DeepRegNet / DeepReg

Medical image registration using deep learning
Apache License 2.0
552 stars 76 forks source link

Incorrect demo script #840

Open IVarha opened 5 months ago

IVarha commented 5 months ago

Hello, I am trying setup your demos, and test run of them and having several errors on code :

python demos/unpaired_ct_lung/demo_train.py

The training can also be launched using the following command.
deepreg_train --gpu '0' --config_path demos/unpaired_ct_lung/unpaired_ct_lung.yaml --log_dir demos/unpaired_ct_lung --exp_name logs_train
=======================================================

Traceback (most recent call last):
  File "demos/unpaired_ct_lung/demo_train.py", line 46, in <module>
    train(
TypeError: train() got an unexpected keyword argument 'exp_name'

further when I try this command

deepreg_train --gpu '0' --config_path demos/unpaired_ct_lung/unpaired_ct_lung.yaml --log_dir demos/unpaired_ct_lung --exp_name logs_train

deepreg_train: error: unrecognized arguments: --exp_name logs_train

when I removed --exp_name and do same run

Traceback (most recent call last):
  File "/usr/local/bin/deepreg_train", line 8, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.8/dist-packages/deepreg/train.py", line 242, in main
    train(
  File "/usr/local/lib/python3.8/dist-packages/deepreg/train.py", line 89, in train
    config, log_dir, ckpt_path = build_config(
  File "/usr/local/lib/python3.8/dist-packages/deepreg/train.py", line 44, in build_config
    config = config_parser.load_configs(config_path)
  File "/usr/local/lib/python3.8/dist-packages/deepreg/config/parser.py", line 45, in load_configs
    loaded_config = config_sanity_check(config)
  File "/usr/local/lib/python3.8/dist-packages/deepreg/config/parser.py", line 87, in config_sanity_check
    if data_config["format"] not in ["nifti", "h5"]:

the config is default

dataset:
  train:
    dir: demos/unpaired_ct_lung/dataset/train
    format: nifti
    labeled: true
  valid:
    dir: demos/unpaired_ct_lung/dataset/test
    format: nifti
    labeled: true
  test:
    dir: demos/unpaired_ct_lung/dataset/test
    format: nifti
    labeled: true
  type: unpaired # paired / unpaired / grouped
  image_shape: [96, 96, 104]
train:
  method: "ddf" # the registration method, value should be ddf / dvf / conditional
  backbone:
    name: "local" # value should be local / global / unet
    num_channel_initial: 32 # number of initial channel in local net, controls the size of the network
    extract_levels: [0, 1, 2, 3]
  loss:
    image:
      name: "ssd"
      weight: 2.0
    label:
      weight: 1.0
      name: "dice"
      scales: [0, 1, 2, 4, 8, 16]
    regularization:
      weight: 1.0
      name: "bending"
  preprocess:
    data_augmentation:
      name: "affine"
    batch_size: 4
    shuffle_buffer_num_batch: 1
  optimizer:
    name: "Adam"
    learning_rate: 1.0e-5
  epochs: 5000
  save_period: 500

my installed version is latest 0.1.2

Maybe it is worth to check. Thank you