In the config file, the number of output class is set to 3. But only probability map and centripetal shift map are needed. Do I miss anything here?
model = dict( type='CT', backbone=dict( type='resnet18', pretrained=True ), neck=dict( type='FPEM_v1', in_channels=(64, 128, 256, 512), out_channels=128 ), detection_head=dict( type='CT_Head', in_channels=512, hidden_dim=128, num_classes=3, loss_kernel=dict( type='DiceLoss', loss_weight=1.0 ), loss_loc=dict( type='SmoothL1Loss', beta=0.1, loss_weight=0.05 ) ) )
In the config file, the number of output class is set to 3. But only probability map and centripetal shift map are needed. Do I miss anything here?
model = dict( type='CT', backbone=dict( type='resnet18', pretrained=True ), neck=dict( type='FPEM_v1', in_channels=(64, 128, 256, 512), out_channels=128 ), detection_head=dict( type='CT_Head', in_channels=512, hidden_dim=128, num_classes=3, loss_kernel=dict( type='DiceLoss', loss_weight=1.0 ), loss_loc=dict( type='SmoothL1Loss', beta=0.1, loss_weight=0.05 ) ) )