yolov7 图解

Ctrl CV

keep learning

8 人赞同了该文章

一、yolov7网络架构图

二、yolov7 ymal， ELAN结构分析

对照论文中的ELAN结构

三、E-ELAN结构

对比yolov7 e6 yaml 与 yolov7 e6e yaml

yolov7 模型缩放

MP1

Conv块 又称为 CBS

class Conv(nn.Module):
    # Standard convolution
    def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):  # ch_in, ch_out, kernel, stride, padding, groups
        super(Conv, self).__init__()
        self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False)
        self.bn = nn.BatchNorm2d(c2)
        self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())
def forward(self, x):
    return self.act(self.bn(self.conv(x)))

def fuseforward(self, x):
    return self.act(self.conv(x))
</code></pre></div><p class="ztext-empty-paragraph"><br></p><p data-pid="EnzX62aN">sppcspc结构</p><figure data-size="normal"><noscript><img src="https://pic1.zhimg.com/v2-5fc005fc207fc6b7c419437045b81774_b.jpg" data-caption="" data-size="normal" data-rawwidth="2174" data-rawheight="1196" class="origin_image zh-lightbox-thumb" width="2174" data-original="https://pic1.zhimg.com/v2-5fc005fc207fc6b7c419437045b81774_r.jpg"/></noscript><div><img src="https://pic1.zhimg.com/80/v2-5fc005fc207fc6b7c419437045b81774_1440w.webp" data-caption="" data-size="normal" data-rawwidth="2174" data-rawheight="1196" class="origin_image zh-lightbox-thumb lazy" width="2174" data-original="https://pic1.zhimg.com/v2-5fc005fc207fc6b7c419437045b81774_r.jpg" data-actualsrc="https://pic1.zhimg.com/v2-5fc005fc207fc6b7c419437045b81774_b.jpg" data-original-token="v2-21a5d69f0847febff325c6443c8eda95" height="1196" data-lazy-status="ok" style=""></div></figure><div class="highlight"><pre><code class="language-text">##### cspnet #####

class SPPCSPC(nn.Module):
# CSP https://github.com/WongKinYiu/CrossStagePartialNetworks
def init(self, c1, c2, n=1, shortcut=False, g=1, e=0.5, k=(5, 9, 13)):
super(SPPCSPC, self).init()
c_ = int(2 * c2 * e) # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
self.cv2 = Conv(c1, c_, 1, 1)
self.cv3 = Conv(c_, c_, 3, 1)
self.cv4 = Conv(c_, c_, 1, 1)
self.m = nn.ModuleList([nn.MaxPool2d(kernel_size=x, stride=1, padding=x // 2) for x in k])
self.cv5 = Conv(4 * c_, c_, 1, 1)
self.cv6 = Conv(c_, c_, 3, 1)
self.cv7 = Conv(2 * c_, c2, 1, 1)

def forward(self, x):
    x1 = self.cv4(self.cv3(self.cv1(x)))
    y1 = self.cv6(self.cv5(torch.cat([x1] + [m(x1) for m in self.m], 1)))
    y2 = self.cv2(x)
    return self.cv7(torch.cat((y1, y2), dim=1))</code></pre></div><p data-pid="71_j5DnC">比较之前的ELAN模块这里有所不同：</p><figure data-size="normal"><noscript><img src="https://pic2.zhimg.com/v2-655133b74bdcc93a3900049e46ba9099_b.jpg" data-caption="" data-size="normal" data-rawwidth="2184" data-rawheight="1232" class="origin_image zh-lightbox-thumb" width="2184" data-original="https://pic2.zhimg.com/v2-655133b74bdcc93a3900049e46ba9099_r.jpg"/></noscript><div><img src="https://pic2.zhimg.com/80/v2-655133b74bdcc93a3900049e46ba9099_1440w.webp" data-caption="" data-size="normal" data-rawwidth="2184" data-rawheight="1232" class="origin_image zh-lightbox-thumb lazy" width="2184" data-original="https://pic2.zhimg.com/v2-655133b74bdcc93a3900049e46ba9099_r.jpg" data-actualsrc="https://pic2.zhimg.com/v2-655133b74bdcc93a3900049e46ba9099_b.jpg" data-original-token="v2-0d281003e9e4eedb43c6eb0a3a97e283" height="1232" data-lazy-status="ok" style=""></div></figure><figure data-size="normal"><noscript><img src="https://pic2.zhimg.com/v2-5c0d0730892cd5b92c3d36023e16b99d_b.jpg" data-caption="" data-size="normal" data-rawwidth="2254" data-rawheight="922" class="origin_image zh-lightbox-thumb" width="2254" data-original="https://pic2.zhimg.com/v2-5c0d0730892cd5b92c3d36023e16b99d_r.jpg"/></noscript><div><img src="https://pic2.zhimg.com/80/v2-5c0d0730892cd5b92c3d36023e16b99d_1440w.webp" data-caption="" data-size="normal" data-rawwidth="2254" data-rawheight="922" class="origin_image zh-lightbox-thumb lazy" width="2254" data-original="https://pic2.zhimg.com/v2-5c0d0730892cd5b92c3d36023e16b99d_r.jpg" data-actualsrc="https://pic2.zhimg.com/v2-5c0d0730892cd5b92c3d36023e16b99d_b.jpg" data-original-token="v2-34d09e564a779cfbfdf2c350f84d40cd" height="922" data-lazy-status="ok" style=""></div></figure><p data-pid="6ZU1zHOU">MP2</p><figure data-size="normal"><noscript><img src="https://pic4.zhimg.com/v2-8c544179619f143db15fee02cfb2ddff_b.jpg" data-caption="" data-size="normal" data-rawwidth="1994" data-rawheight="1140" class="origin_image zh-lightbox-thumb" width="1994" data-original="https://pic4.zhimg.com/v2-8c544179619f143db15fee02cfb2ddff_r.jpg"/></noscript><div><img src="https://pic4.zhimg.com/80/v2-8c544179619f143db15fee02cfb2ddff_1440w.webp" data-caption="" data-size="normal" data-rawwidth="1994" data-rawheight="1140" class="origin_image zh-lightbox-thumb lazy" width="1994" data-original="https://pic4.zhimg.com/v2-8c544179619f143db15fee02cfb2ddff_r.jpg" data-actualsrc="https://pic4.zhimg.com/v2-8c544179619f143db15fee02cfb2ddff_b.jpg" data-original-token="v2-69aaab5f7505458b1741617cb7242003" height="1140" data-lazy-status="ok" style=""></div></figure><p data-pid="CjInuHPu">比较MP1  MP2</p><figure data-size="normal"><noscript><img src="https://pic2.zhimg.com/v2-2a01daa475923528d2d9fd823abc8a39_b.jpg" data-caption="" data-size="normal" data-rawwidth="1966" data-rawheight="752" class="origin_image zh-lightbox-thumb" width="1966" data-original="https://pic2.zhimg.com/v2-2a01daa475923528d2d9fd823abc8a39_r.jpg"/></noscript><div><img src="https://pic2.zhimg.com/80/v2-2a01daa475923528d2d9fd823abc8a39_1440w.webp" data-caption="" data-size="normal" data-rawwidth="1966" data-rawheight="752" class="origin_image zh-lightbox-thumb lazy" width="1966" data-original="https://pic2.zhimg.com/v2-2a01daa475923528d2d9fd823abc8a39_r.jpg" data-actualsrc="https://pic2.zhimg.com/v2-2a01daa475923528d2d9fd823abc8a39_b.jpg" data-original-token="v2-74e23704aba0fa33820c2783be2aad2a" height="752" data-lazy-status="ok" style=""></div></figure><p data-pid="4JSmLSxl">RepConv结构</p><div class="highlight"><pre><code class="language-text">##### repvgg #####

class RepConv(nn.Module):
# Represented convolution
# https://arxiv.org/abs/2101.03697

def __init__(self, c1, c2, k=3, s=1, p=None, g=1, act=True, deploy=False):
    super(RepConv, self).__init__()

    self.deploy = deploy
    self.groups = g
    self.in_channels = c1
    self.out_channels = c2

    assert k == 3
    assert autopad(k, p) == 1

    padding_11 = autopad(k, p) - k // 2

    self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())

    if deploy:
        self.rbr_reparam = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=True)

    else:
        self.rbr_identity = (nn.BatchNorm2d(num_features=c1) if c2 == c1 and s == 1 else None)

        self.rbr_dense = nn.Sequential(
            nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False),
            nn.BatchNorm2d(num_features=c2),
        )

        self.rbr_1x1 = nn.Sequential(
            nn.Conv2d( c1, c2, 1, s, padding_11, groups=g, bias=False),
            nn.BatchNorm2d(num_features=c2),
        )

def forward(self, inputs):
    if hasattr(self, "rbr_reparam"):
        return self.act(self.rbr_reparam(inputs))

    if self.rbr_identity is None:
        id_out = 0
    else:
        id_out = self.rbr_identity(inputs)

    return self.act(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 + biasid,
    )

def _pad_1x1_to_3x3_tensor(self, kernel1x1):
    if kernel1x1 is None:
        return 0
    else:
        return nn.functional.pad(kernel1x1, [1, 1, 1, 1])

def _fuse_bn_tensor(self, branch):
    if branch is None:
        return 0, 0
    if isinstance(branch, nn.Sequential):
        kernel = branch[0].weight
        running_mean = branch[1].running_mean
        running_var = branch[1].running_var
        gamma = branch[1].weight
        beta = branch[1].bias
        eps = branch[1].eps
    else:
        assert isinstance(branch, nn.BatchNorm2d)
        if not hasattr(self, "id_tensor"):
            input_dim = self.in_channels // self.groups
            kernel_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] = 1
            self.id_tensor = torch.from_numpy(kernel_value).to(branch.weight.device)
        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 repvgg_convert(self):
    kernel, bias = self.get_equivalent_kernel_bias()
    return (
        kernel.detach().cpu().numpy(),
        bias.detach().cpu().numpy(),
    )

def fuse_conv_bn(self, conv, bn):

    std = (bn.running_var + bn.eps).sqrt()
    bias = bn.bias - bn.running_mean * bn.weight / std

    t = (bn.weight / std).reshape(-1, 1, 1, 1)
    weights = conv.weight * t

    bn = nn.Identity()
    conv = nn.Conv2d(in_channels = conv.in_channels,
                          out_channels = conv.out_channels,
                          kernel_size = conv.kernel_size,
                          stride=conv.stride,
                          padding = conv.padding,
                          dilation = conv.dilation,
                          groups = conv.groups,
                          bias = True,
                          padding_mode = conv.padding_mode)

    conv.weight = torch.nn.Parameter(weights)
    conv.bias = torch.nn.Parameter(bias)
    return conv

def fuse_repvgg_block(self):    
    if self.deploy:
        return
    print(f"RepConv.fuse_repvgg_block")
            
    self.rbr_dense = self.fuse_conv_bn(self.rbr_dense[0], self.rbr_dense[1])
    
    self.rbr_1x1 = self.fuse_conv_bn(self.rbr_1x1[0], self.rbr_1x1[1])
    rbr_1x1_bias = self.rbr_1x1.bias
    weight_1x1_expanded = torch.nn.functional.pad(self.rbr_1x1.weight, [1, 1, 1, 1])
    
    # Fuse self.rbr_identity
    if (isinstance(self.rbr_identity, nn.BatchNorm2d) or isinstance(self.rbr_identity, nn.modules.batchnorm.SyncBatchNorm)):
        # print(f"fuse: rbr_identity == BatchNorm2d or SyncBatchNorm")
        identity_conv_1x1 = nn.Conv2d(
                in_channels=self.in_channels,
                out_channels=self.out_channels,
                kernel_size=1,
                stride=1,
                padding=0,
                groups=self.groups, 
                bias=False)
        identity_conv_1x1.weight.data = identity_conv_1x1.weight.data.to(self.rbr_1x1.weight.data.device)
        identity_conv_1x1.weight.data = identity_conv_1x1.weight.data.squeeze().squeeze()
        # print(f" identity_conv_1x1.weight = {identity_conv_1x1.weight.shape}")
        identity_conv_1x1.weight.data.fill_(0.0)
        identity_conv_1x1.weight.data.fill_diagonal_(1.0)
        identity_conv_1x1.weight.data = identity_conv_1x1.weight.data.unsqueeze(2).unsqueeze(3)
        # print(f" identity_conv_1x1.weight = {identity_conv_1x1.weight.shape}")

        identity_conv_1x1 = self.fuse_conv_bn(identity_conv_1x1, self.rbr_identity)
        bias_identity_expanded = identity_conv_1x1.bias
        weight_identity_expanded = torch.nn.functional.pad(identity_conv_1x1.weight, [1, 1, 1, 1])            
    else:
        # print(f"fuse: rbr_identity != BatchNorm2d, rbr_identity = {self.rbr_identity}")
        bias_identity_expanded = torch.nn.Parameter( torch.zeros_like(rbr_1x1_bias) )
        weight_identity_expanded = torch.nn.Parameter( torch.zeros_like(weight_1x1_expanded) )            
    

    #print(f"self.rbr_1x1.weight = {self.rbr_1x1.weight.shape}, ")
    #print(f"weight_1x1_expanded = {weight_1x1_expanded.shape}, ")
    #print(f"self.rbr_dense.weight = {self.rbr_dense.weight.shape}, ")

    self.rbr_dense.weight = torch.nn.Parameter(self.rbr_dense.weight + weight_1x1_expanded + weight_identity_expanded)
    self.rbr_dense.bias = torch.nn.Parameter(self.rbr_dense.bias + rbr_1x1_bias + bias_identity_expanded)
            
    self.rbr_reparam = self.rbr_dense
    self.deploy = True

    if self.rbr_identity is not None:
        del self.rbr_identity
        self.rbr_identity = None

    if self.rbr_1x1 is not None:
        del self.rbr_1x1
        self.rbr_1x1 = None

    if self.rbr_dense is not None:
        del self.rbr_dense
        self.rbr_dense = None

yolov7-w6

ReOrg模块

yolov7-e6

DowmC结构

yolov7-d6

yolov7e6e

yolov7-tiny

yolov7训练自己的数据集与yolov5类似，参考之前的文章。

发布于 2023-07-03 20:06 ・IP 属地江苏

YOLO算法

YOLOv7

目标检测

​ 赞同 8​ ​ 2 条评论

​ 分享

​ 喜欢 ​ 收藏 ​ 申请转载

​