Upload to Main

deepcore/nets/mobilenetv3.py

@@ -0,0 +1,304 @@
+import torch.nn as nn
+from torch import set_grad_enabled, flatten, Tensor
+from torchvision.models import mobilenetv3
+from .nets_utils import EmbeddingRecorder
+import math
+
+'''MobileNetV3 in PyTorch.
+Paper： "Inverted Residuals and Linear Bottlenecks:Mobile Networks for Classification, Detection and Segmentation" 
+
+Acknowlegement to:
+https://github.com/d-li14/mobilenetv3.pytorch/blob/master/mobilenetv3.py
+'''
+
+
+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
+    """
+    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
+
+
+class h_sigmoid(nn.Module):
+    def __init__(self, inplace=True):
+        super(h_sigmoid, self).__init__()
+        self.relu = nn.ReLU6(inplace=inplace)
+
+    def forward(self, x):
+        return self.relu(x + 3) / 6
+
+
+class h_swish(nn.Module):
+    def __init__(self, inplace=True):
+        super(h_swish, self).__init__()
+        self.sigmoid = h_sigmoid(inplace=inplace)
+
+    def forward(self, x):
+        return x * self.sigmoid(x)
+
+
+class SELayer(nn.Module):
+    def __init__(self, channel, reduction=4):
+        super(SELayer, self).__init__()
+        self.avg_pool = nn.AdaptiveAvgPool2d(1)
+        self.fc = nn.Sequential(
+            nn.Linear(channel, _make_divisible(channel // reduction, 8)),
+            nn.ReLU(inplace=True),
+            nn.Linear(_make_divisible(channel // reduction, 8), channel),
+            h_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, padding=1):
+    return nn.Sequential(
+        nn.Conv2d(inp, oup, 3, stride, padding, bias=False),
+        nn.BatchNorm2d(oup),
+        h_swish()
+    )
+
+
+def conv_1x1_bn(inp, oup):
+    return nn.Sequential(
+        nn.Conv2d(inp, oup, 1, 1, 0, bias=False),
+        nn.BatchNorm2d(oup),
+        h_swish()
+    )
+
+
+class InvertedResidual(nn.Module):
+    def __init__(self, inp, hidden_dim, oup, kernel_size, stride, use_se, use_hs):
+        super(InvertedResidual, self).__init__()
+        assert stride in [1, 2]
+
+        self.identity = stride == 1 and inp == oup
+
+        if inp == hidden_dim:
+            self.conv = nn.Sequential(
+                # dw
+                nn.Conv2d(hidden_dim, hidden_dim, kernel_size, stride, (kernel_size - 1) // 2, groups=hidden_dim,
+                          bias=False),
+                nn.BatchNorm2d(hidden_dim),
+                h_swish() if use_hs else nn.ReLU(inplace=True),
+                # Squeeze-and-Excite
+                SELayer(hidden_dim) if use_se else nn.Identity(),
+                # 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),
+                nn.BatchNorm2d(hidden_dim),
+                h_swish() if use_hs else nn.ReLU(inplace=True),
+                # dw
+                nn.Conv2d(hidden_dim, hidden_dim, kernel_size, stride, (kernel_size - 1) // 2, groups=hidden_dim,
+                          bias=False),
+                nn.BatchNorm2d(hidden_dim),
+                # Squeeze-and-Excite
+                SELayer(hidden_dim) if use_se else nn.Identity(),
+                h_swish() if use_hs else nn.ReLU(inplace=True),
+                # 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 MobileNetV3_32x32(nn.Module):
+    def __init__(self, cfgs, mode, channel=3, num_classes=1000, record_embedding=False,
+                 no_grad=False, width_mult=1.):
+        super(MobileNetV3_32x32, self).__init__()
+        # setting of inverted residual blocks
+        self.cfgs = cfgs
+        assert mode in ['mobilenet_v3_large', 'mobilenet_v3_small']
+
+        self.embedding_recorder = EmbeddingRecorder(record_embedding)
+        self.no_grad = no_grad
+
+        # building first layer
+        input_channel = _make_divisible(16 * width_mult, 8)
+        layers = [conv_3x3_bn(channel, input_channel, 2, padding=3 if channel == 1 else 1)]
+        # building inverted residual blocks
+        block = InvertedResidual
+        for k, t, c, use_se, use_hs, s in self.cfgs:
+            output_channel = _make_divisible(c * width_mult, 8)
+            exp_size = _make_divisible(input_channel * t, 8)
+            layers.append(block(input_channel, exp_size, output_channel, k, s, use_se, use_hs))
+            input_channel = output_channel
+        self.features = nn.Sequential(*layers)
+        # building last several layers
+        self.conv = conv_1x1_bn(input_channel, exp_size)
+        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
+        output_channel = {'mobilenet_v3_large': 1280, 'mobilenet_v3_small': 1024}
+        output_channel = _make_divisible(output_channel[mode] * width_mult, 8) if width_mult > 1.0 else output_channel[
+            mode]
+        self.classifier = nn.Sequential(
+            nn.Linear(exp_size, output_channel),
+            h_swish(),
+            nn.Dropout(0.2),
+            self.embedding_recorder,
+            nn.Linear(output_channel, num_classes),
+        )
+
+        self._initialize_weights()
+
+    def forward(self, x):
+        with set_grad_enabled(not self.no_grad):
+            x = self.features(x)
+            x = self.conv(x)
+            x = self.avgpool(x)
+            x = x.view(x.size(0), -1)
+            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):
+                m.weight.data.normal_(0, 0.01)
+                m.bias.data.zero_()
+
+    def get_last_layer(self):
+        return self.classifier[-1]
+
+
+class MobileNetV3_224x224(mobilenetv3.MobileNetV3):
+    def __init__(self, inverted_residual_setting, last_channel,
+                 channel=3, num_classes=1000, record_embedding=False, no_grad=False, **kwargs):
+        super(MobileNetV3_224x224, self).__init__(inverted_residual_setting, last_channel,
+                                                  num_classes=num_classes, **kwargs)
+
+        self.embedding_recorder = EmbeddingRecorder(record_embedding)
+
+        self.fc = self.classifier[-1]
+        self.classifier[-1] = self.embedding_recorder
+        self.classifier.add_module("fc", self.fc)
+
+        self.no_grad = no_grad
+
+    def get_last_layer(self):
+        return self.fc
+
+    def _forward_impl(self, x: Tensor) -> Tensor:
+        with set_grad_enabled(not self.no_grad):
+            x = self.features(x)
+            x = self.avgpool(x)
+            x = flatten(x, 1)
+            x = self.classifier(x)
+            return x
+
+
+def MobileNetV3(arch: str, channel: int, num_classes: int, im_size, record_embedding: bool = False,
+                no_grad: bool = False,
+                pretrained: bool = False, **kwargs):
+    arch = arch.lower()
+    if pretrained:
+        if channel != 3:
+            raise NotImplementedError("Network Architecture for current dataset has not been implemented.")
+
+        inverted_residual_setting, last_channel = mobilenetv3._mobilenet_v3_conf(arch)
+        net = MobileNetV3_224x224(inverted_residual_setting=inverted_residual_setting, last_channel=last_channel,
+                                  channel=3, num_classes=1000, record_embedding=record_embedding, no_grad=no_grad,
+                                  **kwargs)
+
+        from torch.hub import load_state_dict_from_url
+        state_dict = load_state_dict_from_url(mobilenetv3.model_urls[arch], progress=True)
+        net.load_state_dict(state_dict)
+
+        if num_classes != 1000:
+            net.fc = nn.Linear(last_channel, num_classes)
+            net.classifier[-1] = net.fc
+
+    elif im_size[0] == 224 and im_size[1] == 224:
+        if channel != 3:
+            raise NotImplementedError("Network Architecture for current dataset has not been implemented.")
+        inverted_residual_setting, last_channel = mobilenetv3._mobilenet_v3_conf(arch)
+        net = MobileNetV3_224x224(inverted_residual_setting=inverted_residual_setting, last_channel=last_channel,
+                                  channel=channel, num_classes=num_classes, record_embedding=record_embedding,
+                                  no_grad=no_grad, **kwargs)
+
+    elif (channel == 1 and im_size[0] == 28 and im_size[1] == 28) or (
+            channel == 3 and im_size[0] == 32 and im_size[1] == 32):
+        if arch == "mobilenet_v3_large":
+            cfgs = [
+                # k, t, c, SE, HS, s
+                [3, 1, 16, 0, 0, 1],
+                [3, 4, 24, 0, 0, 2],
+                [3, 3, 24, 0, 0, 1],
+                [5, 3, 40, 1, 0, 2],
+                [5, 3, 40, 1, 0, 1],
+                [5, 3, 40, 1, 0, 1],
+                [3, 6, 80, 0, 1, 2],
+                [3, 2.5, 80, 0, 1, 1],
+                [3, 2.3, 80, 0, 1, 1],
+                [3, 2.3, 80, 0, 1, 1],
+                [3, 6, 112, 1, 1, 1],
+                [3, 6, 112, 1, 1, 1],
+                [5, 6, 160, 1, 1, 2],
+                [5, 6, 160, 1, 1, 1],
+                [5, 6, 160, 1, 1, 1]
+            ]
+            net = MobileNetV3_32x32(cfgs, arch, channel=channel, num_classes=num_classes,
+                                    record_embedding=record_embedding, no_grad=no_grad)
+        elif arch == "mobilenet_v3_small":
+            cfgs = [
+                # k, t, c, SE, HS, s
+                [3, 1, 16, 1, 0, 2],
+                [3, 4.5, 24, 0, 0, 2],
+                [3, 3.67, 24, 0, 0, 1],
+                [5, 4, 40, 1, 1, 2],
+                [5, 6, 40, 1, 1, 1],
+                [5, 6, 40, 1, 1, 1],
+                [5, 3, 48, 1, 1, 1],
+                [5, 3, 48, 1, 1, 1],
+                [5, 6, 96, 1, 1, 2],
+                [5, 6, 96, 1, 1, 1],
+                [5, 6, 96, 1, 1, 1],
+            ]
+            net = MobileNetV3_32x32(cfgs, arch, channel=channel, num_classes=num_classes,
+                                    record_embedding=record_embedding, no_grad=no_grad)
+        else:
+            raise ValueError("Model architecture not found.")
+    else:
+        raise NotImplementedError("Network Architecture for current dataset has not been implemented.")
+    return net
+
+
+def MobileNetV3Large(channel: int, num_classes: int, im_size, record_embedding: bool = False, no_grad: bool = False,
+                     pretrained: bool = False, **kwargs):
+    return MobileNetV3("mobilenet_v3_large", channel, num_classes, im_size, record_embedding, no_grad,
+                       pretrained, **kwargs)
+
+
+def MobileNetV3Small(channel: int, num_classes: int, im_size, record_embedding: bool = False, no_grad: bool = False,
+                     pretrained: bool = False, **kwargs):
+    return MobileNetV3("mobilenet_v3_small", channel, num_classes, im_size, record_embedding, no_grad,
+                       pretrained, **kwargs)