search

i13302 / learn

Apache License 2.0
1 stars 0 forks source link

Image Processing Library List up #1

Open unomitsu opened 3 years ago

unomitsu commented 3 years ago

Create an environment with containers for each purpose

Library List up

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

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 commented 3 years ago

(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 commented 3 years ago

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

i13302 commented 3 years ago

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

gacky35 commented 3 years ago

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 commented 3 years ago

OK. Use it.

unomitsu commented 3 years ago

(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 commented 3 years ago

(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 commented 3 years ago

(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 commented 3 years ago

(from unomitsu)

The reason is like this.

unomitsu commented 3 years ago

(from I.TAKEHARA)

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

unomitsu commented 3 years ago

(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 commented 3 years ago

(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.

unomitsu commented 3 years ago

(from I.TAKEHARA)

RUN apt-get update -y && apt-get install -y wget Can't you do it with the RUN wget URL?

The base image of Jupyter itself is Debian or Ubuntu.

Bring it off and push it.

unomitsu commented 3 years ago

(from I.TAKEHARA)

Oh, base looks like Ubuntu. The version is specified as expected. https://github.com/jupyter/docker-stacks/blob/32768af31671dc1ab8ec01448ae5bae0c58d2c49/base-notebook/Dockerfile#L7

Image of FROM Jupyter
RUN apt-get update -y && apt-get install -y wget
RUN wget graphViz entity
RUN installation work?
RUN pip install -r requestsment.txt

Isn't it okay? For the time being.

unomitsu commented 3 years ago

(from gacky35)

If it is based on Ubuntu, graphviz itself should be able to be installed with apt.

unomitsu commented 3 years ago

Proposed library selection

[figure make]

[graph make]

[Data Analysis]

unomitsu commented 3 years ago

library list