Image Processing Library List up

[unomitsu]

Create an environment with containers for each purpose

Library List up

requirements.txt : Bulk install the specified package with the specified version

• graphviz : Convert text to graph
• MNIST :
• Tensorflow :
• numpy :
• pandas :
• matplotlib :
• seaborn :
• mecab :

Workflow

1. Sample Code
2. Docker file, Introduce the library
   2.1 pip
   2.2 graphviz
3. check jupyter
   (2 and 3 may be mixed up)

Other

comments with (from [name]) are discussions in other places.

[unomitsu]

(from gacky35)

Proposal for creating a workflow

(from I.TAKEHARA)

1. Preparation of sample code
2. Check only the library installation in the Dockerfile
   2-1. pip
   2-2. graphviz?
3. Check Jupyter
   (2 and 3 may be around)

[i13302]

I think being able to use by used CSSS-EXP2 Code.
But it's not to used graphViz.

[i13302]

@gacky35 can you ready Sample Code using graphViz？
such as careful point, this code will be published in this repository.

[gacky35]

Does it have to be related to ML?
If not, I think we can quote the official sample program.
Please refer to this page.

[i13302]

OK. Use it.

[unomitsu]

(from unomitsu)

Proposed library selection to include.

Put a lot of libraries with the same purpose, You can find your favorite library while comparing.
How is it for learning purposes?

For exmaple, in creating a graph,
matplotlib, plotnin, altair, seaborn, (pixiedust), etc...

[unomitsu]

(from I.TAKEHARA)

How nice.

Ordinary libraries are just written in requirements,
It seems that it can be done in parallel with graphviz.

What about the library version?

[unomitsu]

(from gacky35)

It's the same pattern as the teacher.
I think there is.

In the requirements,
Which library is related to what,
It may be kinder to write something that you can understand at a glance.

The library is up to date for the time being.
If you're having problems with dependencies, I think it's okay to specify it at that time

[unomitsu]

(from unomitsu)

The reason is like this.

• Do not pollute your own PC environment (although Docker will be installed)
• Eliminate the trouble of preparing the environment when trying to study the library

[unomitsu]

(from I.TAKEHARA)

For discussions in other places, it seems better to put the issue in English.

[unomitsu]

(from gacky35)

If you put them all, As a reason, it may be okay to have the contents mentioned by unomitsu.
I think the readme is fine.

[unomitsu]

(from gacky35)

If you want to add graphviz, you can't use the base image Jupyter or Anaconda.

I have to drop the main body.
You can't do that unless you base it on Ubuntu.