Giters

xtreme1-io / xtreme1

Xtreme1 is an all-in-one data labeling and annotation platform for multimodal data training and supports 3D LiDAR point cloud, image, and LLM.

Home Page:https://www.basic.ai

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

some ideas about 'Enable model services' section

Yukun-Cui opened this issue · comments

commented

Enable model services

Make sure you have installed NVIDIA Driver and NVIDIA Container Toolkit. But you do not need to install the CUDA Toolkit, as it already contained in the model image.

# You need set "default-runtime" as "nvidia" in  /etc/docker/daemon.json and restart docker to enable NVIDIA Container > Toolkit
{
 "runtimes": {
   "nvidia": {
     "path": "nvidia-container-runtime",
     "runtimeArgs": []
   }
 },
 "default-runtime": "nvidia"
}

In Enable model services section, you need set default-runtime as nvidia. If you use Docker Desktop + WSL2.0, you have to set it at Docker Engine of Docker Desktop just like the picture below.

企业微信截图_16929318134631

commented

May I ask if you have successfully run the platform through WSL 2.0?
Is it because some components do not support WSL2.0, causing the platform to fail?

commented

Yes, I have successfully run the platform through Docker Desktop + WSL2.0.

commented

Great! I will update it to Readme!

commented

I am trying to annotate images for semantic segmentation. I would like to make use of pretrained models to assist this segmentation.
However, COCO for Object detection is the only available model.

Info:
Processor: 11th Gen Intel(R) Core(TM) i9-11950H @ 2.60GHz 2.61 GHz
Installed RAM: 64.0 GB
OS: Windows 10 Enterprise
GPU: NVIDIA RTX A3000 Laptop
CUDA Version: 12.3
Driver Version: 546.12

WSL2 Description:
Ubuntu 22.04.4 LTS
Release: 22.04
Codename: jammy

image

commented

You need to find a segmentation model.
Pre-trained models for semantic segmentation are generally less accurate, so it's better to label them manually.

commented