import torch.nn as nn
import math
# Modified from https://github.com/tonylins/pytorch-mobilenet-v2/blob/master/MobileNetV2.py.
# In this version, Relu6 is replaced with Relu to make it ONNX compatible.
# BatchNorm Layer is optional to make it easy do batch norm confusion.
def conv_bn(inp, oup, stride, use_batch_norm=True, onnx_compatible=False):
ReLU = nn.ReLU if onnx_compatible else nn.ReLU6
if use_batch_norm:
return nn.Sequential(nn.Conv2d(inp, oup, 3, stride, 1, bias=False), nn.BatchNorm2d(oup), ReLU(inplace=True))
else:
return nn.Sequential(nn.Conv2d(inp, oup, 3, stride, 1, bias=False), ReLU(inplace=True))
def conv_1x1_bn(inp, oup, use_batch_norm=True, onnx_compatible=False):
ReLU = nn.ReLU if onnx_compatible else nn.ReLU6
if use_batch_norm:
return nn.Sequential(nn.Conv2d(inp, oup, 1, 1, 0, bias=False), nn.BatchNorm2d(oup), ReLU(inplace=True))
else:
return nn.Sequential(nn.Conv2d(inp, oup, 1, 1, 0, bias=False), ReLU(inplace=True))
class InvertedResidual(nn.Module):
def __init__(self, inp, oup, stride, expand_ratio, use_batch_norm=True, onnx_compatible=False):
super(InvertedResidual, self).__init__()
ReLU = nn.ReLU if onnx_compatible else nn.ReLU6
self.stride = stride
assert stride in [1, 2]
hidden_dim = round(inp * expand_ratio)
self.use_res_connect = self.stride == 1 and inp == oup
if expand_ratio == 1:
if use_batch_norm:
self.conv = nn.Sequential(
# dw
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
ReLU(inplace=True),
# pw-linear
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
else:
self.conv = nn.Sequential(
# dw
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
ReLU(inplace=True),
# pw-linear
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
)
else:
if use_batch_norm:
self.conv = nn.Sequential(
# pw
nn.Conv2d(inp, hidden_dim, 1, 1, 0, bias=False),
nn.BatchNorm2d(hidden_dim),
ReLU(inplace=True),
# dw
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
ReLU(inplace=True),
# pw-linear
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
else:
self.conv = nn.Sequential(
# pw
nn.Conv2d(inp, hidden_dim, 1, 1, 0, bias=False),
ReLU(inplace=True),
# dw
nn.Conv2d(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
ReLU(inplace=True),
# pw-linear
nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False),
)
def forward(self, x):
if self.use_res_connect:
return x + self.conv(x)
else:
return self.conv(x)
class MobileNetV2(nn.Module):
def __init__(
self,
n_class=1000,
input_size=224,
width_mult=1.0,
dropout_ratio=0.2,
use_batch_norm=True,
onnx_compatible=False,
):
super(MobileNetV2, self).__init__()
block = InvertedResidual
input_channel = 32
last_channel = 1280
interverted_residual_setting = [
# t, c, n, s
[1, 16, 1, 1],
[6, 24, 2, 2],
[6, 32, 3, 2],
[6, 64, 4, 2],
[6, 96, 3, 1],
[6, 160, 3, 2],
[6, 320, 1, 1],
]
# building first layer
assert input_size % 32 == 0
input_channel = int(input_channel * width_mult)
self.last_channel = int(last_channel * width_mult) if width_mult > 1.0 else last_channel
self.features = [conv_bn(3, input_channel, 2, onnx_compatible=onnx_compatible)]
# building inverted residual blocks
for t, c, n, s in interverted_residual_setting:
output_channel = int(c * width_mult)
for i in range(n):
if i == 0:
self.features.append(
block(
input_channel,
output_channel,
s,
expand_ratio=t,
use_batch_norm=use_batch_norm,
onnx_compatible=onnx_compatible,
)
)
else:
self.features.append(
block(
input_channel,
output_channel,
1,
expand_ratio=t,
use_batch_norm=use_batch_norm,
onnx_compatible=onnx_compatible,
)
)
input_channel = output_channel
# building last several layers
self.features.append(
conv_1x1_bn(
input_channel, self.last_channel, use_batch_norm=use_batch_norm, onnx_compatible=onnx_compatible
)
)
# make it nn.Sequential
self.features = nn.Sequential(*self.features)
# building classifier
self.classifier = nn.Sequential(
nn.Dropout(dropout_ratio),
nn.Linear(self.last_channel, n_class),
)
self._initialize_weights()
def forward(self, x):
x = self.features(x)
x = x.mean(3).mean(2)
x = self.classifier(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(2.0 / 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):
n = m.weight.size(1)
m.weight.data.normal_(0, 0.01)
m.bias.data.zero_()
