from torch import nn import torch.utils.model_zoo as model_zoo 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 class ConvBNReLU(nn.Sequential): def __init__(self, in_planes, out_planes, kernel_size=3, stride=1, groups=1): padding = (kernel_size - 1) // 2 super(ConvBNReLU, self).__init__( nn.Conv2d(in_planes, out_planes, kernel_size, stride, padding, groups=groups, bias=False), nn.BatchNorm2d(out_planes), nn.ReLU6(inplace=True) ) class InvertedResidual(nn.Module): def __init__(self, inp, oup, stride, expand_ratio): super(InvertedResidual, self).__init__() self.stride = stride assert stride in [1, 2] hidden_dim = int(round(inp * expand_ratio)) self.use_res_connect = self.stride == 1 and inp == oup layers = [] if expand_ratio != 1: # pw layers.append(ConvBNReLU(inp, hidden_dim, kernel_size=1)) layers.extend([ # dw ConvBNReLU(hidden_dim, hidden_dim, stride=stride, groups=hidden_dim), # pw-linear nn.Conv2d(hidden_dim, oup, 1, 1, 0, bias=False), nn.BatchNorm2d(oup), ]) self.conv = nn.Sequential(*layers) 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, pretrained=True): """ MobileNet V2 main class Args: num_classes (int): Number of classes width_mult (float): Width multiplier - adjusts number of channels in each layer by this amount inverted_residual_setting: Network structure round_nearest (int): Round the number of channels in each layer to be a multiple of this number Set to 1 to turn off rounding block: Module specifying inverted residual building block for mobilenet """ super(MobileNetV2, self).__init__() block = InvertedResidual input_channel = 32 last_channel = 1280 width_mult = 1.0 round_nearest=8 inverted_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], ] # only check the first element, assuming user knows t,c,n,s are required if len(inverted_residual_setting) == 0 or len(inverted_residual_setting[0]) != 4: raise ValueError("inverted_residual_setting should be non-empty " "or a 4-element list, got {}".format(inverted_residual_setting)) # building first layer input_channel = _make_divisible(input_channel * width_mult, round_nearest) self.last_channel = _make_divisible(last_channel * max(1.0, width_mult), round_nearest) features = [ConvBNReLU(3, input_channel, stride=2)] # building inverted residual blocks for t, c, n, s in inverted_residual_setting: output_channel = _make_divisible(c * width_mult, round_nearest) for i in range(n): stride = s if i == 0 else 1 features.append(block(input_channel, output_channel, stride, expand_ratio=t)) input_channel = output_channel # building last several layers features.append(ConvBNReLU(input_channel, self.last_channel, kernel_size=1)) # make it nn.Sequential self.features = nn.Sequential(*features) # building classifier # self.classifier = nn.Sequential( # nn.Dropout(0.2), # nn.Linear(self.last_channel, num_classes), # ) self.toplayer = nn.Conv2d(160, 32, kernel_size=1, stride=1, padding=0) self.latlayer1 = nn.Conv2d(64, 32, kernel_size=1, stride=1, padding=0) self.latlayer2 = nn.Conv2d( 32, 32, kernel_size=1, stride=1, padding=0) self.latlayer3 = nn.Conv2d( 24, 32, kernel_size=1, stride=1, padding=0) self.smooth1 = nn.Conv2d(32, 32, kernel_size=3, stride=1, padding=1) self.smooth2 = nn.Conv2d(32, 32, kernel_size=3, stride=1, padding=1) self.smooth3 = nn.Conv2d(32, 32, kernel_size=3, stride=1, padding=1) self.fpn_selected = [2, 5, 9, 15] # weight initialization for m in self.modules(): if isinstance(m, nn.Conv2d): nn.init.kaiming_normal_(m.weight, mode='fan_out') if m.bias is not None: nn.init.zeros_(m.bias) elif isinstance(m, nn.BatchNorm2d): nn.init.ones_(m.weight) nn.init.zeros_(m.bias) elif isinstance(m, nn.Linear): nn.init.normal_(m.weight, 0, 0.01) nn.init.zeros_(m.bias) if pretrained: self._load_pretrained_model() def _forward_impl(self, x): # This exists since TorchScript doesn't support inheritance, so the superclass method # (this one) needs to have a name other than `forward` that can be accessed in a subclass fpn_features = [] for i, f in enumerate(self.features): x = f(x) if i in self.fpn_selected: fpn_features.append(x) c2, c3, c4, c5 = fpn_features # Top-down p5 = self.toplayer(c5) p4 = nn.functional.upsample(p5, size=c4.size()[2:], mode='bilinear', align_corners=True) + self.latlayer1(c4) p3 = nn.functional.upsample(p4, size=c3.size()[2:], mode='bilinear', align_corners=True) + self.latlayer2(c3) p2 = nn.functional.upsample(p3, size=c2.size()[2:], mode='bilinear', align_corners=True) + self.latlayer3(c2) p4 = self.smooth1(p4) p3 = self.smooth2(p3) p2 = self.smooth3(p2) return p2, p3, p4, p5 # x = self.features(x) # Cannot use "squeeze" as batch-size can be 1 => must use reshape with x.shape[0] # x = nn.functional.adaptive_avg_pool2d(x, 1).reshape(x.shape[0], -1) # x = self.classifier(x) # return x def forward(self, x): return self._forward_impl(x) def _load_pretrained_model(self): pretrain_dict = model_zoo.load_url('https://download.pytorch.org/models/mobilenet_v2-b0353104.pth') model_dict = {} state_dict = self.state_dict() for k, v in pretrain_dict.items(): if k in state_dict: model_dict[k] = v state_dict.update(model_dict) self.load_state_dict(state_dict) def MobileNet_FPN(output_stride=None, BatchNorm=nn.BatchNorm2d, pretrained=True): """Constructs a ResNet-101 model. Args: pretrained (bool): If True, returns a model pre-trained on ImageNet """ model = MobileNetV2(pretrained=pretrained) return model