hi folks,
i’m not sure if this is the right forum. I was curious about the old Eggan Card project. (Links seem to be broken.). Nowadays, AI models can process and interpret images containing handwritten letters, observation logs, annotations, tabular records, etc. and accurately extract text for DB storage and indexing. I have a student working on building a data pipeline proof of concept. The end result will be a searchable database with a front-end dashboard. Interesting opportunities for data mining perhaps. We’re discussing a possible project with the Lowell Observatory here in Flagstaff. I wonder whether there may be similar needs at the AAVSO for this sort of application.
John Rachlin
Assoc Teaching Professor
Northeastern University
User: rachlin
John,
I moved this post over to the data mining forum as it might be a better fit there.
I would be very interested to see what your students could do to help us complete the Eggen Card project! Several years ago, John Crast wrote a paper in JAAVSO about the digitization effort. As you will read, there is considerable variations within the cards themselves, so this probably won’t be a straightforward application of OCR approaches.
Please let us know if you are interested in taking on this effort. We can provide you with access to the relevant databases.
Brian
Thank you Brian. The machine vision APIs for OpenAI models have proven themselves to be quite versatile at handwritten OCR in various formats. The challenge would be not so much extracting Eggen’s hand-written observations, but rather ensuring that the image-processing pipeline is cost-effective given the OpenAI API pricing structure for its various models. By all means, we would love to obtain database access so that we can do some further feasibility testing.
Hi John,
I’ll reach out to you after the annual meeting to make the relevant arrangements.
Brian