Skip to content

Latest commit

 

History

History
107 lines (75 loc) · 4.43 KB

README.md

File metadata and controls

107 lines (75 loc) · 4.43 KB
Raw

Introduction | I want examples now ! | General info | Prerequisites | Installation | How-tos

Introduction

Reference

Intel Realsense SDK. Website. Github.

What's that repo?

Just a Docker image that makes you skip the whole Intel Realsense SDK 2.0 installation process. Simply run a container and start rs- examples. No additional applications, no fancy dependencies... just the source code! Moreover, all the images and their build are tested on 3 different machines with Ubuntu 20.04 to ensure they work properly!

Solved common issues (the real bois will know) :

  • Could not open OpenGL window, please check your graphic drivers or use the textual SDK tools when using rs-capture, rs-multicam like commands ;
  • All the gnupg stuff (package 'gpupgX' has no installation candidate and gnupgX does not seems to be installed etc.)
  • lsb_release: command not found and add-apt-repository: command not found

This repository contains release info and advanced image manipulation. See the project's Dockerhub tags info.

I want examples now

  1. Make sure to have the basic docker dependencies mentioned here.

  2. This will pull the image from Dockerhub and run a container (needs a GPU for Pangolin, container removed after exit)

sudo xhost +local:root && docker run --privileged --name realsense-2-container --rm -p 8084:8084 -e DISPLAY=$DISPLAY -e QT_X11_NO_MITSHM=1 -v /tmp/.X11-unix:/tmp/.X11-unix -v /dev:/dev:ro --gpus all -it lmwafer/realsense-ready:2.0-ubuntu18.04
  1. Run this inside the container to start a real time demo (with Intel D435i plugged in)
rs-multicam

You can run every example that comes along the SDK. Everything in the image is already built!

General info

The image is based on a Ubuntu 18.04 layer.

The images tag follows this template : <image version>-<os name><os version>. <os name> is the name of the Docker os not the system one, same thing for <os version>. <image version> is specific to <os name><os version>. That means image version refers to the work advancement for the Docker version.

Every dependency is installed in /dpds directory. Reach realsense directory with

cd /dpds/librealsense-2.50.0/

You may want better control of what's inside the image. To this matter you will find here :

  • Image Dockerfile.

  • docker-compose.yml to start container automatically and for Kubernetes-like deployement. Note that stopping a container removes it. An external app directory is linked to the containers /app one in order to provide a permanent save point.

  • Makefile to provide usual commands

Image prerequisites

  • Ubuntu 20.04

  • Docker (tested with Docker 20.10.7), see Install Docker Engine

  • Docker Compose (tested with Docker Compose 1.29.2), see Install Docker Compose You may have a /usr/local/bin/docker-compose: no such file or directory error. In this case, use

    sudo mkdir /usr/local/bin/docker-compose

    before restarting the installation process

  • Nvidia Container Toolkit (tested with ubuntu20.04 distribution), see NVIDIA Container Toolkit Installation Guide

  • A PC with GPU. Use the following to list available graphics units

    lshw -c display

Image installation

docker pull lmwafer/realsense-ready:<desired tag>

Image usage

All the commands need to be run in realsense-2-ready directory.

Get inside a freshly new container (basically up + enter)

make

Start an realsense-2-container (uses docker-compose.yml)

make up

Enter running realsense-2-container

make enter

Stop running realsense-2-container (and removes it, by default only data in /app is saved here in app directory)

make down

Build realsense-2-ready image (uses Dockerfile)

make build