mehtadushy / selecsls-pytorch Goto Github PK

Hi, when I run evaluate_timing.py but appeared bugs as following
AssertionError: The NVIDIA driver on your system is too old (found version 9010).
Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install
a PyTorch version that has been compiled with your version of the CUDA driver.

My cuda version of pytorch is 10.1 and I want to know what't the version of your cuda?

Hi，thanks for your great work. Will you release the code and weights of human pose estimation? I'm very interested in training and evaluate code of stage I II III

Hello, will pytorch code of the XNect project be released soon?

Hello, will caffe code of the XNect project be released soon?

Now, MuPoTS-3D and MPI-INF-3DHP dataset only provide the download code of matlab, I would like to ask if these two data sets will also provide the download code of python or c++? In addition， you will other cloud disk download links be provided (for example, Google Drive)？

Hello, In Section"5.3 Kinematic Skeleton Fitting","D = 29 is the number of degrees of freedom (DOF) for one skeleton", how D is calculated?

When I execute XNect C++ Library and take our video as input, an error is reported,

terminate called after throwing an instance of'cv::Exception'
what(): OpenCV(3.4.11) /action_recognition_project/XNect/opencv-3.4.11/modules/calib3d/src/solvepnp.cpp:754: error: (-215:Assertion failed) ((npoints >= 4) | | (npoints == 3 && flags == SOLVEPNP_ITERATIVE && useExtrinsicGuess)) && npoints == std::max(ipoints.checkVector(2, CV_32F), ipoints.checkVector(2, CV_64F)) in function'solvePnPGeneric'.

Because the postCNNSecondNet function is encapsulated in libxnect.a file, I can’t see the implementation. How can I solve this error?

Would it be possible to get a bit more detailed description on how to use the implementation. Specifically, I don't understand exactly which dataset to specify in the parameter (--imagenet_base_path <path_to_imagenet_dataset>).
I am a teacher and in the sense of a research transfer into the application practice I would be gladly ready to write a kind of user manual. But for that I would have to understand it myself first ;-)

Hi Dr. Mehta/ XNECT Team,

First of all, great work with XNECT! I am a PhD candidate at the University of Maryland College Park. I was trying to use XNECT for a paper I am working on. I had a weird issue when there was an overlap of two people in a given frame. In this setup, I have footage of two people overlap and person 1's number of joints detected are less than 4 -- I think this issue happens then due to pnp constraint. Try/catch does not capture this issue as OpenCV catches it and terminates the call. Can you or someone on your team please help resolve this issue?

Steps to reproduce:
Place the attached frame in the images folder path ../../data/images (defined by images_to_load variable)
run XNECT/src/main.cpp

Possible error after function call:

(line193 at XNECT/extern/xnect/include/xnect.hpp)(inside function XNECT::processImg ) :: postCNNSecondNet(m_SecondNet->blob_by_name("pred_3d_pose_scale_root")->cpu_data(), m_ColorSized);  

Command Line output:

terminate called after throwing an instance of 'cv::Exception'
  what():  OpenCV(3.4.9) /root/opencv-3.4.9/modules/calib3d/src/solvepnp.cpp:754: error: (-215:Assertion failed) ( (npoints >= 4) || (npoints == 3 && flags == SOLVEPNP_ITERATIVE && useExtrinsicGuess) ) && npoints == std::max(ipoints.checkVector(2, CV_32F), ipoints.checkVector(2, CV_64F)) in function 'solvePnPGeneric'

Here is a sample frame:

Thanks for your help!

Best,
Snehesh

Hi,
Firstly, thanks for your splendid work and beautiful coding!
I'm a student mainly focusing on real-time 3D pose estimation. Your 3D demo is the best real-time 3D result I've ever seen. So, after the ImageNet demo, will you release 3D pose estimation code and model?

Do you have demo code for camera input like the video shown on the XNect site?

Thanks,
Dan

Hi guys, thanks for the great work.

Just wish to mention that I tried running the model based on the Usage guide and got _pickle.UnpicklingError: invalid load key, '-'. . I found the cause is in the wget command where the flag -o is used (this writes a log file to ./weights/ folder but actually saves the weight outside it). The good one should be -O (with capital O).

Can't download Newer Pretrained Models files. It is always show 404 error.