final_test.py

#!/usr/bin/env python3
from mpunet.models import UNet3D, UNet
from tensorflow import keras
from heartnet.layers import RobustScaler
from heartnet.models.base import BaseModelTraining
for i in range(3):
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=112, out_activation="softmax"), name=f"pad{i}"
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate()
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, out_activation="softmax"), name=f"shrink{i}"
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate()
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, out_activation="softmax"),
    #     name=f"aug-96-0.333-100-{i}"
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate()
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, out_activation="softmax"),
    #     name=f"augmentation-{i}"
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate()
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, out_activation="softmax", depth=4),
    #     name=f"hyper-20-30-1-4-{i}"
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate()
    # base = BaseModelTraining(
    #     UNet(2, depth=4, dim=128, out_activation="softmax", complexity_factor=2),
    #     f"base{i}"
    # )
    # base.batch_size = 64
    # base.setup(True)
    # base.evaluate()
    # base = BaseModelTraining(keras.Sequential([
    #     keras.Input(shape=(128, 128, 1)),
    #     RobustScaler(),
    #     UNet(2,
    #          depth=4,
    #          dim=128,
    #          out_activation="softmax",
    #          complexity_factor=2)
    # ]),
    #                          f"base-data-robust-{i}")
    # base.batch_size = 64
    # base.setup(True)
    # base.evaluate(True)
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=112, out_activation="softmax"), name=f"pad-data-{i}", full=True
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate(True)
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, out_activation="softmax"), name=f"shrink-data-{i}", full=True
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate(True)
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, out_activation="softmax"), name=f"hyper-1-data-{i}", full=True
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate(True)
    # base3D = BaseModelTraining(
    #     UNet3D(2, dim=96, depth=4, out_activation="softmax"), name=f"hyper-2-data-{i}", full=True
    # )
    # base3D.batch_size = 1
    # base3D.setup(True)
    # base3D.evaluate(True)
    base3D = BaseModelTraining(keras.Sequential(
        [keras.Input(shape=(112, 112, 112, 1)),
            RobustScaler(),
        UNet3D(2, dim=112, out_activation="softmax")]),
                            name=f"pad-robust-scaled-{i}")
    base3D.setup(True)
    base3D.evaluate(True)