YOLOV8中加入EMA注意力机制和自创可扩展卷积使得小目标检测提点

1、自创可扩展卷积模块

可扩展卷积与一般卷积的区别是，可以得到不同感受野的大小。这里使用三分支不同的可扩展卷积，对特征图进行提取，得到感受野不同的特征图，有助于图中的小目标信息提取

class RFBconv(nn.Module):
    def __init__(self, c1):
        super(RFBconv, self).__init__()
        # 第一条分支
        c2 = c1
        self.branch1 = nn.Sequential(
            Conv(c1,c2,1,1),
            nn.Conv2d(c2, c2, kernel_size=3, dilation=1, padding=1)
        )

        # 第二条分支
        self.branch2 = nn.Sequential(
            Conv(c1, c2, 1,1),
            nn.Conv2d(c2, c2, kernel_size=3, dilation=1, padding=1),
            nn.Conv2d(c2, c2, kernel_size=3, dilation=3, padding=3)
        )

        # 第三条分支
        self.branch3 = nn.Sequential(
            Conv(c1, c2, 1,1),
            nn.Conv2d(c2, c2, kernel_size=5, dilation=1, padding=2),
            nn.Conv2d(c2, c2, kernel_size=3, dilation=5, padding=5)
        )

        # 1x1卷积用于通道数变换
        self.conv1x1 = Conv(3 * c2, c2, 1)

    def forward(self, x):
        # 分别对三条分支进行前向传播
        out1 = self.branch1(x)
        out2 = self.branch2(x)
        out3 = self.branch3(x)

        # 将三个分支的结果连接在一起
        concatenated = torch.cat((out1, out2, out3), dim=1)

        # 使用1x1卷积进行通道数变换
        result = self.conv1x1(concatenated)
        result = result + x

        return result

2、EMA注意力机制

EMA是CBAM注意力的改进，比CBAM注意力更好

class EMA(nn.Module):
    def __init__(self, channels, factor=8):
        super(EMA, self).__init__()
        self.groups = factor
        assert channels // self.groups > 0
        self.softmax = nn.Softmax(-1)
        self.agp = nn.AdaptiveAvgPool2d((1, 1))
        self.pool_h = nn.AdaptiveAvgPool2d((None, 1))
        self.pool_w = nn.AdaptiveAvgPool2d((1, None))
        self.gn = nn.GroupNorm(channels // self.groups, channels // self.groups)
        self.conv1x1 = nn.Conv2d(channels // self.groups, channels // self.groups, kernel_size=1, stride=1, padding=0)
        self.conv3x3 = nn.Conv2d(channels // self.groups, channels // self.groups, kernel_size=3, stride=1, padding=1)

    def forward(self, x):
        b, c, h, w = x.size()
        group_x = x.reshape(b * self.groups, -1, h, w)  # b*g,c//g,h,w
        x_h = self.pool_h(group_x)
        x_w = self.pool_w(group_x).permute(0, 1, 3, 2)
        hw = self.conv1x1(torch.cat([x_h, x_w], dim=2))
        x_h, x_w = torch.split(hw, [h, w], dim=2)
        x1 = self.gn(group_x * x_h.sigmoid() * x_w.permute(0, 1, 3, 2).sigmoid())
        x2 = self.conv3x3(group_x)
        x11 = self.softmax(self.agp(x1).reshape(b * self.groups, -1, 1).permute(0, 2, 1))
        x12 = x2.reshape(b * self.groups, c // self.groups, -1)  # b*g, c//g, hw
        x21 = self.softmax(self.agp(x2).reshape(b * self.groups, -1, 1).permute(0, 2, 1))
        x22 = x1.reshape(b * self.groups, c // self.groups, -1)  # b*g, c//g, hw
        weights = (torch.matmul(x11, x12) + torch.matmul(x21, x22)).reshape(b * self.groups, 1, h, w)
        return (group_x * weights.sigmoid()).reshape(b, c, h, w)

3、配置文件

# Ultralytics YOLO 🚀, AGPL-3.0 license
# YOLOv8 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect

# Parameters
nc: 20  # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov8n.yaml' will call yolov8.yaml with scale 'n'
  # [depth, width, max_channels]
  # YOLOv8n summary: 225 layers,  3157200 parameters,  3157184 gradients,   8.9 GFLOPs
#  l: [1.00, 1.00, 512]  # YOLOv8s summary: 225 layers, 11166560 parameters, 11166544 gradients,  28.8 GFLOPs
  # s: [0.33, 0.50, 1024]
backbone:
  # [from, repeats, module, args]
  - [-1, 1, Conv, [64, 3, 2]]  # 0-P1/2
  - [-1, 1, Conv, [128, 3, 2]]  # 1-P2/4
  - [-1, 3, C2f, [128, True]]
  - [-1, 1, Conv, [256, 3, 2]]  # 3-P3/8
  - [-1, 6, C2f, [256, True]]
  - [-1, 1, Conv, [512, 3, 2]]  # 5-P4/16
  - [-1, 6, C2f, [512, True]]
  - [-1, 1, Conv, [1024, 3, 2]]  # 7-P5/32
  - [-1, 3, C2f, [1024, True]]
  - [-1, 1, SPPF, [1024, 5]]  # 9

# YOLOv8.0s head
head:
  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
  - [-1,1,EMA,[]]
  - [[-1, 6], 1, Concat, [1]]  # cat backbone P4
  - [-1, 3, C2f, [512]]  # 13

  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
  - [-1,1,EMA,[]]
  - [[-1, 4], 1, Concat, [1]]  # cat backbone P3
  - [-1, 3, C2f, [256]]  # 17 (P3/8-small)

  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
  - [-1,1,EMA,[]]
  - [[-1, 2], 1, Concat, [1]]  # cat backbone P3
  - [-1, 3, C2f, [128]]  # 18 (P2/4-xsmall)
  - [-1,1,RFBconv,[]]  #  22

  - [-1, 1, Conv, [256, 3, 2]]
  - [[-1, 17], 1, Concat, [1]]  # cat head P4
  - [-1, 3, C2f, [256]]  # 21 (P3/8-small)
  - [-1,1,RFBconv,[]]  # 26

  - [-1, 1, Conv, [512, 3, 2]]
  - [[-1, 13], 1, Concat, [1]]  # cat head P5
  - [-1, 3, C2f, [512]]  # 29 (P4/16-medium)

  - [-1, 1, Conv, [512, 3, 2]]
  - [[-1, 9], 1, Concat, [1]]  # cat head P5
  - [-1, 3, C2f, [1024]]  # 32 (P5/32-large)

  - [[22, 26, 29, 32], 1, Detect, [nc]]  # Detect(P2, P3, P4, P5)