fr1ll / ras2las

Well log digitalization

ras2las

Well log digitalization

a.k.a. the young and the ras2las. Started at the geophysics hackathon before SEG 2017 annual meeting put on by Agile Scientific. Slides here

The intention was to adapt scatteract to well logs but this effort is stopped.

Goal:

Web app to digitize well logs. Input PDF --> output LAS

Contributors:

• Diego Castaneda
• Lanre Aboaba
• Chance Sanger
• Suman Gautam
• Will Sanger

Dependencies (partial list)

• Tensorflow
• Google Vision service (for OCR)
• Wand (for reading PDFs)
• welly and lasio (for handling .las format well logs)

Needs:

• More open training datasets with both raster and .las.
• Work on the key components, firstly:
  • Picking curve with CNN
  • Scaling curve picks based on axes labels
    • Method to reject bad axes label picks