AssertionError: assert 'policy' in lr_config

[Shradha2020]

Hello,

Thank you for such a detailed MMSegmentation tutorial.
While trying to run the MMSegmentation tutorial for my custom dataset.
I have provided the dataset as required and have changed the num_classes in the cfg according to my dataset.
While running the training , i get the following error

2021-04-17 22:43:10,314 - mmseg - WARNING - The model and loaded state dict do not match exactly

unexpected key in source state_dict: fc.weight, fc.bias

---

AssertionError Traceback (most recent call last)
in ()
17 mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
18 train_segmentor(model, datasets, cfg, distributed=False, validate=True,
---> 19 meta=dict())

2 frames
/usr/local/lib/python3.7/dist-packages/mmcv/runner/base_runner.py in register_lr_hook(self, lr_config)
372 return
373 elif isinstance(lr_config, dict):
--> 374 assert 'policy' in lr_config
375 policy_type = lr_config.pop('policy')
376 # If the type of policy is all in lower case, e.g., 'cyclic',

Can you please help.

[xvjiarui]

Hi @Shradha2020
May I have your config file? And could you please format your error message? You may consider using error_report template

[Shradha2020]

Hi,

@xvjiarui

Thank you so much for reaching out to me .
I have used the error_report template as below.
I have attached my config file
config.txt

Describe the bug
Facing AssertionError: assert 'policy' in lr_config while trying to run the training on my own custom dataset following the MMSegmentation tutorial.

Reproduction

1. What command or script did you run?

   train_segmentor(model, datasets, cfg, distributed=False, validate=True,
   meta=dict())

2. Did you make any modifications on the code or config? Did you understand what you have modified?
   I just modified num classes of the model in decode/auxiliary head as 3 because my custom dataset has only three classes

3. What dataset did you use?
   I have used a custom dataset of only 10 images, where in the images are manually annotated as fire , nofire and burnt.
   Is there any minimum requirement for the number of images required for training ?

Environment

1. Please run python mmseg/utils/collect_env.py to collect necessary environment infomation and paste it here.
   Using colab pro

sys.platform: linux
Python: 3.7.10 (default, Feb 20 2021, 21:17:23) [GCC 7.5.0]
CUDA available: True
GPU 0: Tesla P100-PCIE-16GB
CUDA_HOME: /usr/local/cuda
NVCC: Build cuda_11.0_bu.TC445_37.28845127_0
GCC: gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
PyTorch: 1.5.0+cu101
PyTorch compiling details: PyTorch built with:

• GCC 7.3
• C++ Version: 201402
• Intel(R) Math Kernel Library Version 2019.0.5 Product Build 20190808 for Intel(R) 64 architecture applications
• Intel(R) MKL-DNN v0.21.1 (Git Hash 7d2fd500bc78936d1d648ca713b901012f470dbc)
• OpenMP 201511 (a.k.a. OpenMP 4.5)
• NNPACK is enabled
• CPU capability usage: AVX2
• CUDA Runtime 10.1
• NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_37,code=compute_37
• CuDNN 7.6.3
• Magma 2.5.2
• Build settings: BLAS=MKL, BUILD_TYPE=Release, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -fopenmp -DNDEBUG -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DUSE_INTERNAL_THREADPOOL_IMPL -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_STATIC_DISPATCH=OFF,

TorchVision: 0.6.0+cu101
OpenCV: 4.1.2
MMCV: 1.3.1
MMCV Compiler: GCC 7.3
MMCV CUDA Compiler: 10.1
MMSegmentation: 0.12.0+67eee62

2. You may add addition that may be helpful for locating the problem, such as
   • I have installed PyTorch using pip in Colab just as the tutorial

Error traceback

If applicable, paste the error trackback here.

2021-04-17 23:00:01,943 - mmseg - INFO - Use load_from_openmmlab loader
2021-04-17 23:00:02,017 - mmseg - WARNING - The model and loaded state dict do not match exactly

unexpected key in source state_dict: fc.weight, fc.bias

---

AssertionError Traceback (most recent call last)
in ()
17 mmcv.mkdir_or_exist(osp.abspath(cfg.work_dir))
18 train_segmentor(model, datasets, cfg, distributed=False, validate=True,
---> 19 meta=dict())

2 frames
/usr/local/lib/python3.7/dist-packages/mmcv/runner/base_runner.py in register_lr_hook(self, lr_config)
372 return
373 elif isinstance(lr_config, dict):
--> 374 assert 'policy' in lr_config
375 policy_type = lr_config.pop('policy')
376 # If the type of policy is all in lower case, e.g., 'cyclic',

AssertionError:

[svtotemxai]

Hi @Shradha2020 ,

If you haven't found the solution yet, you may try following steps, they worked for me.

1. Run config file loading cell again
   something like:
   from mmcv import Config
   cfg = Config.fromfile('configs/pspnet/pspnet_r50-d8_512x1024_40k_cityscapes.py

2. Run the modified config cell and look for lr_config in printed output.
   It should look like: lr_config = dict(policy='poly', power=0.9, min_lr=0.0001, by_epoch=False)

Now you should be able to train and evaluate without assertion error.

[MengzhangLI]

Hi, @Shradha2020:

Have you fixed this error?

Best,