mirror of
https://github.com/netfun2000/dddd_trainer.git
synced 2026-08-18 01:09:41 +08:00
sml2h3
232 lines
6.2 KiB
Python
232 lines
6.2 KiB
Python
"""
|
|
Creates a EfficientNetV2 Model as defined in:
|
|
Mingxing Tan, Quoc V. Le. (2021).
|
|
EfficientNetV2: Smaller Models and Faster Training
|
|
arXiv preprint arXiv:2104.00298.
|
|
import from https://github.com/d-li14/mobilenetv2.pytorch
|
|
"""
|
|
|
|
import torch
|
|
import torch.nn as nn
|
|
import math
|
|
|
|
__all__ = ['effnetv2_s', 'effnetv2_m', 'effnetv2_l', 'effnetv2_xl']
|
|
|
|
|
|
def _make_divisible(v, divisor, min_value=None):
|
|
"""
|
|
This function is taken from the original tf repo.
|
|
It ensures that all layers have a channel number that is divisible by 8
|
|
It can be seen here:
|
|
https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet/mobilenet.py
|
|
:param v:
|
|
:param divisor:
|
|
:param min_value:
|
|
:return:
|
|
"""
|
|
if min_value is None:
|
|
min_value = divisor
|
|
new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
|
|
# Make sure that round down does not go down by more than 10%.
|
|
if new_v < 0.9 * v:
|
|
new_v += divisor
|
|
return new_v
|
|
|
|
|
|
# SiLU (Swish) activation function
|
|
if hasattr(nn, 'SiLU'):
|
|
SiLU = nn.SiLU
|
|
else:
|
|
# For compatibility with old PyTorch versions
|
|
class SiLU(nn.Module):
|
|
def forward(self, x):
|
|
return x * torch.sigmoid(x)
|
|
|
|
|
|
class SELayer(nn.Module):
|
|
def __init__(self, inp, oup, reduction=4):
|
|
super(SELayer, self).__init__()
|
|
self.avg_pool = nn.AdaptiveAvgPool2d(1)
|
|
self.fc = nn.Sequential(
|
|
nn.Linear(oup, _make_divisible(inp // reduction, 8)),
|
|
SiLU(),
|
|
nn.Linear(_make_divisible(inp // reduction, 8), oup),
|
|
nn.Sigmoid()
|
|
)
|
|
|
|
def forward(self, x):
|
|
b, c, _, _ = x.size()
|
|
y = self.avg_pool(x).view(b, c)
|
|
y = self.fc(y).view(b, c, 1, 1)
|
|
return x * y
|
|
|
|
|
|
def conv_3x3_bn(inp, oup, stride):
|
|
return nn.Sequential(
|
|
nn.Conv2d(inp, oup, 3, stride, 1, bias=False),
|
|
nn.BatchNorm2d(oup),
|
|
SiLU()
|
|
)
|
|
|
|
|
|
def conv_1x1_bn(inp, oup):
|
|
return nn.Sequential(
|
|
nn.Conv2d(inp, oup, 1, 1, 0, bias=False),
|
|
nn.BatchNorm2d(oup),
|
|
SiLU()
|
|
)
|
|
|
|
|
|
class MBConv(nn.Module):
|
|
def __init__(self, inp, oup, stride, expand_ratio, use_se):
|
|
super(MBConv, self).__init__()
|
|
assert stride in [1, 2]
|
|
|
|
hidden_dim = round(inp * expand_ratio)
|
|
self.identity = stride == 1 and inp == oup
|
|
if use_se:
|
|
self.conv = nn.Sequential(
|
|
# pw
|
|
nn.Conv2d(inp, hidden_dim, 1, 1, 0, bias=False),
|
|
nn.BatchNorm2d(hidden_dim),
|
|
SiLU(),
|
|
# dw
|
|
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
|
|
nn.BatchNorm2d(hidden_dim),
|
|
SiLU(),
|
|
SELayer(inp, hidden_dim),
|
|
# pw-linear
|
|
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
|
|
nn.BatchNorm2d(oup),
|
|
)
|
|
else:
|
|
self.conv = nn.Sequential(
|
|
# fused
|
|
nn.Conv2d(inp, hidden_dim, 3, stride, 1, bias=False),
|
|
nn.BatchNorm2d(hidden_dim),
|
|
SiLU(),
|
|
# pw-linear
|
|
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
|
|
nn.BatchNorm2d(oup),
|
|
)
|
|
|
|
def forward(self, x):
|
|
if self.identity:
|
|
return x + self.conv(x)
|
|
else:
|
|
return self.conv(x)
|
|
|
|
|
|
class EffNetV2(nn.Module):
|
|
def __init__(self, cfgs, nc=3, width_mult=1.):
|
|
super(EffNetV2, self).__init__()
|
|
self.cfgs = cfgs
|
|
|
|
# building first layer
|
|
input_channel = _make_divisible(24 * width_mult, 8)
|
|
layers = [conv_3x3_bn(nc, input_channel, 2)]
|
|
# building inverted residual blocks
|
|
block = MBConv
|
|
for t, c, n, s, use_se in self.cfgs:
|
|
output_channel = _make_divisible(c * width_mult, 8)
|
|
for i in range(n):
|
|
layers.append(block(input_channel, output_channel, s if i == 0 else 1, t, use_se))
|
|
input_channel = output_channel
|
|
self.features = nn.Sequential(*layers)
|
|
|
|
self._initialize_weights()
|
|
|
|
def forward(self, x):
|
|
x = self.features(x)
|
|
return x
|
|
|
|
def _initialize_weights(self):
|
|
for m in self.modules():
|
|
if isinstance(m, nn.Conv2d):
|
|
n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
|
|
m.weight.data.normal_(0, math.sqrt(2. / n))
|
|
if m.bias is not None:
|
|
m.bias.data.zero_()
|
|
elif isinstance(m, nn.BatchNorm2d):
|
|
m.weight.data.fill_(1)
|
|
m.bias.data.zero_()
|
|
elif isinstance(m, nn.Linear):
|
|
m.weight.data.normal_(0, 0.001)
|
|
m.bias.data.zero_()
|
|
|
|
|
|
def effnetv2_s(**kwargs):
|
|
"""
|
|
Constructs a EfficientNetV2-S model
|
|
"""
|
|
cfgs = [
|
|
# t, c, n, s, SE
|
|
[1, 24, 2, 1, 0],
|
|
[4, 48, 4, 2, 0],
|
|
[4, 64, 4, 2, 0],
|
|
[4, 128, 6, 2, 1],
|
|
[6, 160, 9, 1, 1],
|
|
[6, 256, 15, 2, 1],
|
|
]
|
|
return EffNetV2(cfgs, **kwargs)
|
|
|
|
|
|
def effnetv2_m(**kwargs):
|
|
"""
|
|
Constructs a EfficientNetV2-M model
|
|
"""
|
|
cfgs = [
|
|
# t, c, n, s, SE
|
|
[1, 24, 3, 1, 0],
|
|
[4, 48, 5, 2, 0],
|
|
[4, 80, 5, 2, 0],
|
|
[4, 160, 7, 2, 1],
|
|
[6, 176, 14, 1, 1],
|
|
[6, 304, 18, 2, 1],
|
|
[6, 512, 5, 1, 1],
|
|
]
|
|
return EffNetV2(cfgs, **kwargs)
|
|
|
|
|
|
def effnetv2_l(**kwargs):
|
|
"""
|
|
Constructs a EfficientNetV2-L model
|
|
"""
|
|
cfgs = [
|
|
# t, c, n, s, SE
|
|
[1, 32, 4, 1, 0],
|
|
[4, 64, 7, 2, 0],
|
|
[4, 96, 7, 2, 0],
|
|
[4, 192, 10, 2, 1],
|
|
[6, 224, 19, 1, 1],
|
|
[6, 384, 25, 2, 1],
|
|
[6, 640, 7, 1, 1],
|
|
]
|
|
return EffNetV2(cfgs, **kwargs)
|
|
|
|
|
|
def effnetv2_xl(**kwargs):
|
|
"""
|
|
Constructs a EfficientNetV2-XL model
|
|
"""
|
|
cfgs = [
|
|
# t, c, n, s, SE
|
|
[1, 32, 4, 1, 0],
|
|
[4, 64, 8, 2, 0],
|
|
[4, 96, 8, 2, 0],
|
|
[4, 192, 16, 2, 1],
|
|
[6, 256, 24, 1, 1],
|
|
[6, 512, 32, 2, 1],
|
|
[6, 640, 8, 1, 1],
|
|
]
|
|
return EffNetV2(cfgs, **kwargs)
|
|
|
|
def test():
|
|
net = effnetv2_s(nc=1)
|
|
x = torch.randn(1, 1, 128, 128)
|
|
y = net(x)
|
|
print(y.size())
|
|
|
|
|
|
if __name__ == '__main__':
|
|
test() |