About the use of MixConv2d

[Zengyf-CVer]

@glenn-jocher
How should I use this MixConv2d method? I did an experiment and replaced Conv directly with the MixConv2d method, and found that it couldn't run:

backbone:
  # [from, number, module, args]
  [[-1, 1, Conv, [64, 6, 2, 2]],  # 0-P1/2
   [-1, 1, MixConv2d, [128, 3, 2]],  # 1-P2/4
   [-1, 3, C3, [128]],
   [-1, 1, MixConv2d, [256, 3, 2]],  # 3-P3/8
   [-1, 6, C3, [256]],
   [-1, 1, MixConv2d, [512, 3, 2]],  # 5-P4/16
   [-1, 9, C3, [512]],
   [-1, 1, MixConv2d, [1024, 3, 2]],  # 7-P5/32
   [-1, 3, C3, [1024]],
   [-1, 1, SPPF, [1024, 5]],  # 9
  ]

Console output information:

I found that the problem with this k, how do I set this k?

yolov5/models/experimental.py

Line 53 in ed887b5

groups = len(k)

[glenn-jocher]

@Zengyf-CVer see MixConv2d module for input requirements:

yolov5/models/experimental.py

Lines 49 to 71 in ed887b5

	class MixConv2d(nn.Module):
	# Mixed Depth-wise Conv https://arxiv.org/abs/1907.09595
	def __init__(self, c1, c2, k=(1, 3), s=1, equal_ch=True):
	super().__init__()
	groups = len(k)
	if equal_ch: # equal c_ per group
	i = torch.linspace(0, groups - 1E-6, c2).floor() # c2 indices
	c_ = [(i == g).sum() for g in range(groups)] # intermediate channels
	else: # equal weight.numel() per group
	b = [c2] + [0] * groups
	a = np.eye(groups + 1, groups, k=-1)
	a -= np.roll(a, 1, axis=1)
	a *= np.array(k) ** 2
	a[0] = 1
	c_ = np.linalg.lstsq(a, b, rcond=None)[0].round() # solve for equal weight indices, ax = b
	self.m = nn.ModuleList([nn.Conv2d(c1, int(c_[g]), k[g], s, k[g] // 2, bias=False) for g in range(groups)])
	self.bn = nn.BatchNorm2d(c2)
	self.act = nn.LeakyReLU(0.1, inplace=True)
	def forward(self, x):
	return x + self.act(self.bn(torch.cat([m(x) for m in self.m], 1)))

[glenn-jocher]

@Zengyf-CVer Conv() module for comparison:

yolov5/models/common.py

Lines 36 to 46 in ed887b5

	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)
	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)))

[Zengyf-CVer]

@glenn-jocher
I used a closed test and found a problem. Is this a bug?

import torch
from utils.torch_utils import profile
from models.experimental import MixConv2d

m = MixConv2d(128, 256, (3, 5), 1)
results = profile(input=torch.randn(16, 128, 80, 80), ops=[m], n=1)

Error message:

The size of tensor a (128) must match the size of tensor b (256) at non-singleton dimension 1

Is there a bug for MixConv2d?

[glenn-jocher]

@Zengyf-CVer I cam able to reproduce this error (thank you for the code to reproduce!), may be a bug, will investigate.