lxy5513 / cvtoolkit Goto Github PK

Traceback (most recent call last):
  File "demo.py", line 408, in <module>
    light_track_camera(pose_estimator, video_capture)
  File "demo.py", line 111, in light_track_camera
    human_candidates = get_human_bbox(cur_img, yolo_model)
  File "demo.py", line 57, in get_human_bbox
    bboxs, probs = inference(image, model)
  File "C:\Pytorch\cvToolkit\pose_track\lighttrack\human_detector.py", line 137, in inference
    im_dim_list = torch.index_select(im_dim_list, 0, output[:,0].long())
TypeError: 'NoneType' object is not subscriptable

When trying to run it on a video

Hi,
I have converted HRNET model for pose estimations from pytorch to coreml successfully. But the output of the converted model is not as in pytorch model.
I mean: the output of the model is the body joints keypoints, the issue in converted coreml is that the predicted keypoints are not exactly at the same position on human on the camera. (All keypoints are there but a little bit moved when you visualize them).

Any suggestion?

Hi @lxy5513 , i really appreciate the effort you put in to make cvToolBox I was interested to compile Hrnet but i get always get some error . I had tried to use your standalone Hrnet implementation but failed to run it. Instead i had to try other implementation smpleHrNEt. And i guess i have to compile Hrnet to run the the pose track ? For me , i dont know what is the issue and why i cant compile it straight forward. So i was wondering if there is something i am missing totally ?

Error shown now

cd nms; python setup_linux.py build_ext --inplace; rm -rf build; cd ../../
Traceback (most recent call last):
  File "setup_linux.py", line 12, in <module>
    from Cython.Distutils import build_ext
ImportError: No module named Cython.Distutils

It would be great otherwise to use the work you have done . I hope you understand .

Make version

make --version
GNU Make 4.1
Built for x86_64-pc-linux-gnu
Copyright (C) 1988-2014 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.

And my installed packages

.      Successfully uninstalled tqdm-4.32.2
Successfully installed Cython-0.29.11 Markdown-2.6.11 Pillow-5.1.0 PyYAML-5.1.1 Werkzeug-0.14.1 absl-py-0.2.2 astor-0.6.2 backcall-0.1.0 bleach-1.5.0 certifi-2018.4.16 easydict-1.9 gast-0.2.0 grpcio-1.12.0 html5lib-0.9999999 imageio-2.3.0 ipdb-0.12 ipython-7.6.1 jedi-0.14.0 kiwisolver-1.0.1 matplotlib-3.1.1 numpy-1.14.3 opencv-python-4.1.0.25 parso-0.5.0 pexpect-4.7.0 prompt-toolkit-2.0.9 protobuf-3.5.2 python-dateutil-2.7.3 pytz-2018.4 setproctitle-1.1.10 tensorboard-1.8.0 tornado-5.0.2 tqdm-4.32.1

In Ubuntu 18.04

朋友你这个代码里面有些逻辑有问题，什么时候更新一下版本啊

hi~ I notice that your folder kps2d_detection has the open_pose subfolder, while you don't use it in your file demo.py. Is it ok to replace pose-estimator with open_pose?

Hi
I was wondering if we can add the pose track module to the simple HrNEt. In your implementation , you use CUDA to compile and exeture simple hrnet. I was wondering if we could use this simpler module simplehrnetcpu. Then it gives the versatility to the whole toolkit. If i understand correctly you can run posetrack it as a standalone module right ?

请问如何对gcn进行训练？

Hi, thanks for your work
when i tested the hrnet and higher hrnet for inference. the drawn avatar was in different location than the camera.
when i test on video or image, things are good and in same location, it only differ for inference and this case only with hrnet (not for other models).
any suggestion on this?