目录

1.RFB-Net介绍

2. RFB引入到yolov5

2.1修改commmon.py

2.2 修改yolo.py

2.3 yolov5s_REF.yaml

---

1.RFB-Net介绍

 论文：https://arxiv.org/pdf/1711.07767.pdf

 代码：https://github.com/ruinmessi/RFBNet

         受启发于人类视觉的Receptive Fields结构，本文提出RFB，将RFs的尺度、离心率纳入考虑范围，使用轻量级主干网也能提取到高判别性特征，使得检测器速度快、精度高；具体地，RFB基于RFs的不同尺度，使用不同的卷积核，设计了多分支的conv、pooling操作（makes use of multi-branch pooling with varying kernels），并通过空洞卷积（dilated conv）来控制感受野的离心率，最后一步reshape操作后，形成生成的特征

 RFs也已被深入研究，如Inception、ASPP、Deformable CNN：

RFB模块是一个多分支的卷积模块，它的内部结构被划分为两部分：

1.多分支卷积层：根据RF的定义，使用多种尺寸的卷积核来实现比固定尺寸更好。具体设计：1.瓶颈结构，1x1-s2的卷积减少通道特征，然后加上一个nxn卷积。2.用5x5卷积替换为2个3x3的卷积去减少参数，这样可得到非线性结构更好的层。3.为了输出，卷积经常有stride=2或者是减少通道，所有直连层为了匹配维度用一个不带激活函数的1x1卷积层。
2.dilated 卷积层：在保持参数量可扩大感受野，用来获取更高分辨率的特征。下图展示了两种RFB结构：RFB和RFB-s。每个分支都是一个正常卷积后面加一个dilated卷积，主要尺寸和dilated因子不同。（a）RFB整体上借鉴了Inception的思想，主要不同点在于引入了3个dilated卷积层。（b）RFB-s和RFB相比主要有两个改进，一方面用3x3的卷积层代替5x5卷积层，另一方面用1x3和3x1的卷积来代替3x3卷积，主要目的是为了减少计算量，类似Inception后期版本对Inception结构的改进。

 实验结果

RFB模块：在table 2中，原始的SSD300实现了77.2%的mAP，通过简单的用RFB-max Pooling替代最后一个卷积层，我们将结果提升到了79.1%，获得了1.9%的提高，这表明了RFB模块的高效性。

2. RFB引入到yolov5

2.1修改common.py

class BasicRFB(nn.Module):

    def __init__(self, in_planes, out_planes, stride=1, scale=0.1, map_reduce=8, vision=1, groups=1):
        super(BasicRFB, self).__init__()
        self.scale = scale
        self.out_channels = out_planes
        inter_planes = in_planes // map_reduce

        self.branch0 = nn.Sequential(
            BasicConv(in_planes, inter_planes, kernel_size=1, stride=1, groups=groups, relu=False),
            BasicConv(inter_planes, 2 * inter_planes, kernel_size=(3, 3), stride=stride, padding=(1, 1), groups=groups),
            BasicConv(2 * inter_planes, 2 * inter_planes, kernel_size=3, stride=1, padding=vision + 1,
                      dilation=vision + 1, relu=False, groups=groups)
        )
        self.branch1 = nn.Sequential(
            BasicConv(in_planes, inter_planes, kernel_size=1, stride=1, groups=groups, relu=False),
            BasicConv(inter_planes, 2 * inter_planes, kernel_size=(3, 3), stride=stride, padding=(1, 1), groups=groups),
            BasicConv(2 * inter_planes, 2 * inter_planes, kernel_size=3, stride=1, padding=vision + 2,
                      dilation=vision + 2, relu=False, groups=groups)
        )
        self.branch2 = nn.Sequential(
            BasicConv(in_planes, inter_planes, kernel_size=1, stride=1, groups=groups, relu=False),
            BasicConv(inter_planes, (inter_planes // 2) * 3, kernel_size=3, stride=1, padding=1, groups=groups),
            BasicConv((inter_planes // 2) * 3, 2 * inter_planes, kernel_size=3, stride=stride, padding=1,
                      groups=groups),
            BasicConv(2 * inter_planes, 2 * inter_planes, kernel_size=3, stride=1, padding=vision + 4,
                      dilation=vision + 4, relu=False, groups=groups)
        )

        self.ConvLinear = BasicConv(6 * inter_planes, out_planes, kernel_size=1, stride=1, relu=False)
        self.shortcut = BasicConv(in_planes, out_planes, kernel_size=1, stride=stride, relu=False)
        self.relu = nn.ReLU(inplace=False)

    def forward(self, x):
        x0 = self.branch0(x)
        x1 = self.branch1(x)
        x2 = self.branch2(x)

        out = torch.cat((x0, x1, x2), 1)
        out = self.ConvLinear(out)
        short = self.shortcut(x)
        out = out * self.scale + short
        out = self.relu(out)

        return out

2.2 修改yolo.py

if m in {
                Conv, GhostConv, Bottleneck, GhostBottleneck, SPP, SPPF,DWConv, MixConv2d, Focus, CrossConv,
                BottleneckCSP, C3, C3TR, C3SPP, C3Ghost, CNeB, nn.ConvTranspose2d, DWConvTranspose2d, C3x, C2f,BasicRFB}

2.3 yolov5s_REF.yaml3.

# YOLOv5 🚀 by Ultralytics, GPL-3.0 license

# Parameters
nc: 1  # number of classes
depth_multiple: 0.33  # model depth multiple
width_multiple: 0.50  # 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
# CSPNet-v5
backbone:
  # [from, number, module, args]
  [[-1, 1, Conv, [64, 6, 2, 2]],  # 0-P1/2
   [-1, 1, Conv, [128, 3, 2]],  # 1-P2/4
   [-1, 3, C3, [128]],
   [-1, 1, Conv, [256, 3, 2]],  # 3-P3/8
   [-1, 6, C3, [256]],
   [-1, 1, Conv, [512, 3, 2]],  # 5-P4/16
   [-1, 9, C3, [512]],
   [-1, 1, Conv, [1024, 3, 2]],  # 7-P5/32
   [-1, 3, C3, [1024]],
   [-1, 1, SPPF, [1024, 5]],  # 9
  ]

# YOLOv5 v6.0 head
head:
  [[-1, 1, Conv, [512, 1, 1]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [[-1, 6], 1, Concat, [1]],  # cat backbone P4
   [-1, 3, C3, [512, False]],  # 13

   [-1, 1, Conv, [256, 1, 1]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [[-1, 4], 1, Concat, [1]],  # cat backbone P3
   [-1, 3, C3, [256, False]],  # 17 (P3/8-small)
   [-1, 1, BasicRFB, [256]],  # 18

   [-1, 1, Conv, [256, 3, 2]],
   [[-1, 14], 1, Concat, [1]],  # cat head P4
   [-1, 3, C3, [512, False]],  # 21 (P4/16-medium)
   [-1, 1, BasicRFB, [512]], # 22

   [-1, 1, Conv, [512, 3, 2]],
   [[-1, 10], 1, Concat, [1]],  # cat head P5
   [-1, 3, C3, [1024, False]],  # 25 (P5/32-large)
   [-1, 1, BasicRFB, [1024]], # 26

   [[18, 22, 26], 1, Detect, [nc, anchors]],  # Detect(P3, P4, P5)
  ]