Install add-on using command line with no GUI

105 次查看（过去 30 天）
Jan Jaap van Assen
评论： Henrique Chaves 2021-1-7
I have a question concerning the installations of add-ons, in this case AlexNet for the Neural Network Toolbox. I'm running my neural network code on a Linux system without GUI. How do install an add-on without a GUI?
Every Matlab document like here and here refers to the GUI menu which I can't access on my system.
Matlab Version: 9.2.0.556344 (R2017a), Neural Network Toolbox, Version 10.0 (R2017a).
I hope this is possible.

请先登录，再回答此问题。

回答（6 个）

Serge Lamberty
Serge Lamberty 2019-11-7
I just had the same problem, I found the following work-around, not too pretty but it works for now:
  • I installed the support packages on another pc.
  • I found the packages under /usr/local/MATLAB/R2019b/toolbox/nnet/supportpackages
  • I copied the whole folder "supportpackages" to the same location on the server where its supposed to run
  • On the server, add the "supportpackages"-folder (and subfolders) to the path
  • Done. Training running right now.
  2 个评论
Serge Lamberty
Serge Lamberty 2020-10-22
This does not sound like you are on a Linux system? /usr/local should always exist.

请先登录，再进行评论。

Sindhu Yerragunta
Sindhu Yerragunta 2017-8-10
编辑：Sindhu Yerragunta 2017-8-10
Hi Jan,
Installing MATLAB and/or a new toolbox onto an existing MATLAB installation without a graphical user interface or X server follows the same process as installing MATLAB with the same conditions. You will need to utilize the silent installer to complete the installation. Please refer to the link for in depth instructions:
http://in.mathworks.com/help/install/ug/install-noninteractively-silent-installation.html
Hope this helps.
-Sindhu
  3 个评论
显示 1更早的评论
Jan Jaap van Assen
Jan Jaap van Assen 2017-8-10
编辑：Jan Jaap van Assen 2017-8-10
Here another support package where people in the comments report the same issues. How to install this using commandline? One clear example would be great, they are all in the *.mlpkginstall format.
Mário Moraes
Mário Moraes 2017-8-11
编辑：Mário Moraes 2017-8-12
Please! I have the same issue. If you find the solution tell me damhur92@gmail.com

请先登录，再进行评论。

Yujin Hu
Yujin Hu 2017-11-28
Hi, did you guys fix it yet? I'm using Arch Linux with MATLAB 2017b, and I could not install the googlenet add-on by double click it, either. I just wonder if I could install it by command line or via matlab function. I know there is a function `matlab.addons.toolbox.installToolbox` to install toolbox, but it's not available for add-on.
  1 个评论
Jan Jaap van Assen
Jan Jaap van Assen 2017-11-28
In my case, using the AlexNet add-on, I ended up installing it on a system that did have a GUI and then just exported the data as a regular *.mat file which was enough for me. Never received a proper solution from Mathworks.

请先登录，再进行评论。

Yujin Hu
Yujin Hu 2017-11-30
I have not figure out any solution to this yet. But one might download add-on with a GUI (Windows, Linux, MacOS), and get the url of the pretrained model file in the downloaded file. It's in one of the xml file. And then one could just download the .mat file, and load it.
I'm wondering maybe the url is something like https://www.mathworks.com/supportfiles/nnet/data/vgg19.tar.gz, one might replace the filename and download other models.
Sumitabha Banerjee
Sumitabha Banerjee 2017-12-24
Just go to Mathworks Community File Exchange. Type in alexnet in the searchbox and download the file from there.
Taylor Johnson
Taylor Johnson 2019-9-10
We also have this issue (specifically to set up a tool to run on CodeOcean), and still, as of September 2019, I have not been able to find a reasonable solution beyond the workaround proposed above of exporting the data as a .mat file.
Is there any way to install these support packages without a GUI and directly through a script/command line?

类别

在 Help Center 和 File Exchange 中查找有关 Introduction to Installation and Licensing 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by