[Bug] 'dict' object has no attribute 'net'

[Pinery-lee]

我在本地运行给的变化检测的例子时，在运行run_task.py进行训练的时候，报错如下：
...
use_vdl: True
2023-04-17 20:20:17 [INFO] 1024 samples in file ./data/levircd/val.txt
2023-04-17 20:20:20 [INFO] 7120 samples in file ./data/levircd/train.txt
Traceback (most recent call last):
File "run_task.py", line 108, in
cfg['optimizer'].args['parameters'] = model.net.parameters()
AttributeError: 'dict' object has no attribute 'net'
这是怎么回事呢？model是个dict吗？

[Bobholamovic]

我这边好像没办法复现这个问题。请问你修改了代码嘛？如果没有修改的话，请提供使用的PaddleRS版本~

[Pinery-lee]

是因为版本问题吗？
(paddle_env) PS D:\Change-detection-using-deep-learning-in-remote-sensing\PyCharmProjects> conda list paddlers
packages in environment at C:\Users\lx\anaconda3\envs\paddle_env:
Name Version Build Channel
paddlers 1.0b0 pypi_0 pypi

[Pinery-lee]

1. 后来，我将paddlers clone 到了本地，使用脚本训练train_cd.py和预测predict_cd.py成功跑通。
2. 但是，如您所见，使用run_task.py时报错AttributeError: 'dict' object has no attribute 'net'。
3. 详细记录如下：

(paddle_env) PS D:\Change-detection-using-deep-learning-in-remote-sensing\PyCharmProjects> python run_task.py train cd --config "D:/Change-detection-using-deep-learning-in-remote-sensing
/PyCharmProjects/data/levircd/configs/levircd/levircd.yaml" 2>&1 | tee "D:/Change-detection-using-deep-learning-in-remote-sensing/PyCharmProjects/logs/try_logs.txt"
python : C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:121: DeprecationWarning: pkg_resources is deprecated as an API
所在位置 行:1 字符: 1
+ python run_task.py train cd --config "D:/Change-detection-using-deep- ...
+ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CategoryInfo          : NotSpecified: (C:\Users\lx\ana...cated as an API:String) [], RemoteException
    + FullyQualifiedErrorId : NativeCommandError
 
  warnings.warn("pkg_resources is deprecated as an API", DeprecationWarning)
C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:2870: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('google')`.
Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#k 
eyword-namespace-packages
  declare_namespace(pkg)
C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:2870: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('google.cloud')`.
Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#k 
eyword-namespace-packages
  declare_namespace(pkg)
C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:2349: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('google')`.
Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#k
eyword-namespace-packages
  declare_namespace(parent)
C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:2870: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('google.logging')`.       
Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#k 
eyword-namespace-packages
  declare_namespace(pkg)
C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:2870: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('mpl_toolkits')`.
Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#k 
eyword-namespace-packages
  declare_namespace(pkg)
C:\Users\lx\anaconda3\envs\paddle_env\lib\site-packages\pkg_resources\__init__.py:2870: DeprecationWarning: Deprecated call to `pkg_resources.declare_namespace('zope')`.
Implementing implicit namespace packages (as specified in PEP 420) is preferred to `pkg_resources.declare_namespace`. See https://setuptools.pypa.io/en/latest/references/keywords.html#k
eyword-namespace-packages
  declare_namespace(pkg)
D:\Change-detection-using-deep-learning-in-remote-sensing\PyCharmProjects\paddlers\transforms\functions.py:57: DeprecationWarning: invalid escape sequence \ 
  """
D:\Change-detection-using-deep-learning-in-remote-sensing\PyCharmProjects\paddlers\transforms\functions.py:169: DeprecationWarning: invalid escape sequence \ 
  """
Can not use `conditional_random_field`. Please install pydensecrf first.
Can not import `map_display`. This is probably because GDAL is not properly installed.
Warning: Unable to use OC-SORT, please install filterpy, for example: `pip install filterpy`, see https://github.com/rlabbe/filterpy
Warning: import ppdet from source directory without installing, run 'python setup.py install' to install ppdet firstly
2023-04-19 10:57:00,483-WARNING: post-quant-hpo is not support in system other than linux
cmd: train
config: D:/Change-detection-using-deep-learning-in-remote-sensing/PyCharmProjects/data/levircd/configs/levircd/levircd.yaml
datasets:
 eval:
  type: CDDataset
  module:
  args:
   binarize_labels: True
   data_dir: ./data/levircd/
   file_list: ./data/levircd/val.txt
   label_list: None
   num_workers: 0
   shuffle: False
   with_seg_labels: False
 train:
  type: CDDataset
  module:
  args:
   binarize_labels: True
   data_dir: ./data/levircd/
   file_list: ./data/levircd/train.txt
   label_list: None
   num_workers: 2
   shuffle: True
   with_seg_labels: False
download_on: False
download_path: ./
download_url:
early_stop: False
early_stop_patience: 5
inherit_off: False
learning_rate: 0.002
log_interval_steps: 50
losses:

model:

num_epochs: 50
optimizer:
 type: Adam
 module:
 args:
  learning_rate:
   type: StepDecay
   module: paddle.optimizer.lr
   args:
    gamma: 0.2
    learning_rate: 0.002
    step_size: 35000
resume_checkpoint:
save_dir: ./exp/levircd/
save_interval_epochs: 5
seed: None
task: cd
train:

train_batch_size: 8
transforms:
 eval:
  -
   type: DecodeImg
   module:
   args:

  -
   type: Normalize
   module:
   args:
    mean:
     - 0.5
     - 0.5
     - 0.5
    std:
     - 0.5
     - 0.5
     - 0.5
 train:
  -
   type: DecodeImg
   module:
   args:

  -
   type: RandomFlipOrRotate
   module:
   args:
    probs:
     - 0.35
     - 0.35
    probsf:
     - 0.5
     - 0.5
     - 0
     - 0
     - 0
    probsr:
     - 0.33
     - 0.34
     - 0.33
  -
   type: Normalize
   module:
   args:
    mean:
     - 0.5
     - 0.5
     - 0.5
    std:
     - 0.5
     - 0.5
     - 0.5
use_vdl: True
2023-04-19 10:57:00 [INFO]      1024 samples in file ./data/levircd/val.txt
2023-04-19 10:57:02 [INFO]      7120 samples in file ./data/levircd/train.txt
Traceback (most recent call last):
  File "run_task.py", line 108, in <module>
    cfg['optimizer'].args['parameters'] = model.net.parameters()
AttributeError: 'dict' object has no attribute 'net'

[Bobholamovic]

看起来你用的是1.0-beta版本，examples/rs_research仅支持1.0正式版以上的版本，请根据文档安装~