seo网站推广优化费用,wordpress后台登录地址,网站单页生成器,软文自助发稿平台oem一、本文介绍
本文给大家带来的最新改进机制是2024最新深度可分卷积与多尺度卷积的结合的模块MSCB#xff0c;其核心机制是 Multi-scale Depth-wise Convolution (MSDC) 是一种改进的 卷积神经网络 #xff08;CNN#xff09;结构#xff0c;旨在提升卷积操作的多尺度特征…一、本文介绍
本文给大家带来的最新改进机制是2024最新深度可分卷积与多尺度卷积的结合的模块MSCB其核心机制是 Multi-scale Depth-wise Convolution (MSDC) 是一种改进的 卷积神经网络 CNN结构旨在提升卷积操作的多尺度特征提取能力。它的核心思想是通过在多个尺度下进行卷积操作以捕获不同层级的图像特征同时保持 深度可分卷积Depth-wise Convolution 的计算效率我将其和C3k2进行结合多种结合方式分别为辅助yolov11进行特征提取能力和特征融合能力本文内容为独家创新文章内涵代码和添加方法。
训练信息YOLO11-C3k2-MSCB1 summary: 395 layers, 2,555,235 parameters, 2,555,219 gradients, 6.3 GFLOPs训练信息YOLO11-C3k2-MSCB2 summary: 410 layers, 2,358,867 parameters, 2,358,851 gradients, 6.2 GFLOPs未优化版本YOLO11 summary: 319 layers, 2,590,035 parameters, 2,590,019 gradients, 6.4 GFLOPs
系列专栏
YOLOv11改进更换卷积、添加注意力、更换主干网络、图像去噪、去雾、增强等涨点系列------发论文必备https://blog.csdn.net/m0_58941767/category_12987736.html?spm1001.2014.3001.5482https://blog.csdn.net/m0_58941767/category_12987736.html?spm1001.2014.3001.5482 目录
一、本文介绍
二、原理介绍
1. 深度可分卷积Depth-wise Convolution
2. 多尺度卷积核的引入
3. 深度可分卷积与多尺度卷积的结合
三、核心代码
四、使用方式
4.1 修改一
4.2 修改二
4.3 修改三
4.4 修改四
五、正式训练
5.1 yaml文件1
5.2 yaml文件2
5.3 训练代码
5.4 训练过程截图
五、本文总结 二、原理介绍 论文地址 官方论文地址
代码地址 官方代码地址 Multi-scale Depth-wise Convolution (MSDC) 是一种改进的卷积神经网络 CNN 结构旨在提升卷积操作的多尺度特征提取能力。它的核心思想是通过在多个尺度下进行卷积操作以捕获不同层级的图像特征同时保持 深度可分卷积Depth-wise Convolution 的计算效率。 下面我们了解一下MSDC的 基本原理 1. 深度可分卷积Depth-wise Convolution 在传统卷积中卷积核会对输入的每个通道进行操作然后得到一个新的特征图而这种操作往往涉及大量的计算尤其是在输入数据和卷积核的通道数较多时计算量会急剧增加。 深度可分卷Depth-wise Convolution将传统卷积操作分解成 两个阶段 1逐通道卷积 对于输入的每个通道使用一个卷积核单独进行卷积操作。每个卷积核只处理一个通道通常使用较小的卷积核例如3x3。 2逐点卷积 在每个通道的输出特征图上使用一个1x1卷积核进行线性组合即通过逐点卷积将每个通道的输出进行融合。 2. 多尺度卷积核的引入 传统的卷积神经网络通常使用固定大小的卷积核如3x3、5x5等来提取特征。然而单一尺寸的卷积核往往只能捕捉到固定尺度的特征无法全面地捕捉图像中不同大小、不同尺度的物体特征。为了解决这个问题 多尺度卷积核Multi-scale Kernels 被引入到 MSDC 中。 在 MSDC 中通过引入多种尺度尺寸不同的卷积核来提取图像的 多尺度特征 。常见的做法是使用不同尺寸的卷积核并行处理输入特征图比如使用 的卷积核捕捉局部的细节信息。使用 或 的卷积核捕捉较大的上下文信息。这些不同尺度的卷积核能够捕捉到图像中的多种尺度特征例如小物体、大物体等尤其是在图像中物体尺度变化较大的情况下多尺度卷积能够帮助网络提高对不同尺度特征的感知能力。 3. 深度可分卷积与多尺度卷积的结合 将深度可分卷积与多尺度卷积结合MSDC 在保持计算效率的同时能够有效地捕捉图像中不同尺度的特征。具体而言MSDC 使用多尺度卷积核例如 等分别对输入的特征图进行卷积操作并通过深度可分卷积的方式对每个尺度的卷积核进行操作最后将这些不同尺度的特征进行融合例如通过拼接或加和的方式。 这种设计方式的具体流程如下 1. 逐通道卷积 每个卷积核不同尺度只作用于输入的每一个通道分别对不同尺度的特征进行处理。 2. 多尺度特征提取 多个不同尺度的卷积核会在同一层次上并行工作每个卷积核从不同的感受野范围内提取特征。 3. 特征融合 通过连接concatenation或加和summation等方式将来自不同尺度卷积核的输出特征进行融合得到包含多尺度信息的特征图。 三、核心代码
import torch
import torch.nn as nn
from functools import partial
import math
from timm.models.layers import trunc_normal_tf_
from timm.models.helpers import named_apply__all__ [C3k2_MSCB1, C3k2_MSCB2]def gcd(a, b):while b:a, b b, a % breturn a# Other types of layers can go here (e.g., nn.Linear, etc.)
def _init_weights(module, name, scheme):if isinstance(module, nn.Conv2d) or isinstance(module, nn.Conv3d):if scheme normal:nn.init.normal_(module.weight, std.02)if module.bias is not None:nn.init.zeros_(module.bias)elif scheme trunc_normal:trunc_normal_tf_(module.weight, std.02)if module.bias is not None:nn.init.zeros_(module.bias)elif scheme xavier_normal:nn.init.xavier_normal_(module.weight)if module.bias is not None:nn.init.zeros_(module.bias)elif scheme kaiming_normal:nn.init.kaiming_normal_(module.weight, modefan_out, nonlinearityrelu)if module.bias is not None:nn.init.zeros_(module.bias)else:# efficientnet likefan_out module.kernel_size[0] * module.kernel_size[1] * module.out_channelsfan_out // module.groupsnn.init.normal_(module.weight, 0, math.sqrt(2.0 / fan_out))if module.bias is not None:nn.init.zeros_(module.bias)elif isinstance(module, nn.BatchNorm2d) or isinstance(module, nn.BatchNorm3d):nn.init.constant_(module.weight, 1)nn.init.constant_(module.bias, 0)elif isinstance(module, nn.LayerNorm):nn.init.constant_(module.weight, 1)nn.init.constant_(module.bias, 0)def act_layer(act, inplaceFalse, neg_slope0.2, n_prelu1):# activation layeract act.lower()if act relu:layer nn.ReLU(inplace)elif act relu6:layer nn.ReLU6(inplace)elif act leakyrelu:layer nn.LeakyReLU(neg_slope, inplace)elif act prelu:layer nn.PReLU(num_parametersn_prelu, initneg_slope)elif act gelu:layer nn.GELU()elif act hswish:layer nn.Hardswish(inplace)else:raise NotImplementedError(activation layer [%s] is not found % act)return layerdef channel_shuffle(x, groups):batchsize, num_channels, height, width x.data.size()channels_per_group num_channels // groups# reshapex x.view(batchsize, groups,channels_per_group, height, width)x torch.transpose(x, 1, 2).contiguous()# flattenx x.view(batchsize, -1, height, width)return x# Multi-scale depth-wise convolution (MSDC)
class MSDC(nn.Module):def __init__(self, in_channels, kernel_sizes, stride, activationrelu6, dw_parallelTrue):super(MSDC, self).__init__()self.in_channels in_channelsself.kernel_sizes kernel_sizesself.activation activationself.dw_parallel dw_parallelself.dwconvs nn.ModuleList([nn.Sequential(nn.Conv2d(self.in_channels, self.in_channels, kernel_size, stride, kernel_size // 2,groupsself.in_channels, biasFalse),nn.BatchNorm2d(self.in_channels),act_layer(self.activation, inplaceTrue))for kernel_size in self.kernel_sizes])self.init_weights(normal)def init_weights(self, scheme):named_apply(partial(_init_weights, schemescheme), self)def forward(self, x):# Apply the convolution layers in a loopoutputs []for dwconv in self.dwconvs:dw_out dwconv(x)outputs.append(dw_out)if self.dw_parallel False:x x dw_out# You can return outputs based on what you intend to do with themreturn outputsclass MSCB(nn.Module):Multi-scale convolution block (MSCB)def __init__(self, in_channels, out_channels, shortcutFalse, stride1, kernel_sizes[1, 3, 5], expansion_factor2, dw_parallelTrue, activationrelu6):super(MSCB, self).__init__()add shortcutself.in_channels in_channelsself.out_channels out_channelsself.stride strideself.kernel_sizes kernel_sizesself.expansion_factor expansion_factorself.dw_parallel dw_parallelself.add addself.activation activationself.n_scales len(self.kernel_sizes)# check stride valueassert self.stride in [1, 2]# Skip connection if stride is 1self.use_skip_connection True if self.stride 1 else False# expansion factorself.ex_channels int(self.in_channels * self.expansion_factor)self.pconv1 nn.Sequential(# pointwise convolutionnn.Conv2d(self.in_channels, self.ex_channels, 1, 1, 0, biasFalse),nn.BatchNorm2d(self.ex_channels),act_layer(self.activation, inplaceTrue))self.msdc MSDC(self.ex_channels, self.kernel_sizes, self.stride, self.activation,dw_parallelself.dw_parallel)if self.add True:self.combined_channels self.ex_channels * 1else:self.combined_channels self.ex_channels * self.n_scalesself.pconv2 nn.Sequential(# pointwise convolutionnn.Conv2d(self.combined_channels, self.out_channels, 1, 1, 0, biasFalse),nn.BatchNorm2d(self.out_channels),)if self.use_skip_connection and (self.in_channels ! self.out_channels):self.conv1x1 nn.Conv2d(self.in_channels, self.out_channels, 1, 1, 0, biasFalse)self.init_weights(normal)def init_weights(self, scheme):named_apply(partial(_init_weights, schemescheme), self)def forward(self, x):pout1 self.pconv1(x)msdc_outs self.msdc(pout1)if self.add True:dout 0for dwout in msdc_outs:dout dout dwoutelse:dout torch.cat(msdc_outs, dim1)dout channel_shuffle(dout, gcd(self.combined_channels, self.out_channels))out self.pconv2(dout)if self.use_skip_connection:if self.in_channels ! self.out_channels:x self.conv1x1(x)return x outelse:return outdef autopad(k, pNone, d1): # kernel, padding, dilationPad to same shape outputs.if d 1:k d * (k - 1) 1 if isinstance(k, int) else [d * (x - 1) 1 for x in k] # actual kernel-sizeif p is None:p k // 2 if isinstance(k, int) else [x // 2 for x in k] # auto-padreturn pclass Conv(nn.Module):Standard convolution with args(ch_in, ch_out, kernel, stride, padding, groups, dilation, activation).default_act nn.SiLU() # default activationdef __init__(self, c1, c2, k1, s1, pNone, g1, d1, actTrue):Initialize Conv layer with given arguments including activation.super().__init__()self.conv nn.Conv2d(c1, c2, k, s, autopad(k, p, d), groupsg, dilationd, biasFalse)self.bn nn.BatchNorm2d(c2)self.act self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity()def forward(self, x):Apply convolution, batch normalization and activation to input tensor.return self.act(self.bn(self.conv(x)))def forward_fuse(self, x):Perform transposed convolution of 2D data.return self.act(self.conv(x))class Bottleneck(nn.Module):Standard bottleneck.def __init__(self, c1, c2, shortcutTrue, g1, k(3, 3), e0.5):Initializes a standard bottleneck module with optional shortcut connection and configurable parameters.super().__init__()c_ int(c2 * e) # hidden channelsself.cv1 Conv(c1, c_, k[0], 1)self.cv2 Conv(c_, c2, k[1], 1, gg)self.add shortcut and c1 c2def forward(self, x):Applies the YOLO FPN to input data.return x self.cv2(self.cv1(x)) if self.add else self.cv2(self.cv1(x))class C2f(nn.Module):Faster Implementation of CSP Bottleneck with 2 convolutions.def __init__(self, c1, c2, n1, shortcutFalse, g1, e0.5):Initializes a CSP bottleneck with 2 convolutions and n Bottleneck blocks for faster processing.super().__init__()self.c int(c2 * e) # hidden channelsself.cv1 Conv(c1, 2 * self.c, 1, 1)self.cv2 Conv((2 n) * self.c, c2, 1) # optional actFReLU(c2)self.m nn.ModuleList(Bottleneck(self.c, self.c, shortcut, g, k((3, 3), (3, 3)), e1.0) for _ in range(n))def forward(self, x):Forward pass through C2f layer.y list(self.cv1(x).chunk(2, 1))y.extend(m(y[-1]) for m in self.m)return self.cv2(torch.cat(y, 1))def forward_split(self, x):Forward pass using split() instead of chunk().y list(self.cv1(x).split((self.c, self.c), 1))y.extend(m(y[-1]) for m in self.m)return self.cv2(torch.cat(y, 1))class C3(nn.Module):CSP Bottleneck with 3 convolutions.def __init__(self, c1, c2, n1, shortcutTrue, g1, e0.5):Initialize the CSP Bottleneck with given channels, number, shortcut, groups, and expansion values.super().__init__()c_ int(c2 * e) # hidden channelsself.cv1 Conv(c1, c_, 1, 1)self.cv2 Conv(c1, c_, 1, 1)self.cv3 Conv(2 * c_, c2, 1) # optional actFReLU(c2)self.m nn.Sequential(*(Bottleneck(c_, c_, shortcut, g, k((1, 1), (3, 3)), e1.0) for _ in range(n)))def forward(self, x):Forward pass through the CSP bottleneck with 2 convolutions.return self.cv3(torch.cat((self.m(self.cv1(x)), self.cv2(x)), 1))class C3k(C3):C3k is a CSP bottleneck module with customizable kernel sizes for feature extraction in neural networks.def __init__(self, c1, c2, n1, shortcutTrue, g1, e0.5, k3):Initializes the C3k module with specified channels, number of layers, and configurations.super().__init__(c1, c2, n, shortcut, g, e)c_ int(c2 * e) # hidden channels# self.m nn.Sequential(*(RepBottleneck(c_, c_, shortcut, g, k(k, k), e1.0) for _ in range(n)))self.m nn.Sequential(*(Bottleneck(c_, c_, shortcut, g, k(k, k), e1.0) for _ in range(n)))class C3k_MSCB(C3):C3k is a CSP bottleneck module with customizable kernel sizes for feature extraction in neural networks.def __init__(self, c1, c2, n1, shortcutTrue, g1, e0.5, k3):Initializes the C3k module with specified channels, number of layers, and configurations.super().__init__(c1, c2, n, shortcut, g, e)c_ int(c2 * e) # hidden channels# self.m nn.Sequential(*(RepBottleneck(c_, c_, shortcut, g, k(k, k), e1.0) for _ in range(n)))self.m nn.Sequential(*(MSCB(c_, c_, shortcut) for _ in range(n)))class C3k2_MSCB1(C2f):Faster Implementation of CSP Bottleneck with 2 convolutions.def __init__(self, c1, c2, n1, c3kFalse, e0.5, g1, shortcutTrue):Initializes the C3k2 module, a faster CSP Bottleneck with 2 convolutions and optional C3k blocks.super().__init__(c1, c2, n, shortcut, g, e)self.m nn.ModuleList(C3k(self.c, self.c, 2, shortcut, g) if c3k else MSCB(self.c, self.c, shortcut) for _ in range(n))# 解析 c3k在主干和网络最后一个C3k2的时候设置True走的是C3k, 否则我们走的是MSBlockclass C3k2_MSCB2(C2f):Faster Implementation of CSP Bottleneck with 2 convolutions.def __init__(self, c1, c2, n1, c3kFalse, e0.5, g1, shortcutTrue):Initializes the C3k2 module, a faster CSP Bottleneck with 2 convolutions and optional C3k blocks.super().__init__(c1, c2, n, shortcut, g, e)self.m nn.ModuleList(C3k_MSCB(self.c, self.c, 2, shortcut, g) if c3k else Bottleneck(self.c, self.c, shortcut, g) for _ in range(n))if __name__ __main__:# Generating Sample imageimage_size (1, 64, 240, 240)image torch.rand(*image_size)image_size1 (1, 64, 480, 480)image1 torch.rand(*image_size1)# Modelmobilenet_v1 MSCB(64, 64,)out mobilenet_v1(image)print(out.size()) 四、使用方式
4.1 修改一
第一还是建立文件我们找到如下ultralytics/nn文件夹下建立一个目录名字呢就是Addmodules文件夹 然后在其内部建立一个新的py文件将核心代码复制粘贴进去即可。 4.2 修改二
第二步我们在该目录下创建一个新的py文件名字为__init__.py然后在其内部导入我们的检测头如下图所示。 4.3 修改三
第三步我门中到如下文件ultralytics/nn/tasks.py进行导入和注册我们的模块 4.4 修改四
按照我的添加在parse_model里添加即可。 到此就修改完成了大家可以复制下面的yaml文件运行。 五、正式训练
5.1 yaml文件1 训练信息YOLO11-C3k2-MSCB1 summary: 395 layers, 2,555,235 parameters, 2,555,219 gradients, 6.3 GFLOPs # Ultralytics YOLO , AGPL-3.0 license
# YOLO11 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect# Parameters
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. modelyolo11n.yaml will call yolo11.yaml with scale n# [depth, width, max_channels]n: [0.50, 0.25, 1024] # summary: 319 layers, 2624080 parameters, 2624064 gradients, 6.6 GFLOPss: [0.50, 0.50, 1024] # summary: 319 layers, 9458752 parameters, 9458736 gradients, 21.7 GFLOPsm: [0.50, 1.00, 512] # summary: 409 layers, 20114688 parameters, 20114672 gradients, 68.5 GFLOPsl: [1.00, 1.00, 512] # summary: 631 layers, 25372160 parameters, 25372144 gradients, 87.6 GFLOPsx: [1.00, 1.50, 512] # summary: 631 layers, 56966176 parameters, 56966160 gradients, 196.0 GFLOPs# YOLO11n backbone
backbone:# [from, repeats, module, args]- [-1, 1, Conv, [64, 3, 2]] # 0-P1/2- [-1, 1, Conv, [128, 3, 2]] # 1-P2/4- [-1, 2, C3k2_MSCB1, [256, False, 0.25]]- [-1, 1, Conv, [256, 3, 2]] # 3-P3/8- [-1, 2, C3k2_MSCB1, [512, False, 0.25]]- [-1, 1, Conv, [512, 3, 2]] # 5-P4/16- [-1, 2, C3k2_MSCB1, [512, True]]- [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32- [-1, 2, C3k2_MSCB1, [1024, True]]- [-1, 1, SPPF, [1024, 5]] # 9- [-1, 2, C2PSA, [1024]] # 10# YOLO11n head
head:- [-1, 1, nn.Upsample, [None, 2, nearest]]- [[-1, 6], 1, Concat, [1]] # cat backbone P4- [-1, 2, C3k2_MSCB1, [512, False]] # 13- [-1, 1, nn.Upsample, [None, 2, nearest]]- [[-1, 4], 1, Concat, [1]] # cat backbone P3- [-1, 2, C3k2_MSCB1, [256, False]] # 16 (P3/8-small)- [-1, 1, Conv, [256, 3, 2]]- [[-1, 13], 1, Concat, [1]] # cat head P4- [-1, 2, C3k2_MSCB1, [512, False]] # 19 (P4/16-medium)- [-1, 1, Conv, [512, 3, 2]]- [[-1, 10], 1, Concat, [1]] # cat head P5- [-1, 2, C3k2_MSCB1, [1024, True]] # 22 (P5/32-large)- [[16, 19, 22], 1, Detect, [nc]] # Detect(P3, P4, P5) 5.2 yaml文件2 训练信息YOLO11-C3k2-MSCB2 summary: 410 layers, 2,358,867 parameters, 2,358,851 gradients, 6.2 GFLOPs # Ultralytics YOLO , AGPL-3.0 license
# YOLO11 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect# Parameters
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. modelyolo11n.yaml will call yolo11.yaml with scale n# [depth, width, max_channels]n: [0.50, 0.25, 1024] # summary: 319 layers, 2624080 parameters, 2624064 gradients, 6.6 GFLOPss: [0.50, 0.50, 1024] # summary: 319 layers, 9458752 parameters, 9458736 gradients, 21.7 GFLOPsm: [0.50, 1.00, 512] # summary: 409 layers, 20114688 parameters, 20114672 gradients, 68.5 GFLOPsl: [1.00, 1.00, 512] # summary: 631 layers, 25372160 parameters, 25372144 gradients, 87.6 GFLOPsx: [1.00, 1.50, 512] # summary: 631 layers, 56966176 parameters, 56966160 gradients, 196.0 GFLOPs# YOLO11n backbone
backbone:# [from, repeats, module, args]- [-1, 1, Conv, [64, 3, 2]] # 0-P1/2- [-1, 1, Conv, [128, 3, 2]] # 1-P2/4- [-1, 2, C3k2_MSCB2, [256, False, 0.25]]- [-1, 1, Conv, [256, 3, 2]] # 3-P3/8- [-1, 2, C3k2_MSCB2, [512, False, 0.25]]- [-1, 1, Conv, [512, 3, 2]] # 5-P4/16- [-1, 2, C3k2_MSCB2, [512, True]]- [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32- [-1, 2, C3k2_MSCB2, [1024, True]]- [-1, 1, SPPF, [1024, 5]] # 9- [-1, 2, C2PSA, [1024]] # 10# YOLO11n head
head:- [-1, 1, nn.Upsample, [None, 2, nearest]]- [[-1, 6], 1, Concat, [1]] # cat backbone P4- [-1, 2, C3k2_MSCB2, [512, False]] # 13- [-1, 1, nn.Upsample, [None, 2, nearest]]- [[-1, 4], 1, Concat, [1]] # cat backbone P3- [-1, 2, C3k2_MSCB2, [256, False]] # 16 (P3/8-small)- [-1, 1, Conv, [256, 3, 2]]- [[-1, 13], 1, Concat, [1]] # cat head P4- [-1, 2, C3k2_MSCB2, [512, False]] # 19 (P4/16-medium)- [-1, 1, Conv, [512, 3, 2]]- [[-1, 10], 1, Concat, [1]] # cat head P5- [-1, 2, C3k2_MSCB2, [1024, True]] # 22 (P5/32-large)- [[16, 19, 22], 1, Detect, [nc]] # Detect(P3, P4, P5) 5.3 训练代码
大家可以创建一个py文件将我给的代码复制粘贴进去配置好自己的文件路径即可运行。 import warnings
warnings.filterwarnings(ignore)
from ultralytics import YOLOif __name__ __main__:model YOLO(模型配置文件)# 如何切换模型版本, 上面的ymal文件可以改为 yolov8s.yaml就是使用的v8s,# 类似某个改进的yaml文件名称为yolov8-XXX.yaml那么如果想使用其它版本就把上面的名称改为yolov8l-XXX.yaml即可改的是上面YOLO中间的名字不是配置文件的# model.load(yolov8n.pt) # 是否加载预训练权重,科研不建议大家加载否则很难提升精度model.train(datarC:\Users\Administrator\PycharmProjects\yolov5-master\yolov5-master\Construction Site Safety.v30-raw-images_latestversion.yolov8\data.yaml,# 如果大家任务是其它的ultralytics/cfg/default.yaml找到这里修改task可以改成detect, segment, classify, posecacheFalse,imgsz640,epochs150,single_clsFalse, # 是否是单类别检测batch16,close_mosaic0,workers0,device0,optimizerSGD, # using SGD# resumeruns/train/exp21/weights/last.pt, # 如过想续训就设置last.pt的地址ampTrue, # 如果出现训练损失为Nan可以关闭ampprojectruns/train,nameexp,) 5.4 训练过程截图 五、本文总结
到此本文的正式分享内容就结束了在这里给大家推荐我的YOLOv11改进有效涨点专栏本专栏目前为新开的平均质量分98分后期我会根据各种最新的前沿顶会进行论文复现也会对一些老的改进机制进行补充如果大家觉得本文帮助到你了订阅本专栏关注后续更多的更新~
