addcb13fb3dcd2cdbc32ac81611a78ac1c19b02f,niftynet/network/dense_vnet.py,DenseVNet,layer_op,#DenseVNet#Any#Any#Any#,70

Before Change

        segmentation=image_resize(segmentation,input_size[1:-1])
        tf.summary.scalar("segmentation".format(idx), tf.reduce_mean(tf.square(segmentation)), [logging.LOG])
        logging.image3_axial("seg",tf.nn.softmax(segmentation)[:,:,:,:,1:]*255.,3,[logging.LOG])
        logging.image3_axial("img",tf.minimum(255.,tf.maximum(0.,(tf.to_float(downsampled_img)/2.+1.)*127.)),3,[logging.LOG])
        return segmentation

def image_resize(image,output_size):

After Change

                tf.summary.scalar("down{}".format(idx),tf.reduce_mean(tf.square(down)),[tf.GraphKeys.SUMMARIES])
            all_segmentation_features.append(image_resize(skip,output_shape))
        hack=True
        if hack == True:
            all_segmentation_features=[all_segmentation_features[0]]+all_segmentation_features
        segmentation = ConvolutionalLayer(
                 self.num_classes+1,
                 kernel_size=hp["final_kernel"],
                 with_bn=False,

Instances