict)print(f'loading weights... {idx}/{len(model_dict)} items')return model_dictdef SwinTransformer_Tiny(weights=''):model = SwinTransformer(depths=[2, 2, 6, 2], num_heads=[3, 6, 12, 24])if weights:model.load_state_dict(update_weight(model.state_dict(), torch.load(weights)['model']))return modelif __name__ == '__main__':device = torch.device('cuda:0')model = SwinTransformer().to(device)model.half()# model.load_state_dict(update_weight(model.state_dict(), torch.load('swin_tiny_patch4_window7_224_22k.pth')['model']))inputs = torch.randn((1, 3, 640, 512)).to(device).half()res = model(inputs)for i in res:print(i.size())print(model.channel)

Backbone替换

yolo.py修改

def parse_model函数

def parse_model(d, ch):  # model_dict, input_channels(3)# Parse a YOLOv5 model.yaml dictionaryLOGGER.info(f"\n{'':>3}{'from':>18}{'n':>3}{'params':>10}  {'module':<40}{'arguments':<30}")anchors, nc, gd, gw, act = d['anchors'], d['nc'], d['depth_multiple'], d['width_multiple'], d.get('activation')if act:Conv.default_act = eval(act)  # redefine default activation, i.e. Conv.default_act = nn.SiLU()LOGGER.info(f"{colorstr('activation:')} {act}")  # printna = (len(anchors[0]) // 2) if isinstance(anchors, list) else anchors  # number of anchorsno = na * (nc + 5)  # number of outputs = anchors * (classes + 5)is_backbone = Falselayers, save, c2 = [], [], ch[-1]  # layers, savelist, ch outfor i, (f, n, m, args) in enumerate(d['backbone'] + d['head']):  # from, number, module, argstry:t = mm = eval(m) if isinstance(m, str) else m  # eval stringsexcept:passfor j, a in enumerate(args):with contextlib.suppress(NameError):try:args[j] = eval(a) if isinstance(a, str) else a  # eval stringsexcept:args[j] = an = n_ = max(round(n * gd), 1) if n > 1 else n  # depth gainif m in {Conv, GhostConv, Bottleneck, GhostBottleneck, SPP, SPPF, DWConv, MixConv2d, Focus, CrossConv,BottleneckCSP, C3, C3TR, C3SPP, C3Ghost, nn.ConvTranspose2d, DWConvTranspose2d, C3x}:c1, c2 = ch[f], args[0]if c2 != no:  # if not outputc2 = make_divisible(c2 * gw, 8)args = [c1, c2, *args[1:]]if m in {BottleneckCSP, C3, C3TR, C3Ghost, C3x}:args.insert(2, n)  # number of repeatsn = 1elif m is nn.BatchNorm2d:args = [ch[f]]elif m is Concat:c2 = sum(ch[x] for x in f)# TODO: channel, gw, gdelif m in {Detect, Segment}:args.append([ch[x] for x in f])if isinstance(args[1], int):  # number of anchorsargs[1] = [list(range(args[1] * 2))] * len(f)if m is Segment:args[3] = make_divisible(args[3] * gw, 8)elif m is Contract:c2 = ch[f] * args[0] ** 2elif m is Expand:c2 = ch[f] // args[0] ** 2elif isinstance(m, str):t = mm = timm.create_model(m, pretrained=args[0], features_only=True)c2 = m.feature_info.channels()elif m in {SwinTransformer_Tiny}: #添加Backbonem = m(*args)c2 = m.channelelse:c2 = ch[f]if isinstance(c2, list):is_backbone = Truem_ = mm_.backbone = Trueelse:m_ = nn.Sequential(*(m(*args) for _ in range(n))) if n > 1 else m(*args)  # modulet = str(m)[8:-2].replace('__main__.', '')  # module typenp = sum(x.numel() for x in m_.parameters())  # number paramsm_.i, m_.f, m_.type, m_.np = i + 4 if is_backbone else i, f, t, np  # attach index, 'from' index, type, number paramsLOGGER.info(f'{i:>3}{str(f):>18}{n_:>3}{np:10.0f}  {t:<40}{str(args):<30}')  # printsave.extend(x % (i + 4 if is_backbone else i) for x in ([f] if isinstance(f, int) else f) if x != -1)  # append to savelistlayers.append(m_)if i == 0:ch = []if isinstance(c2, list):ch.extend(c2)for _ in range(5 - len(ch)):ch.insert(0, 0)else:ch.append(c2)return nn.Sequential(*layers), sorted(save)

def _forward_once函数

def _forward_once(self, x, profile=False, visualize=False):y, dt = [], []  # outputsfor m in self.model:if m.f != -1:  # if not from previous layerx = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f]  # from earlier layersif profile:self._profile_one_layer(m, x, dt)if hasattr(m, 'backbone'):x = m(x)for _ in range(5 - len(x)):x.insert(0, None)for i_idx, i in enumerate(x):if i_idx in self.save:y.append(i)else:y.append(None)x = x[-1]else:x = m(x)  # runy.append(x if m.i in self.save else None)  # save outputif visualize:feature_visualization(x, m.type, m.i, save_dir=visualize)return x

创建.yaml配置文件

# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
# Parameters
nc: 80  # number of classes
depth_multiple: 0.33  # model depth multiple
width_multiple: 0.25  # layer channel multiple
anchors:- [10,13, 16,30, 33,23]  # P3/8- [30,61, 62,45, 59,119]  # P4/16- [116,90, 156,198, 373,326]  # P5/32# 0-P1/2
# 1-P2/4
# 2-P3/8
# 3-P4/16
# 4-P5/32# YOLOv5 v6.0 backbone
backbone:# [from, number, module, args][[-1, 1, SwinTransformer_Tiny, [False]], # 4[-1, 1, SPPF, [1024, 5]],  # 5]# YOLOv5 v6.0 head
head:[[-1, 1, Conv, [512, 1, 1]], # 6[-1, 1, nn.Upsample, [None, 2, 'nearest']], # 7[[-1, 3], 1, Concat, [1]],  # cat backbone P4 8[-1, 3, C3, [512, False]],  # 9[-1, 1, Conv, [256, 1, 1]], # 10[-1, 1, nn.Upsample, [None, 2, 'nearest']], # 11[[-1, 2], 1, Concat, [1]],  # cat backbone P3 12[-1, 3, C3, [256, False]],  # 13 (P3/8-small)[-1, 1, Conv, [256, 3, 2]], # 14[[-1, 10], 1, Concat, [1]],  # cat head P4 15[-1, 3, C3, [512, False]],  # 16 (P4/16-medium)[-1, 1, Conv, [512, 3, 2]], # 17[[-1, 5], 1, Concat, [1]],  # cat head P5 18[-1, 3, C3, [1024, False]],  # 19 (P5/32-large)[[13, 16, 19], 1, Detect, [nc, anchors]],  # Detect(P3, P4, P5)]