def __init__(self, class_num=62):
super(res18, self).__init__()
model_ft = ResNet(BasicBlock, [2, 2, 2, 2])
self.base_model = nn.Sequential(*list(model_ft.children())[:-3])
# attention schema
self.avgpool = nn.AdaptiveAvgPool2d(1)
self.maxpool = nn.AdaptiveMaxPool2d(1)
self.sign = nn.Sigmoid()
in_plances = 256
ratio = 8
self.a_fc1 = nn.Conv2d(in_plances,in_plances//ratio,1,bias=False)
self.a_relu = nn.ReLU()
self.a_fc2 = nn.Conv2d(in_plances//ratio, in_plances, 1, bias=False)
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.max_pool = nn.AdaptiveMaxPool2d(1)
self.reduce_layer = nn.Conv2d(512, 256, 1)
# self.classifier = ClassBlock(512, 1024)
self.fc1 = nn.Sequential(nn.Dropout(0.5),
nn.Linear(256, class_num))
self.fc2 = nn.Sequential(nn.Dropout(0.5),
nn.Linear(256, class_num))
self.fc3 = nn.Sequential(nn.Dropout(0.5),
nn.Linear(256, class_num))
self.fc4 = nn.Sequential(nn.Dropout(0.5),
nn.Linear(256, class_num))
return out
def se_resnet50(num_classes=1_000, pretrained=True):
"""Constructs a ResNet-50 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(SEBottleneck, [3, 4, 6, 3], num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
if pretrained:
model.load_state_dict(
load_state_dict_from_url(
"https://github.com/moskomule/senet.pytorch/releases/download/archive/seresnet50-60a8950a85b2b.pkl"
))
for child in model.children():
for param in child.parameters():
param.requires_grad = False
return model
class vae_fc(nn.Module):
def __init__(self):
super(vae_fc, self).__init__()
self.fc1 = nn.Linear(1000, 512)
self.fc21 = nn.Linear(512, 256)
self.fc22 = nn.Linear(512, 256)
def encode(self, x):
h1 = F.relu(self.fc1(x))
