question about “mmdet”

[YeTaoY]

I have successfully installed mmdet by:

pip install -r requirements/build.txt pip install "git+https://github.com/open-mmlab/cocoapi.git#subdirectory=pycocotools" pip install -v -e .

However, I got next error when I verify that mmskeleton and mmdetection by "python mmskl.py pose_demo --gpus 1" :

_(open-mmlab) ty@ubuntu:~/KEYPOINT_GCN/mmskeleton$ python mmskl.py pose_demo --gpus 1 /home/ty/anaconda3/lib/python3.7/site-packages/Cython/Distutils/old_build_ext.py:41: UserWarning: Cython.Distutils.old_build_ext does not properly handle dependencies and is deprecated. "Cython.Distutils.old_build_ext does not properly handle dependencies " No CUDA runtime is found, using CUDA_HOME='/usr/local/cuda' Load configuration information from ./configs/pose_estimation/pose_demo.yaml

Pose estimation: Traceback (most recent call last): File "mmskl.py", line 123, in main() File "mmskl.py", line 117, in main call_obj(**cfg.processor_cfg) File "/home/ty/KEYPOINT_GCN/mmskeleton/mmskeleton/utils/importer.py", line 24, in call_obj return import_obj(type)(**kwargs) File "/home/ty/KEYPOINT_GCN/mmskeleton/mmskeleton/processor/pose_demo.py", line 78, in inference model = init_pose_estimator(detection_cfg, estimation_cfg, device=0) File "/home/ty/KEYPOINT_GCN/mmskeleton/mmskeleton/apis/estimation.py", line 18, in init_pose_estimator detection_model = mmdet.apis.init_detector(detection_model_file, File "/home/ty/anaconda3/lib/python3.7/site-packages/lazy_import-0.2.2-py3.7.egg/lazy_import/init.py", line 156, in getattribute _load_module(self) File "/home/ty/anaconda3/lib/python3.7/site-packages/lazy_import-0.2.2-py3.7.egg/lazy_import/init.py", line 537, in _load_module msg.format(**modclass.lazy_import_error_strings)), None) File "", line 3, in raise_from ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.

more information:

1. My mmdetection folde is under mmskeleton folder
2. I notice that the new mmdetection uses PyTorch 1.3+, while mmskeleton uses PyTorch 1.2. Should I turn back to pytorch 1.2

[happysheep224]

same error

[happysheep224]

I modify the mmskeleton/utils/third_party.py, add "import mmdet "in the first row,but it show me another error:

TypeError: Passed module is not a LazyModule instance.

[happysheep224]

My cuda verison is 10.2, is it matter?

[CamilleMaurice]

I am having the same trouble

[happysheep224]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

[mytk2012]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc Should mmdet version be 1.0 or others?

[mytk2012]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

The latest version of mmskeleton is 0.7?

[happysheep224]

@mytk2012 zhe latest version of mmskeleton is 0.7rc1. the version of mmdet I installed is 1.2, because mmdet1.2 support torch1.2.0

[BozhouZha]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

[BozhouZha]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

Solved, nvm.

For friends who may run into the same issue "ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.", please check your mmcv version via $ pip list.

Previously, mine is 1.0.0. I uninstall it via $ pip uninstall mmcv, then $ pip install mmcv==0.4.3 fixed this issue.

[YLTsai0609]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

Solved, nvm.

For friends who may run into the same issue "ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.", please check your mmcv version via $ pip list.

Previously, mine is 1.0.0. I uninstall it via $ pip uninstall mmcv, then $ pip install mmcv==0.4.3 fixed this issue.

Successfully installed with following:

mmcv                      0.4.3                    pypi_0    pypi
mmdet                     1.0rc1+unknown           pypi_0    pypi
mmskeleton                0.7rc1+82bf68a            dev_0    <develop>
pytorch                   1.2.0           cuda100py37h938c94c_0  
torchvision               0.4.0           cuda100py37hecfc37a_0
cudatoolkit               10.0.130                      0  

With Cuda 10.1 on the system path.

echo $PATH

/usr/local/cuda/bin:my_path/anaconda3/envs/py_37_torch_12_open-mmlab/bin:my_path/anaconda3/condabin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin

echo $LD_LIBRARY_PATH /usr/local/cuda/lib64 echo $CUDAHOME /usr/local/cuda

[happysheep224]

thank you for sharing

[zjr666]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

Solved, nvm.

For friends who may run into the same issue "ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.", please check your mmcv version via $ pip list.

Previously, mine is 1.0.0. I uninstall it via $ pip uninstall mmcv, then $ pip install mmcv==0.4.3 fixed this issue.

my mmcv version is 0.4.3, But i still get this error ，How can I solve this problem?

[xiaowo1996]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

Solved, nvm. For friends who may run into the same issue "ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.", please check your mmcv version via $ pip list. Previously, mine is 1.0.0. I uninstall it via $ pip uninstall mmcv, then $ pip install mmcv==0.4.3 fixed this issue.

my mmcv version is 0.4.3, But i still get this error ，How can I solve this problem?

same problem , Did you slove the probelm?

[xiaowo1996]

met the same problem, althougth install mmcv0.4.3,but didn't work ,how can i slove the issue

[xiaowo1996]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

Solved, nvm. For friends who may run into the same issue "ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.", please check your mmcv version via $ pip list. Previously, mine is 1.0.0. I uninstall it via $ pip uninstall mmcv, then $ pip install mmcv==0.4.3 fixed this issue.

Successfully installed with following:

mmcv                      0.4.3                    pypi_0    pypi
mmdet                     1.0rc1+unknown           pypi_0    pypi
mmskeleton                0.7rc1+82bf68a            dev_0    <develop>
pytorch                   1.2.0           cuda100py37h938c94c_0  
torchvision               0.4.0           cuda100py37hecfc37a_0
cudatoolkit               10.0.130                      0  

With Cuda 10.1 on the system path.

echo $PATH

/usr/local/cuda/bin:my_path/anaconda3/envs/py_37_torch_12_open-mmlab/bin:my_path/anaconda3/condabin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin

echo $LD_LIBRARY_PATH /usr/local/cuda/lib64 echo $CUDAHOME /usr/local/cuda

same envirment, but still report the error

[SKBL5694]

I have fixed this problem by rollback the verision of mmskeleton to 0.7rc

Hi there, may I know how you rollback to the 0.7rc version?

Solved, nvm. For friends who may run into the same issue "ImportError: mmskeleton.utils.third_party attempted to use a functionality that requires module mmdet.apis, but it couldn't be loaded. Please install mmdet and retry.", please check your mmcv version via $ pip list. Previously, mine is 1.0.0. I uninstall it via $ pip uninstall mmcv, then $ pip install mmcv==0.4.3 fixed this issue.

Successfully installed with following:

mmcv                      0.4.3                    pypi_0    pypi
mmdet                     1.0rc1+unknown           pypi_0    pypi
mmskeleton                0.7rc1+82bf68a            dev_0    <develop>
pytorch                   1.2.0           cuda100py37h938c94c_0  
torchvision               0.4.0           cuda100py37hecfc37a_0
cudatoolkit               10.0.130                      0  

With Cuda 10.1 on the system path. echo $PATH

/usr/local/cuda/bin:my_path/anaconda3/envs/py_37_torch_12_open-mmlab/bin:my_path/anaconda3/condabin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin

echo $LD_LIBRARY_PATH /usr/local/cuda/lib64 echo $CUDAHOME /usr/local/cuda

same envirment, but still report the error

when you reinstall those environment, you should execute pip install -e .. So you should run this command 3 times during mmcv, mmdet and mmskeleton

[YAERxc]

same problem!how to install mmdet==1.0