niftynet.contrib.csv_reader.highres3dnet_features module

class HighRes3DNetFeatures(w_initializer=None, w_regularizer=None, b_initializer=None, b_regularizer=None, acti_func='prelu', name='HighRes3DNet')[source]


implementation of HighRes3DNet:
Li et al., “On the compactness, efficiency, and representation of 3D convolutional networks: Brain parcellation as a pretext task”, IPMI ‘17
layer_op(images, is_training, layer_id=-1)[source]
class HighResBlock(n_output_chns, kernels=(3, 3), acti_func='relu', w_initializer=None, w_regularizer=None, with_res=True, name='HighResBlock')[source]

Bases: niftynet.layer.base_layer.TrainableLayer

This class define a high-resolution block with residual connections kernels

  • specify kernel sizes of each convolutional layer
  • e.g.: kernels=(5, 5, 5) indicate three conv layers of kernel_size 5


  • whether to add residual connections to bypass the conv layers
layer_op(input_tensor, is_training)[source]