Getting error about nms cpp extension

[Junhan Wang]

Thanks for your great job! When i run demo i get error like this:
Traceback (most recent call last):
File "tools/demo.py", line 204, in
a2v = Activity2Vec(args.mode, cfg, logger)
File "tools/demo.py", line 33, in init
self.alphapose = AlphaPose(cfg.DEMO.DETECTOR, cfg.DEMO.YOLO_CFG, cfg.DEMO.YOLO_WEIGHT, cfg.DEMO.POSE_CFG, cfg.DEMO.POSE_WEIGHT, cfg.DEMO.TRACKER_WEIGHT, logger)
File "/home/wjh/Desktop/video/HAKE-Action-Torch/tools/inference_tools/pose_inference.py", line 230, in init
self.det_loader = DetectionLoader(get_detector(self.args.detector, self.args), self.cfg, self.args)
File "/home/wjh/Desktop/video/HAKE-Action-Torch/tools/inference_tools/detector/apis.py", line 12, in get_detector
from .yolo_api import YOLODetector
File "/home/wjh/Desktop/video/HAKE-Action-Torch/tools/inference_tools/detector/yolo_api.py", line 27, in
from .nms import nms_wrapper
File "/home/wjh/Desktop/video/HAKE-Action-Torch/tools/inference_tools/detector/nms/init.py", line 1, in
from .nms_wrapper import nms, soft_nms
File "/home/wjh/Desktop/video/HAKE-Action-Torch/tools/inference_tools/detector/nms/nms_wrapper.py", line 6, in
from . import nms_cpu, nms_cuda
ImportError: libtorch_cpu.so: cannot open shared object file: No such file or directory

This error seems related to nms cpp extension. I can not fix it.

[Hongwei Fan]

Thanks for playing with Activity2Vec! To solve this problem, we need the detailed information about your environment. Could you please provide the output of the following command?

python -c 'from torch.utils.collect_env import main; main()'

[Junhan Wang]

I got these:
Collecting environment information...
PyTorch version: 1.4.0
Is debug build: No
CUDA used to build PyTorch: 10.1

OS: Ubuntu 18.04.5 LTS
GCC version: (Ubuntu 6.5.0-2ubuntu1~18.04) 6.5.0 20181026
CMake version: Could not collect

Python version: 3.7
Is CUDA available: Yes
CUDA runtime version: 10.1.105
GPU models and configuration: GPU 0: GeForce RTX 2060
Nvidia driver version: 460.39
cuDNN version: /usr/local/cuda-10.1/targets/x86_64-linux/lib/libcudnn.so.7

Versions of relevant libraries:
[pip3] numpy==1.20.1
[pip3] torch==1.4.0
[pip3] torchfile==0.1.0
[pip3] torchvision==0.5.0
[conda] torch 1.4.0 pypi_0 pypi
[conda] torchfile 0.1.0 pypi_0 pypi
[conda] torchvision 0.5.0 pypi_0 pypi

[Hongwei Fan]

Thanks! It seems that the nms module of Activity2Vec needs to be rebuilt.
Please remove the "build" folder in the root directory of Activity2Vec and re-run the setup command:

rm -r build/
python setup.py build develop

[Junhan Wang]

Thanks, problem solved by reinstall torch1.4.0, and rebuild twice.