contextual_model

Model

• contextual_model.py

SENet_Segmentation - ver.PyTorch

Prerequisites

• [Python]
• [PyTorch]
• Visdom

Model

• 2d_unetse_model

model = torch.nn.DataParallel(
    Dilated_UNet(inChannel=2, num_classes=2, init_features=32, network_depth=3, bottleneck_layers=3))
model.cuda()

• 2d_densenetse_model

model=torch.nn.DataParallel(FCDenseNet103(2))
model.cuda()

• 3d_unetse_model

model = torch.nn.DataParallel(
    UNet3D_SE(inChannel=2, num_classes=2, init_features=32, network_depth=3, bottleneck_layers=2))
model.cuda()

Visualizatoin

• you should run visdom before training

python -m visdom.server

click the URL http://localhost:8097