How to read the coordinates of a graph automatically

[WalkerSource]

No description provided.

[markummitchell]

There are 2 answeres to your question:

1. Currently you cannot run Engauge automatically - it requires a person manually guiding the process
2. In the future, it may be possible to run Engauge automatically 'mostly'

Engauge currently uses a manual process. I do not think Engauge itself can be upgraded to be automated since (1) it is already 60,000 lines of code and that upgrade would involve reworking most of that code which would take months of development and testing, and (2) the existing interface already has many features and adding automation would make the user interface more confusing.

However, there is a possibility that an automated preprocessor could perform most of the work and then pass the DIG file to Engauge for manual inspection and cleanup. There are so many things that can go wrong during automation (like an 8 looking like a 3) that some human oversight is usually needed. If anybody wants to develop such a preprocessor, I would do what I can to help.Personally I would lean to fast.ai and python for the implementation, but other ideas may be worthwhile too.

[taner45]

did you read paper from stanford university " Auto-Digitizer for Fast Graph-to Data conversion"?
I belive you can easly apply for your program " this Auto-digitizer..."
thanks in advance.

[WalkerSource]

@markummitchell @taner45 Thank you so much for your replies. This is my first try in Github. Your replies have increased my confidence and fun in this community.

[markummitchell]

Good input. I am unable to work on this in the immediate future. However, I am thinking somebody else might be interested in leading the task. I will put an advertisement on the web page to see if anybody volunteers.

[WalkerSource]

@markummitchell Great! This should be a very meaningful work!

[wilthan]

We use Engauge a lot for collecting data for our data repository of thermophysical properties for alloys (https://trc.nist.gov/metals_data/).
To solve the same automation problem we recently started to work on an AI system (https://aircconline.com/ijaia/V11N4/11420ijaia03.pdf, DOI: 10.5121/ijaia .2020.11403) and also created a training data set. Collaborators are welcome to join us in that effort and it would be great if we could include all that power in Engauge! (TRCalloy@nist.gov)