YOLOv5是Ultralytics公司的开源项目,GitHub地址:https://github.com/ultralytics/yolov5,更新速度非常快,最新版的v6.1于2022年2月22日发布,目前star数22.7k。
YOLOv5更新日志:
下面对v6.x版本的网络模型及各个组件,结合源码和网络框图进行解析。
可以看出,相比于之前v5.x,最新版的v6.x网络结构更加精简(以提高速度和推理性能),主要有以下更新:
YOLOv5网络搭建的各个组件主要在model/common.py文件中
网络中的标准卷积层,有2D卷积+BN层+激活函数(SiLU)组成,在之后的Bottleneck、C3、SPPF等结构中都会被调用。
源码:
# 标准卷积操作:conv2D+BN+SiLU
# 在Focus、Bottleneck、BottleneckCSP、C3、SPP、DWConv、TransformerBloc等模块中调用
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().__init__()
self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False)
self.bn = nn.BatchNorm2d(c2)
# 这里的nn.Identity()不改变input,直接return input
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)))
# 前向加速推理模块
# 用于Model类的fuse函数,融合conv+bn 加速推理 一般用于测试/验证阶段
def forward_fuse(self, x):
return self.act(self.conv(x))
Focus模块是作者自己设计出来,为了减少浮点数和提高速度,而不是增加featuremap的,本质就是将图像进行切片,类似于下采样取值,将原图像的宽高信息切分,聚合到channel通道中。
源码:
class Focus(nn.Module):
# Focus wh information into c-space
"""理论:从高分辨率图像中,周期性的抽出像素点重构到低分辨率图像中,即将图像相邻的四个位置进行堆叠,
聚焦wh维度信息到c通道中,增大每个点的感受野,减少原始信息的丢失,该模块的设计主要是减少计算量加快速度
Focus wh information into c-space 把宽度w和高度h的信息整合到c空间中
1. 先做4个slice 再concat 最后再做Conv
2. slice后 (b,c1,w,h) -> 分成4个slice 每个slice(b,c1,w/2,h/2)
3. concat(dim=1)后 4个slice(b,c1,w/2,h/2)) -> (b,4c1,w/2,h/2)
4. conv后 (b,4c1,w/2,h/2) -> (b,c2,w/2,h/2)
"""
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True): # ch_in, ch_out, kernel, stride, padding, groups
super().__init__()
self.conv = Conv(c1 * 4, c2, k, s, p, g, act)
# self.contract = Contract(gain=2)
def forward(self, x): # x(b,c,w,h) -> y(b,4c,w/2,h/2)
# 假设x = [1,2,3,4,5,6,7,8,9] x[::2] = [1,3,5,7,9] 间隔2个取样
# x[1::2] = [2, 4, 6, 8] 从第二个数据开始,间隔2个取样
return self.conv(torch.cat([x[..., ::2, ::2], x[..., 1::2, ::2], x[..., ::2, 1::2], x[..., 1::2, 1::2]], 1))
# return self.conv(self.contract(x))
标准的bottleneck模块,用在构建BottleneckCSP和C3等模块中,包含shortcut,起到加深网络的作用。
源码:
class Bottleneck(nn.Module):
# Standard bottleneck
def __init__(self, c1, c2, shortcut=True, g=1, e=0.5): # ch_in, ch_out, shortcut, groups, expansion
super().__init__()
c_ = int(c2 * e) # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
self.cv2 = Conv(c_, c2, 3, 1, g=g)
self.add = shortcut and c1 == c2
def forward(self, x):
return x + self.cv2(self.cv1(x)) if self.add else self.cv2(self.cv1(x))
C3模块其实是简化版的BottleneckCSP,该部分除了Bottleneck之外,只有3个卷积模块,可以减少参数,所以取名C3,作者用意为:C3() is an improved version of CSPBottleneck(). It is simpler, faster and and lighter with similar performance and better fuse characteristics.
源码:
class C3(nn.Module):
# C3() is an improved version of CSPBottleneck()
# It is simpler, faster and lighter with similar performance and better fuse characteristics
# CSP Bottleneck with 3 convolutions
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5): # ch_in, ch_out, number, shortcut, groups, expansion
super().__init__()
c_ = int(c2 * e) # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
self.cv2 = Conv(c1, c_, 1, 1)
self.cv3 = Conv(2 * c_, c2, 1) # act=FReLU(c2)
self.m = nn.Sequential(*(Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)))
# self.m = nn.Sequential(*[CrossConv(c_, c_, 3, 1, g, 1.0, shortcut) for _ in range(n)])
def forward(self, x):
return self.cv3(torch.cat((self.m(self.cv1(x)), self.cv2(x)), dim=1))
Spatial Pyramid Pooling (SPP) layer https://arxiv.org/abs/1406.4729
SPP层将更多不同分辨率的特征进行融合,在送入网络neck之前能够得到更多的信息。
源码:
class SPP(nn.Module):
# Spatial Pyramid Pooling (SPP) layer https://arxiv.org/abs/1406.4729
def __init__(self, c1, c2, k=(5, 9, 13)):
super().__init__()
c_ = c1 // 2 # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
# cv2的输入channel数,等于c_乘以4(4个不同的分辨率的feature map进行融合)
self.cv2 = Conv(c_ * (len(k) + 1), c2, 1, 1)
self.m = nn.ModuleList([nn.MaxPool2d(kernel_size=x, stride=1, padding=x // 2) for x in k])
def forward(self, x):
x = self.cv1(x)
with warnings.catch_warnings():
warnings.simplefilter('ignore') # suppress torch 1.9.0 max_pool2d() warning
return self.cv2(torch.cat([x] + [m(x) for m in self.m], 1))
Spatial Pyramid Pooling - Fast (SPPF) layer for YOLOv5 by Glenn Jocher
SPP-Fast顾名思义就是为了保证准确率相似的条件下爱,减少计算量,以提高速度,使用3个5×5的最大池化,代替原来的5×5、9×9、13×13最大池化。
源码:
class SPPF(nn.Module):
# Spatial Pyramid Pooling - Fast (SPPF) layer for YOLOv5 by Glenn Jocher
def __init__(self, c1, c2, k=5): # equivalent to SPP(k=(5, 9, 13))
super().__init__()
c_ = c1 // 2 # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
self.cv2 = Conv(c_ * 4, c2, 1, 1)
self.m = nn.MaxPool2d(kernel_size=k, stride=1, padding=k // 2)
def forward(self, x):
x = self.cv1(x)
with warnings.catch_warnings():
warnings.simplefilter('ignore') # suppress torch 1.9.0 max_pool2d() warning
y1 = self.m(x)
y2 = self.m(y1)
return self.cv2(torch.cat([x, y1, y2, self.m(y2)], 1))