import torch.nn as nn
import torch
import os
import math
from lib.utils.quantize_utils import QConv2d, QLinear
__all__ = ['MobileNetV2', 'mobilenetv2', 'qmobilenetv2']
def conv_bn(inp, oup, stride, conv_layer=nn.Conv2d, half_wave=True):
if conv_layer == nn.Conv2d:
return nn.Sequential(
conv_layer(inp, oup, 3, stride, 1, bias=False),
nn.BatchNorm2d(oup),
nn.ReLU6(inplace=True)
)
else:
return nn.Sequential(
conv_layer(inp, oup, 3, stride, 1, bias=False, half_wave=half_wave),
nn.BatchNorm2d(oup),
nn.ReLU6(inplace=True)
)
def conv_1x1_bn(inp, oup, conv_layer=nn.Conv2d, half_wave=True):
if conv_layer == nn.Conv2d:
return nn.Sequential(
conv_layer(inp, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
nn.ReLU6(inplace=True)
)
else:
return nn.Sequential(
conv_layer(inp, oup, 1, 1, 0, bias=False, half_wave=half_wave),
nn.BatchNorm2d(oup),
nn.ReLU6(inplace=True)
)
def make_divisible(x, divisible_by=8):
import numpy as np
return int(np.ceil(x * 1. / divisible_by) * divisible_by)
class InvertedResidual(nn.Module):
def __init__(self, inp, oup, stride, expand_ratio, conv_layer=nn.Conv2d):
super(InvertedResidual, self).__init__()
self.stride = stride
assert stride in [1, 2]
hidden_dim = int(inp * expand_ratio)
self.use_res_connect = self.stride == 1 and inp == oup
if expand_ratio == 1:
self.conv = nn.Sequential(
# dw
conv_layer(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU6(inplace=True),
# pw-linear
conv_layer(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
elif conv_layer == nn.Conv2d:
self.conv = nn.Sequential(
# pw
conv_layer(inp, hidden_dim, 1, 1, 0, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU6(inplace=True),
# dw
conv_layer(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU6(inplace=True),
# pw-linear
conv_layer(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
else:
self.conv = nn.Sequential(
# pw
conv_layer(inp, hidden_dim, 1, 1, 0, bias=False, half_wave=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU6(inplace=True),
# dw
conv_layer(hidden_dim, hidden_dim, 3, stride, 1, groups=hidden_dim, bias=False),
nn.BatchNorm2d(hidden_dim),
nn.ReLU6(inplace=True),
# pw-linear
conv_layer(hidden_dim, oup, 1, 1, 0, bias=False),
nn.BatchNorm2d(oup),
)
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, num_classes=1000, input_size=224, width_mult=1., block=InvertedResidual, conv_layer=nn.Conv2d):
super(MobileNetV2, self).__init__()
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 = make_divisible(input_channel * width_mult) # first channel is always 32!
self.last_channel = make_divisible(last_channel * width_mult) if width_mult > 1.0 else last_channel
self.features = [conv_bn(3, input_channel, 2, conv_layer=conv_layer)]
# building inverted residual blocks
for t, c, n, s in interverted_residual_setting:
if conv_layer is not nn.Conv2d:
output_channel = make_divisible(c * width_mult)
else:
output_channel = make_divisible(c * width_mult) if t > 1 else c
for i in range(n):
if i == 0:
self.features.append(block(input_channel, output_channel, s, expand_ratio=t, conv_layer=conv_layer))
else:
self.features.append(block(input_channel, output_channel, 1, expand_ratio=t, conv_layer=conv_layer))
input_channel = output_channel
# building last several layers
self.features.append(conv_1x1_bn(input_channel, self.last_channel, conv_layer=conv_layer, half_wave=False))
# make it nn.Sequential
self.features = nn.Sequential(*self.features)
# building classifier
if conv_layer == nn.Conv2d:
self.classifier = nn.Linear(self.last_channel, num_classes)
else:
self.classifier = QLinear(self.last_channel, num_classes)
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. / 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_()
def mobilenetv2(pretrained=False, **kwargs):
model = MobileNetV2(**kwargs)
if pretrained:
# Load pretrained model.
path = 'pretrained/imagenet/mobilenetv2-150.pth.tar'
print('==> load pretrained mobilenetv2 model..')
assert os.path.isfile(path), 'Error: no checkpoint directory found!'
ch = torch.load(path)
ch = {n.replace('module.', ''): v for n, v in ch['state_dict'].items()}
model.load_state_dict(ch, strict=False)
return model
def qmobilenetv2(pretrained=False, num_classes=1000, **kwargs):
# model = MobileNetV2(conv_layer=QConv2d, **kwargs)
model = MobileNetV2(conv_layer=QConv2d, num_classes=1000, **kwargs)
if pretrained:
# Load pretrained model.
path = 'pretrained/imagenet/mobilenetv2-150.pth.tar'
print('==> load pretrained mobilenetv2 model..')
assert os.path.isfile(path), 'Error: no checkpoint directory found!'
ch = torch.load(path)
ch = {n.replace('module.', ''): v for n, v in ch['state_dict'].items()}
model.load_state_dict(ch, strict=False)
return model
if __name__ == '__main__':
# from ops.profile import profile
net = mobilenetv2()
flops, param = profile(net, input_size=(1, 3, 224, 224))
print(flops/1e9, param/1e6)
