cannot import name 'deform_conv_cuda' from 'mmdet.ops.dcn'

[EddieEduardo]

Hi， thanks for sharing your code.

These days I tried to run it, but I met with some errors as follows:

Traceback (most recent call last):
File "htc_deepsort.py", line 13, in
from mmdet.apis import init_detector, inference_detector, show_result
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\apis_init_.py", line 1, in
from .inference import (async_inference_detector, inference_detector,
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\apis\inference.py", line 11, in
from mmdet.core import get_classes
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\core_init_.py", line 7, in
from .post_processing import * # noqa: F401, F403
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\core\post_processing_init_.py", line 1, in
from .bbox_nms import multiclass_nms
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\core\post_processing\bbox_nms.py", line 3, in
from mmdet.ops.nms import nms_wrapper
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\ops_init_.py", line 2, in
from .conv import build_conv_layer
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\ops\conv.py", line 4, in
from .dcn import DeformConvPack, ModulatedDeformConvPack
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\ops\dcn_init_.py", line 1, in
from .deform_conv import (DeformConv, DeformConvPack, ModulatedDeformConv,
File "J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\ops\dcn\deform_conv.py", line 11, in
from . import deform_conv_cuda
ImportError: cannot import name 'deform_conv_cuda' from 'mmdet.ops.dcn' (J:\MOTS\MOT CODES\pedestrian_tracking-master\mmdet\ops\dcn_init_.py)

Could you please tell me how to fix it ? Thanks very much.

[FinalFlowers]

It seems that mmdetection was not correctly compiled.
To use deformable convolution, you need to compile the project on Linux by
pip install -v -e . # or "python setup.py develop"
You can get more information about the compilation process through install.md of mmdetection

Also, CUDA should be configured correctly. You can check to see if CUDA is configured correctly in the following code
import torch
print torch.cuda.is_available()