varunnr/camera_and_pose

Reconstructing 3D Human Pose from 2D Image Landmarks

camera_and_pose's Issues

Use own labeled data

Since im new to this field, im struggling to test my own labeled data with your tool. I was wondering if you could provide a workflow, on how to use your tool or better, on how to integrate own labeled images, if this is even possible? I understand, that the examples have stored the x/y coordinates of the anatomical landmarks, but i dont know, in which order shall these landmarks be stored in the mat. double, also i dont understand, what the im file stands for, which is also provided in the example file.
In any case, i dont want to steal your time, but would be glad for an answer.

How to use this code ?

Hi
I'm very interesting this code and your research. Therefore, i want to reference the code but I don't know how add the outside input to the code. If i want to use another input, how can pre-processing the data ?
Thank you very much!

xy data coordinates

Hi, I am wondering in what joint order is your 2D xy data array stored? I.e. which index of xy in your demo corresponds to which joint in the labeled image?

I am attempting to run your program using my own labeled data. I was unable to find the solution in the paper or comments.

Thanks

Which key points in CMU Mocap dataset did you use?

Hi,
I am very interested in your work and I want to do some research based on it, especially those bases. However, I don't know which key points in CMU Mocap Data that you used for your work. Therefore, could you please tell me something about it? Thanks a lot!