Yolov10s网络模型及代码分析

原创

已于 2024-06-26 09:41:58 修改 · 955 阅读

13 ·

CC 4.0 BY-SA版权

文章标签：

#YOLO

于 2024-06-24 23:15:32 首次发布

一、网络模型

"""
backbone:
  # [from, repeats, module, args]
  - [-1, 1, Conv, [32, 3, 2]]           # 0-P1/2   320*320*32
  - [-1, 1, Conv, [64, 3, 2]]           # 1-P2/4   160*160*64
  - [-1, 1, C2f, [64, True]]            # 2
  - [-1, 1, Conv, [128, 3, 2]]          # 3-P3/8   80*80*128
  - [-1, 2, C2f, [128, True]]           # 4
  - [-1, 1, SCDown, [256, 3, 2]]        # 5-P4/16  40*40*256
  - [-1, 2, C2f, [256, True]]           # 6
  - [-1, 1, SCDown, [512, 3, 2]]        # 7-P5/32  20*20*512
  - [-1, 1, C2fCIB, [512, True, True]]  # 8
  - [-1, 1, SPPF, [512, 5]]             # 9
  - [-1, 1, PSA, [512]]                 # 10       20*20*512

head:
  - [-1, 1, nn.Upsample, [None, 2, "nearest"]]  # 11                 40*40*512
  - [[-1, 6], 1, Concat, [1]]                   # cat backbone P4    40*40*768
  - [-1, 1, C2f, [256]]                         # 13                 40*40*256

  - [-1, 1, nn.Upsample, [None, 2, "nearest"]]  # 14                 80*80*256
  - [[-1, 4], 1, Concat, [1]]                   # cat backbone P3    80*80*384
  - [-1, 1, C2f, [128]]                         # 16 (P3/8-small)    80*80*128

  - [-1, 1, Conv, [128, 3, 2]]                  # 17                 40*40*128
  - [[-1, 13], 1, Concat, [1]]                  # cat head P4        40*40*384
  - [-1, 1, C2f, [256]]                         # 19 (P4/16-medium)  40*40*256

  - [-1, 1, SCDown, [256, 3, 2]]                # 20                 20*20*256
  - [[-1, 10], 1, Concat, [1]]                  # cat head P5        20*20*768
  - [-1, 1, C2fCIB, [512, True, True]]          # 22 (P5/32-large)   20*20*512

  - [[16, 19, 22], 1, v10Detect, [nc]]          # Detect(P3, P4, P5) 80*80*85 40*40*85 20*20*85
"""

二、特殊结构分解

1、C2f

160*160*64 --> 160*160*64

class C2f(nn.Module):
    """Faster Implementation of CSP Bottleneck with 2 convolutions."""

    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):
        """Initialize CSP bottleneck layer with two convolutions with arguments ch_in, ch_out, number, shortcut, groups,
        expansion.
        """
        super().__init__()
        self.c = int(c2 * e)  # hidden channels
        self.cv1 = Conv(c1, 2 * self.c, 1, 1)
        self.cv2 = Conv((2 + n) * self.c, c2, 1)  # optional act=FReLU(c2)
        self.m = nn.ModuleList(Bottleneck(self.c, self.c, shortcut, g, k=((3, 3), (3, 3)), e=1.0) for _ in range(n))

    def forward(self, x):
        """Forward pass through C2f layer."""
        y = list(self.cv1(x).chunk(2, 1))
        y.extend(m(y[-1]) for m in self.m)
        return self.cv2(torch.cat(y, 1))