Show HN: Extract Data from Line Chart Image

github.com

19 points·tdsone3··7 comments

Hey HN! Plextract extracts datapoints from a line chart image so you can compare and collect data from publications without manual effort! Enjoy and leave feedback.

Open articleView on HN

Great to see new tools in this space. So far we have been using WebPlotDigitizer, it has recently released new versions that stopped being open source but moved to be a paid software, requiring login, plastered with AI advertising terms.

https://automeris.io/

Would be interested in your use case if you don't mind sharing!

Most of the time, I just use it to scrape original data from screenshots taken from research papers where the data is not provided as a table. I would say 7/8 of these figures are scatter plots from experimental measurements.

Glad to see some work in this space. While I was doing my PhD all I had was https://datathief.org/, which usually did the job, but had some limitations and was Java-based. Definitely did some manual extraction from time to time.

Been there too haha! I wanted to do large scale data extraction from literature and manual approaches would have just been too slow.

looks great. Although in your example it seems like the y-axis is a bit off.

start: 0.05 vs 0.1

end: 0.65 vs 0.7

It's a difficult example!

Thanks! Yes, some images have hick-ups. Would probably need a UI that allows for easy post-prediction editing.

Reply on news.ycombinator.com