ModuleNotFoundError: No module named '_ext'

[AndrewGuo0930]

After I ran the command python train.py --model_name DSFNet --gpus 0,1 --lr 1.25e-4 --lr_step 30,45 --num_epochs 55 --batch_size 4 --val_intervals 5 --test_large_size True --datasetname rsdata --data_dir ./data/RsCarData/, it said that ModuleNotFoundError: No module named '_ext'. Here is my error log.

Traceback (most recent call last):
  File "train.py", line 13, in <module>
    from lib.models.stNet import get_det_net,load_model, save_model
  File "/cluster/home/it_stu12/main/SatVideoDT/Moving-object-detection-DSFNet-main/lib/models/stNet.py", line 7, in <module>
    from lib.models.DSFNet_with_Static import DSFNet_with_Static
  File "/cluster/home/it_stu12/main/SatVideoDT/Moving-object-detection-DSFNet-main/lib/models/DSFNet_with_Static.py", line 16, in <module>
    from lib.models.DCNv2.dcn_v2 import DCN
  File "/cluster/home/it_stu12/main/SatVideoDT/Moving-object-detection-DSFNet-main/lib/models/DCNv2/dcn_v2.py", line 13, in <module>
    import _ext as _backend
ModuleNotFoundError: No module named '_ext'

Additionally, here is my environment information.

Python 3.7.0 | packaged by conda-forge | (default, Nov 12 2018, 20:15:55) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.version
'3.7.0 | packaged by conda-forge | (default, Nov 12 2018, 20:15:55) \n[GCC 7.3.0]'
>>> import torch
>>> torch.__version__
'1.7.1'
>>> import torchvision
>>> torchvision.__version__
'0.8.2'

[ChaoXiao12]

It seems that you did not compile the DCN. You can refer to the INSTALL of CenterNet to build the conda environment step by step.

[AndrewGuo0930]

Thank you so much. Is the DCNv2 used in this code compiled with PyTorch 1.7.1?

[WesLee88524]

Thank you so much. Is the DCNv2 used in this code compiled with PyTorch 1.7.1?

Maybe u can solve this problem by this way : https://github.com/lbin/DCNv2.
It works for me !