这个图是有点问题的,在GiraffeNeckV2代码中只有了5个Fusion Block(图中有6个)

https://github.com/tinyvision/DAMO-YOLO/blob/master/damo/base_models/necks/giraffe_fpn_btn.py

代码中只有5个CSPStage

所以我自己画了一个总体图,在github上提了个issue,得到了原作者的肯定

I think the pictures in your paper are not rigorous in several places · Issue #91 · tinyvision/DAMO-YOLO · GitHub

想要看懂Neck部分,只需要看懂Fusion Block在做什么就行了,其他部分和PAN差不太多

class CSPStage(nn.Module): def __init__(self, block_fn, ch_in, ch_hidden_ratio, ch_out, n, act='swish', spp=False):super(CSPStage, self).__init__()split_ratio = 2ch_first = int(ch_out // split_ratio)ch_mid = int(ch_out - ch_first)self.conv1 = ConvBNAct(ch_in, ch_first, 1, act=act)self.conv2 = ConvBNAct(ch_in, ch_mid, 1, act=act)self.convs = nn.Sequential()next_ch_in = ch_midfor i in range(n):if block_fn == 'BasicBlock_3x3_Reverse':self.convs.add_module(str(i),BasicBlock_3x3_Reverse(next_ch_in, ch_hidden_ratio, ch_mid, act=act, shortcut=True))else:raise NotImplementedErrorif i == (n - 1) // 2 and spp:self.convs.add_module('spp', SPP(ch_mid * 4, ch_mid, 1, [5, 9, 13], act=act))next_ch_in = ch_midself.conv3 = ConvBNAct(ch_mid * n + ch_first, ch_out, 1, act=act)def forward(self, x):y1 = self.conv1(x)y2 = self.conv2(x)mid_out = [y1]for conv in self.convs:y2 = conv(y2)mid_out.append(y2)y = torch.cat(mid_out, axis=1)y = self.conv3(y)return y

以上是CSPStage的代码,要想看懂,我们得先看懂ConvBNAct、BasicBlock_3x3_Reverse这两个类

class ConvBNAct(nn.Module):"""A Conv2d -> Batchnorm -> silu/leaky relu block"""def __init__(self,in_channels,out_channels,ksize,stride=1,groups=1,bias=False,act='silu',norm='bn',reparam=False,):super().__init__()# same paddingpad = (ksize - 1) // 2self.conv = nn.Conv2d(in_channels,out_channels,kernel_size=ksize,stride=stride,padding=pad,groups=groups,bias=bias,)if norm is not None:self.bn = get_norm(norm, out_channels, inplace=True)if act is not None:self.act = get_activation(act, inplace=True)self.with_norm = norm is not Noneself.with_act = act is not Nonedef forward(self, x):x = self.conv(x)if self.with_norm:x = self.bn(x)if self.with_act:x = self.act(x)return xdef fuseforward(self, x):return self.act(self.conv(x))

ConvBNAct还是很好看懂的,Conv +BN + SiLU就完事了(也可用别的激活函数,文章用SiLU)

如果设置了groups参数就变成了组卷积了

class BasicBlock_3x3_Reverse(nn.Module):def __init__(self, ch_in, ch_hidden_ratio, ch_out, act='relu', shortcut=True):super(BasicBlock_3x3_Reverse, self).__init__()assert ch_in == ch_outch_hidden = int(ch_in * ch_hidden_ratio)self.conv1 = ConvBNAct(ch_hidden, ch_out, 3, stride=1, act=act)self.conv2 = RepConv(ch_in, ch_hidden, 3, stride=1, act=act)self.shortcut = shortcutdef forward(self, x):y = self.conv2(x)y = self.conv1(y)if self.shortcut:return x + yelse:return y

要看懂BasicBlock_3x3_Reverse这个类,就得了解RepConv类,这个类就是根据RepVGG网络的RepVGGBlock改的

class RepConv(nn.Module):'''RepConv is a basic rep-style block, including training and deploy statusCode is based on https://github.com/DingXiaoH/RepVGG/blob/main/repvgg.py'''def __init__(self, in_channels, out_channels, kernel_size=3, stride=1, padding=1, dilation=1, groups=1, padding_mode='zeros', deploy=False, act='relu', norm=None):super(RepConv, self).__init__()self.deploy = deployself.groups = groupsself.in_channels = in_channelsself.out_channels = out_channelsassert kernel_size == 3assert padding == 1padding_11 = padding - kernel_size // 2if isinstance(act, str):self.nonlinearity = get_activation(act)else:self.nonlinearity = actif deploy:self.rbr_reparam = nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=kernel_size, stride=stride, padding=padding, dilation=dilation, groups=groups, bias=True, padding_mode=padding_mode)else:self.rbr_identity = Noneself.rbr_dense = conv_bn(in_channels=in_channels, out_channels=out_channels, kernel_size=kernel_size, stride=stride, padding=padding, groups=groups)self.rbr_1x1 = conv_bn(in_channels=in_channels, out_channels=out_channels, kernel_size=1, stride=stride, padding=padding_11, groups=groups)def forward(self, inputs):'''Forward process'''if hasattr(self, 'rbr_reparam'):return self.nonlinearity(self.rbr_reparam(inputs))if self.rbr_identity is None:id_out = 0else:id_out = self.rbr_identity(inputs)return self.nonlinearity(self.rbr_dense(inputs) + self.rbr_1x1(inputs) + id_out)def get_equivalent_kernel_bias(self):kernel3x3, bias3x3 = self._fuse_bn_tensor(self.rbr_dense)kernel1x1, bias1x1 = self._fuse_bn_tensor(self.rbr_1x1)kernelid, biasid = self._fuse_bn_tensor(self.rbr_identity)return kernel3x3 + self._pad_1x1_to_3x3_tensor(kernel1x1) + kernelid, bias3x3 + bias1x1 + biasiddef _pad_1x1_to_3x3_tensor(self, kernel1x1):if kernel1x1 is None:return 0else:return torch.nn.functional.pad(kernel1x1, [1, 1, 1, 1])def _fuse_bn_tensor(self, branch):if branch is None:return 0, 0if isinstance(branch, nn.Sequential):kernel = branch.conv.weightrunning_mean = branch.bn.running_meanrunning_var = branch.bn.running_vargamma = branch.bn.weightbeta = branch.bn.biaseps = branch.bn.epselse:assert isinstance(branch, nn.BatchNorm2d)if not hasattr(self, 'id_tensor'):input_dim = self.in_channels // self.groupskernel_value = np.zeros((self.in_channels, input_dim, 3, 3),dtype=np.float32)for i in range(self.in_channels):kernel_value[i, i % input_dim, 1, 1] = 1self.id_tensor = torch.from_numpy(kernel_value).to(branch.weight.device)kernel = self.id_tensorrunning_mean = branch.running_meanrunning_var = branch.running_vargamma = branch.weightbeta = branch.biaseps = branch.epsstd = (running_var + eps).sqrt()t = (gamma / std).reshape(-1, 1, 1, 1)return kernel * t, beta - running_mean * gamma / stddef switch_to_deploy(self):if hasattr(self, 'rbr_reparam'):returnkernel, bias = self.get_equivalent_kernel_bias()self.rbr_reparam = nn.Conv2d(in_channels=self.rbr_dense.conv.in_channels,out_channels=self.rbr_dense.conv.out_channels,kernel_size=self.rbr_dense.conv.kernel_size,stride=self.rbr_dense.conv.stride,padding=self.rbr_dense.conv.padding,dilation=self.rbr_dense.conv.dilation,groups=self.rbr_dense.conv.groups,bias=True)self.rbr_reparam.weight.data = kernelself.rbr_reparam.bias.data = biasfor para in self.parameters():para.detach_()self.__delattr__('rbr_dense')self.__delattr__('rbr_1x1')if hasattr(self, 'rbr_identity'):self.__delattr__('rbr_identity')if hasattr(self, 'id_tensor'):self.__delattr__('id_tensor')self.deploy = True

RepConv的特点是结构重参数化,训练时采用三条分支,推理时将三个分支融合在一起,大大减少了推理时间(建议看看RepVGG的讲解视频),我图画得太丑了

RepConv采用的两分支的结构(a)

其他细节有缘再更,代码不难,慢慢看完全能懂。有写的不对的地方请见谅