For testing stereo vision Can I use Jetson Xavier nx inplace of 1080Ti(12G) for processing 512X1024 image?

Robotics & Edge Computing Jetson & Embedded Systems Jetson Xavier NX
1

I am Having Jetson Xavier Nx and two CSI IMX219 160FOV cameras.
I want to achieve stereo vision.
I am preferring this link-> GitHub - albert100121/360SD-Net: Pytorch implementation of ICRA 2020 paper "360° Stereo Depth Estimation with Learnable Cost Volume"

There is a note in this link that:-

  • The training process will cost a lot of GPU memory. Please make sure you have a GPU with 32G or larger memory.
  • For testing, 1080Ti (12G) is enough for a 512 x 1024 image.

My question is that Can I execute the above link through Jetson Xavier nx or Is there any problem or better option to do it?

PLease tell me as soon as possible.

1 Like
3

Hi,

Please noted that XavierNX only has 8GB memory which doesn’t meet the requirement.
You can find some detailed spec of XavierNX on the below page:

Usually, we recommend the user do the training job on a desktop GPU.
Since the Jetson platform is designed for runtime inferencing rather than training tasks.

Thanks.