[Feature Request] Using GPU, CUDA

Hello i have a question about tracking skeleton.
I want using 5 cameras to detect skeleton, but my CPU is too weak. I though if it possible to move calculation from CPU to GPU ?

Hi @Netif, currently skeletal tracking could be performed only on CPU.
Please tell what GPU model do you have ?

I have Geforce RTX 3080 TI

Do you use Nuitrack Pro skeletal tracking or Nuitrack AI (deep-learning based) ? Also please tell what CPU model do you have ?
Currently only object detection could be delegated to GPU, I’m now trying to figure out how we could help with your needs.

I’m using Nuitrack AI, and i have intel i9 12th.
I need run (5-8) applications using nuitrack AI with (60FPS and maximum resolutions). But, my processor i to weak to get 60 FPS, then i thought, can i run my 5-8 applications using GPU ?

And what happened next? I have the same GPU and CPU and working on a similar application

I’ve also encountered this problem, how did you solve this problem in the end

Hi @TAG,

I’ve also encountered this problem. Can the deep learning based skeletal tracking be delegated to GPU yet?

Hi there, GPU acceleration isn’t yet available for AI tracking engine, but taking all your replies into account - it should be prioritized :slight_smile:

@luluTS could you kindly specify - what particular hardware (CPU/GPU) do you use with AI tracking engine currently ? Is CPU not powerful enough to perform AI tracking ?

I am currently using an Intel® Core™ i7-8700K CPU @ 3.70GHz × 12. I am also using a GeForce GTX 1080 GPU. At the moment CPU is powerful enough. However, I was hoping to offload the computation to the GPU so I can run other applications in parallel as well as potentially getting 60 fps.