How to install Ikomia STUDIO on Linux

Here are the steps to install Ikomia STUDIO on Linux Ubuntu 18.04 but it’s the same process whatever your distro.

Ikomia is a cross-platform software dedicated to use, share and reproduce state of the art Computer Vision algorithms (see here for more details).

At first, you need to download Ikomia STUDIO here. For that, you have to sign-up for free or register with your credentials.

Note: The size of the installer is important because we bundle our own Python framework, all OpenCV libs and NVIDIA CUDNN dependencies.

Moreover, your Ikomia account gives you some advantages:

  • Access to state of the art algorithms from the HUB
  • Features like plugin integration and algorithms sharing
  • Access to the Q&A system to get support from the community

So now, you need to make the installer executable by right-clicking on it → permission → check “Allow execution” or by command line with

sudo chmod +x /path/to/IkomiaSetup

Then the next step is to launch IkomiaSetup and follow the instructions. Thus it creates two folders in your home directory:

  1. IkomiaInstall: it contains the EULA license and the uninstaller (maintenanceTool)
  2. Ikomia: it contains all the stuff needed for the Ikomia platform

Once it’s installed, search Ikomia and launch it.

In this version, we use the following tools:

  • OpenSSL 1.1
  • OpenCV 4.5.0 (with CUDA enabled)
  • VTK 8.1
  • Cudnn 7.6.5
  • Python 3.7
  • G’mic 2.8.0

Note: you must install NVIDIA CUDA toolkit 10.2  to leverage GPU computing with deep learning framework like PyTorch or Tensorflow.

Tested on these Linux distributions:

  • Manjaro
  • Linux Mint 19|20
  • Debian 10
  • Ubuntu 18.04|20.04
  • Elementary OS 5.0
  • Solus 4.0
  • Fedora 32
  • Deepin 15.11
  • Centos 7
  • Archlinux


Want to know more? Follow us on Twitter, LinkedIn and Youtube !!

Feel free to report issues or feature requests in our Github repository.

Please visit our website at

Categories: STUDIO


Leave a Reply

Avatar placeholder

Your email address will not be published.