yolov7tiny引入轻量化骨干网络MobileOne（个人备忘录）

参考链接：http://t.csdn.cn/J7iCI

http://t.csdn.cn/bwtMV

一、在models→common.py文件中加入

#====MobileOne====#
import copy as copy2   # 为防止与common原来引入的copy冲突, for mobileone reparameterize
from typing import Optional, List, Tuple

class SEBlock(nn.Module):
    """ Squeeze and Excite module.
        https://arxiv.org/pdf/1709.01507.pdf
    """

    def __init__(self, in_channels: int, rd_ratio: float = 0.0625) -> None:
        """ Construct a Squeeze and Excite Module.
        :param in_channels: Number of input channels.
        :param rd_ratio: Input channel reduction ratio.
        """
        super(SEBlock, self).__init__()
        self.reduce = nn.Conv2d(in_channels=in_channels,out_channels=int(in_channels * rd_ratio), kernel_size=1, stride=1, bias=True)
        self.expand = nn.Conv2d(in_channels=int(in_channels * rd_ratio),out_channels=in_channels, kernel_size=1, stride=1, bias=True)

    def forward(self, inputs: torch.Tensor) -> torch.Tensor:
        """ Apply forward pass. """
        b, c, h, w = inputs.size()
        x = F.avg_pool2d(inputs, kernel_size=[h, w])
        x = self.reduce(x)
        x = F.relu(x)
        x = self.expand(x)
        x = torch.sigmoid(x)
        x = x.view(-1, c, 1, 1)
        return inputs * x


class MobileOneBlock(nn.Module):
    """ MobileOne building block. https://arxiv.org/pdf/2206.04040.pdf
    """
    def __init__(self, in_channels: int, out_channels: int, kernel_size: int, stride: int = 1,
                 padding: int = 0, dilation: int = 1, groups: int = 1, use_se: bool = False, num_conv_branches: int = 1, inference_mode: bool = False) -> None:
        """ Construct a MobileOneBlock module.
        :param in_channels: Number of channels in the input.
        :param out_channels: Number of channels produced by the block.
        :param kernel_size: Size of the convolution kernel.
        :param stride: Stride size.
        :param padding: Zero-padding size.
        :param dilation: Kernel dilation factor.
        :param groups: Group number.
        :param inference_mode: If True, instantiates model in inference mode.
        :param use_se: Whether to use SE-ReLU activations.
        :param num_conv_branches: Number of linear conv branches.
        """
        super(MobileOneBlock, self).__init__()
        self.inference_mode = inference_mode
        self.groups = groups
        self.stride = stride
        self.kernel_size = kernel_size
        self.in_channels = in_channels
        self.out_channels = out_channels
        self.num_conv_branches = num_conv_branches  # 4

        # Check if SE-ReLU is requested
        if use_se:
            self.se = SEBlock(out_channels)
        else:
            self.se = nn.Identity()
        self.activation = nn.ReLU()

        if inference_mode:
            self.reparam_conv = nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=kernel_size,stride=stride, padding=padding, dilation=dilation, groups=groups, bias=True)
        else:
            # Re-parameterizable skip connection
            self.rbr_skip = nn.BatchNorm2d(num_features=in_channels) if out_channels == in_channels and stride == 1 else None   # BN skip

            # Re-parameterizable conv branches
            rbr_conv = list()
            for _ in range(self.num_conv_branches):
                rbr_conv.append(self._conv_bn(kernel_size=kernel_size, padding=padding))
            self.rbr_conv = nn.ModuleList(rbr_conv)

            # Re-parameterizable scale branch
            self.rbr_scale = None
            if kernel_size > 1:
                self.rbr_scale = self._conv_bn(kernel_size=1, padding=0)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """ Apply forward pass. """
        # Inference mode forward pass.
        if self.inference_mode:
            return self.activation(self.se(self.reparam_conv(x)))

        # Multi-branched train-time forward pass.
        # Skip branch output
        identity_out = 0
        if self.rbr_skip is not None:
            identity_out = self.rbr_skip(x)

        # Scale branch output
        scale_out = 0
        if self.rbr_scale is not None:
            scale_out = self.rbr_scale(x)

        # Other branches
        out = scale_out + identity_out
        for ix in range(self.num_conv_branches):
            out += self.rbr_conv[ix](x)

        return self.activation(self.se(out))

    def reparameterize(self):
        """ Following works like `RepVGG: Making VGG-style ConvNets Great Again` -
        https://arxiv.org/pdf/2101.03697.pdf. We re-parameterize multi-branched
        architecture used at training time to obtain a plain CNN-like structure
        for inference.
        """
        if self.inference_mode:
            return
        kernel, bias = self._get_kernel_bias()
        self.reparam_conv = nn.Conv2d(in_channels=self.rbr_conv[0].conv.in_channels,
                                      out_channels=self.rbr_conv[0].conv.out_channels,
                                      kernel_size=self.rbr_conv[0].conv.kernel_size,
                                      stride=self.rbr_conv[0].conv.stride,
                                      padding=self.rbr_conv[0].conv.padding,
                                      dilation=self.rbr_conv[0].conv.dilation,
                                      groups=self.rbr_conv[0].conv.groups,
                                      bias=True)
        self.reparam_conv.weight.data = kernel
        self.reparam_conv.bias.data = bias

        # Delete un-used branches
        for para in self.parameters():
            para.detach_()
        self.__delattr__('rbr_conv')
        self.__delattr__('rbr_scale')
        if hasattr(self, 'rbr_skip'):
            self.__delattr__('rbr_skip')

        self.inference_mode = True

    def _get_kernel_bias(self) -> Tuple[torch.Tensor, torch.Tensor]:
        """ Method to obtain re-parameterized kernel and bias.
        Reference: https://github.com/DingXiaoH/RepVGG/blob/main/repvgg.py#L83
        :return: Tuple of (kernel, bias) after fusing branches.
        """
        # get weights and bias of scale branch
        kernel_scale = 0
        bias_scale = 0
        if self.rbr_scale is not None:
            kernel_scale, bias_scale = self._fuse_bn_tensor(self.rbr_scale)
            # Pad scale branch kernel to match conv branch kernel size.
            pad = self.kernel_size // 2
            kernel_scale = torch.nn.functional.pad(kernel_scale, [pad, pad, pad, pad])

        # get weights and bias of skip branch
        kernel_identity = 0
        bias_identity = 0
        if self.rbr_skip is not None:
            kernel_identity, bias_identity = self._fuse_bn_tensor(self.rbr_skip)

        # get weights and bias of conv branches
        kernel_conv = 0
        bias_conv = 0
        for ix in range(self.num_conv_branches):
            _kernel, _bias = self._fuse_bn_tensor(self.rbr_conv[ix])
            kernel_conv += _kernel
            bias_conv += _bias

        kernel_final = kernel_conv + kernel_scale + kernel_identity
        bias_final = bias_conv + bias_scale + bias_identity
        return kernel_final, bias_final

    def _fuse_bn_tensor(self, branch) -> Tuple[torch.Tensor, torch.Tensor]:
        """ Method to fuse batchnorm layer with preceeding conv layer.
        Reference: https://github.com/DingXiaoH/RepVGG/blob/main/repvgg.py#L95

        :param branch:
        :return: Tuple of (kernel, bias) after fusing batchnorm.
        """
        if isinstance(branch, nn.Sequential):
            kernel = branch.conv.weight
            running_mean = branch.bn.running_mean
            running_var = branch.bn.running_var
            gamma = branch.bn.weight
            beta = branch.bn.bias
            eps = branch.bn.eps
        else:
            assert isinstance(branch, nn.BatchNorm2d)
            if not hasattr(self, 'id_tensor'):
                input_dim = self.in_channels // self.groups
                kernel_value = torch.zeros((self.in_channels, input_dim, self.kernel_size, self.kernel_size),
                                           dtype=branch.weight.dtype, device=branch.weight.device)
                for i in range(self.in_channels):
                    kernel_value[i, i % input_dim,self.kernel_size // 2, self.kernel_size // 2] = 1
                self.id_tensor = kernel_value
            kernel = self.id_tensor
            running_mean = branch.running_mean
            running_var = branch.running_var
            gamma = branch.weight
            beta = branch.bias
            eps = branch.eps
        std = (running_var + eps).sqrt()
        t = (gamma / std).reshape(-1, 1, 1, 1)
        return kernel * t, beta - running_mean * gamma / std

    def _conv_bn(self, kernel_size: int, padding: int) -> nn.Sequential:
        """ Helper method to construct conv-batchnorm layers.
        :param kernel_size: Size of the convolution kernel.
        :param padding: Zero-padding size.
        :return: Conv-BN module.
        """
        mod_list = nn.Sequential()
        mod_list.add_module('conv', nn.Conv2d(in_channels=self.in_channels,out_channels=self.out_channels,
                                              kernel_size=kernel_size, stride=self.stride, padding=padding, groups=self.groups, bias=False))
        mod_list.add_module('bn', nn.BatchNorm2d(num_features=self.out_channels))
        return mod_list

class MobileOne(nn.Module):
    """ MobileOne Model  https://arxiv.org/pdf/2206.04040.pdf """
    def __init__(self,
                 in_channels, out_channels,
                 num_blocks_per_stage = 2, num_conv_branches: int = 1,
                 use_se: bool = False, num_se: int = 0,
                 inference_mode: bool = False, ) -> None:
        """ Construct MobileOne model.
        :param num_blocks_per_stage: List of number of blocks per stage.
        :param num_classes: Number of classes in the dataset.
        :param width_multipliers: List of width multiplier for blocks in a stage.
        :param inference_mode: If True, instantiates model in inference mode.
        :param use_se: Whether to use SE-ReLU activations.
        :param num_conv_branches: Number of linear conv branches.
        """
        super().__init__()
        self.inference_mode = inference_mode
        self.use_se = use_se
        self.num_conv_branches = num_conv_branches

        self.stage = self._make_stage(in_channels, out_channels, num_blocks_per_stage, num_se_blocks= num_se if use_se else 0)

    # planes指输出通道
    def _make_stage(self, in_channels, out_channels,  num_blocks: int, num_se_blocks: int) -> nn.Sequential:
        """ Build a stage of MobileOne model.

        :param planes: Number of output channels.
        :param num_blocks: Number of blocks in this stage.
        :param num_se_blocks: Number of SE blocks in this stage.
        :return: A stage of MobileOne model.
        """
        # Get strides for all layers
        strides = [2] + [1]*(num_blocks-1)
        blocks = []
        for ix, stride in enumerate(strides):  # 用于训练几个blocks
            use_se = False
            if num_se_blocks > num_blocks:
                raise ValueError("Number of SE blocks cannot " "exceed number of layers.")
            if ix >= (num_blocks - num_se_blocks):
                use_se = True

            # Depthwise conv
            blocks.append(MobileOneBlock(in_channels=in_channels, out_channels=in_channels,
                                         kernel_size=3, stride=stride, padding=1, groups=in_channels,
                                         inference_mode=self.inference_mode, use_se=use_se, num_conv_branches=self.num_conv_branches))
            # Pointwise conv
            blocks.append(MobileOneBlock(in_channels=in_channels, out_channels=out_channels,
                                         kernel_size=1, stride=1, padding=0, groups=1,
                                         inference_mode=self.inference_mode, use_se=use_se, num_conv_branches=self.num_conv_branches))
            in_channels = out_channels
        return nn.Sequential(*blocks)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """ Apply forward pass. """
        x = self.stage(x)
        return x

def reparameterize_model(model: torch.nn.Module) -> nn.Module:
    """ Method returns a model where a multi-branched structure
        used in training is re-parameterized into a single branch
        for inference.

    :param model: MobileOne model in train mode.
    :return: MobileOne model in inference mode.
    """
    # Avoid editing original graph
    model = copy2.deepcopy(model)
    for module in model.modules():
        if hasattr(module, 'reparameterize'):
            module.reparameterize()
    return model

二、在models→yolo.py文件中，def parse_model(d, ch):里面加入

  elif m in [MobileOneBlock, MobileOne]:
            c1, c2 = ch[f], args[0]
            args = [c1, c2, *args[1:]]

位置如图：

 三、写新的yolov7tiny.yaml文件

# parameters
nc: 3  # number of classes
depth_multiple: 1.0  # model depth multiple
width_multiple: 1.0  # layer channel multiple

# anchors
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

# yolov7-tiny backbone
backbone:
  # [from, number, module, args] c2, k=1, s=1, p=None, g=1, act=True
  [
    [-1, 1, MobileOneBlock, [48, 3, 2, 1]],           # 0

    [-1, 1, MobileOne, [48, 2, 4, False, 0]],   # MobileOne [out_channels, num_blocks, num_conv_branches, use_se, num_se, inference_mode]

    [-1, 1, MobileOne, [128, 8, 4, False, 0]],

    [-1, 1, MobileOne, [256, 10, 4, False, 0]],

    [ -1, 1, MobileOne, [512, 1, 4, False, 0]],  # 4

  ]

# yolov7-tiny head
head:
  [[-1, 1, Conv, [256, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-2, 1, Conv, [256, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, SP, [5]],
   [-2, 1, SP, [9]],
   [-3, 1, SP, [13]],
   [[-1, -2, -3, -4], 1, Concat, [1]],
   [-1, 1, Conv, [256, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, -7], 1, Concat, [1]],
   [-1, 1, Conv, [256, 1, 1, None, 1, nn.LeakyReLU(0.1)]],  # 13


   [-1, 1, Conv, [128, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [3, 1, Conv, [128, 1, 1, None, 1, nn.LeakyReLU(0.1)]], # route backbone P4
   [[-1, -2], 1, Concat, [1]],

   [-1, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-2, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [64, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [64, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, -2, -3, -4], 1, Concat, [1]],
   [-1, 1, Conv, [128, 1, 1, None, 1, nn.LeakyReLU(0.1)]],  # 23

   [-1, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [2, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, -2], 1, Concat, [1]],                              # 27

   [-1, 1, Conv, [32, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-2, 1, Conv, [32, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [32, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [32, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, -2, -3, -4], 1, Concat, [1]],
   [-1, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],  # 33

   [-1, 1, Conv, [128, 3, 2, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, 23], 1, Concat, [1]],

   [-1, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-2, 1, Conv, [64, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [64, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [64, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, -2, -3, -4], 1, Concat, [1]],
   [-1, 1, Conv, [128, 1, 1, None, 1, nn.LeakyReLU(0.1)]],  # 41

   [-1, 1, Conv, [256, 3, 2, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, 13], 1, Concat, [1]],

   [-1, 1, Conv, [128, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-2, 1, Conv, [128, 1, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [128, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [-1, 1, Conv, [128, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [[-1, -2, -3, -4], 1, Concat, [1]],
   [-1, 1, Conv, [256, 1, 1, None, 1, nn.LeakyReLU(0.1)]],  # 49

   [33, 1, Conv, [128, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [41, 1, Conv, [256, 3, 1, None, 1, nn.LeakyReLU(0.1)]],
   [49, 1, Conv, [512, 3, 1, None, 1, nn.LeakyReLU(0.1)]], # 52

   [[50,51,52], 1, IDetect, [nc, anchors]],   # Detect(P3, P4, P5)
  ]

四、继续在yolo.py文件中fuse中添加如下部分

            elif isinstance(m, (MobileOne, MobileOneBlock)) and hasattr(m, 'reparameterize'):
                m.reparameterize()

位置：

img-size默认是[640, 640]，通常数据集采用这个尺寸就行了，img-size值越大，占用显存就越多，训练速度就更慢，对显卡的要求就越高，这次我用[320，320]

我的模型参数量：7295872

23.9 GFLOPS

GPU消耗0.898G

最好结果p：0.337      r：0.998     map0.5：0.339    怪怪的，先搁着