5b5ea7b5d864e53894f5342f19becfba8f8035e5,niftynet/network/interventional_dense_net.py,INetDense,layer_op,#INetDense#Any#Any#Any#Any#,49

Before Change

        up_res_3 = UpRes(self.fea[0], **self.res_param)(
            up_res_2, conv_0_0, is_training)

        conv_5 = Conv(n_output_chns=spatial_rank,
                      kernel_size=self.k_conv,
                      with_bias=True,
                      with_bn=False,
                      acti_func=None,
                      **self.disp_param)(up_res_3)

        if base_grid is None:
            in_spatial_size = [None] * spatial_rank
            out_spatial_size = conv_5.get_shape().as_list()[1:-1]
            base_grid = _create_affine_features(output_shape=out_spatial_size,
                                                source_shape=in_spatial_size)
            base_grid = np.asarray(base_grid[:-1])
            base_grid = np.reshape(
                base_grid.T, [-1] + out_spatial_size + [spatial_rank])
            base_grid = tf.constant(base_grid, dtype=conv_5.dtype)
        dense_field = base_grid + conv_5
        // TODO filtering
        if self.smoothing_func is not None:
            dense_field = self.smoothing_func(dense_field, spatial_rank)
        return dense_field

After Change

        if self.multi_scale_fusion:
            output_list = [up_res_3, up_res_2, up_res_1, up_res_0, conv_4]
        else:
            output_list = [up_res_3]

        // converting all output layers to displacement fields 
        dense_fields = []
        for scale_out in output_list:

Instances