laxmaniron/3d-animation

2D to 3D motion capture

🌴 Hey there, I'm Laxman 🤝

A Dedicated Machine Learning Engineer proficient in Python and PyTorch, with a proven track record of architecting and optimizing scalable deep learning training frameworks in distributed environments. Experienced in working with large codebases, I am adept at collaborating with interdisciplinary teams to drive cutting-edge ML solutions. My passion lies in harnessing technology to design innovative solutions that transform user experiences.

Always enthusiastic about the open-source community, I am on the lookout to contribute to collaborative machine learning projects. So, if you're keen on harnessing my expertise or envision a collaborative endeavor, please feel free to reach out. Let's create, innovate, and elevate together!

---

Connect with me 🔗	Skills 🎩

3d-animation's Issues

What FPS can we expect?

Is it possible to get this model running on a webcam? What FPS would be possible in this case and does it depend on the number of people in the video?

Uplifting 2D to 3D using a different method than 3d-pose-baseline

Hi,
I do not have access to the Human3.6M dataset and therefore cannot use 3d-pose-baseline.
I found another network to do uplifting named VideoPose3D, but I don't know how the output of 3d-pose-baseline, which is the input to unity, should look like.
Can you please share an example output of 3d-pose-baseline so I can understand how to manipulate the output I receive from VideoPose3D?

Thanks

Help with animation in unity

Hi, Thank you for making your code available. I read the explanation on the medium link and created a new project with your code in Unity. However, I dont know how to create the animation with your code. It would be really helpful if you could please provide more detailed steps for creating the animation in Unity with your code, especially for beginners of Unity.

thanks

Confusion regarding the intermediate outputs in overall process

Hi, I have recently started working into Pose estimation and Unity. The repository seems very interesting and I wanted to create the same things but due to non clarity in the intermediate inputs and output from each model I am stuck in between.

As per the instruction-

1. Run AlphaPose and use the output in DeepSort. Should I use the Alphapose-output.json as input to DeepSort?

2. What specific output do I require from DeepSort ?

Can I possibly get some help or references in understanding the overall steps with outputs.

Thanks

FormatException: Input string was not in a correct format

Thank you for this repository, I have been trying for so long to animate an character with some 3D keypoints of human in Unity unsuccessfully. Your repository is really helping. But I run into an issue when I copied your results in the folder "my_code" in the assets folder. I get the following error: FormatException: Input string was not in a correct format.

I don't know if you still maintain this repo but could you please point me to what needs to be corrected?

JSON from AlphaPose to TXT for SORT ?

Hi there, such a simple question, but I can't find an obvious script to do this. It seems AlphaPose produces a JSON file decsribing the detections, but it's unclear how to convert that to the TXT file that SORT seems to require (e.g. test/MOT16-06/det/det.txt
The ini file is of course easy to create.
Anyone still playing with this and able to point me in the right direction? Thanks :-)