Skip to content

Various qdgga and symbolscraper fixes and improvements

Ayush Kumar Shah requested to merge qdgga_fixes into master

QD-GGA and SymbolScraper fixes in the pipeline

  1. Parse symbol scraper info using xml parser (ET)
  • Use xml.etree.ElementTree to parse the SymbolScraper output files
  • Remove number of CCs limit after extraction. So such formulas not skipped.
  • Count number of pages using system calls instead of using SS so that program works if no SS output available.
  1. Add translation to support lg2mml for infty
  2. Add pytest framework for test scripts: CC extraction and lg2mml
  3. Correct different scale input test images
  4. Replace Out-Of-Memory infty_strokes_newBest model by newly trained error free infty_contour model and update QD-GGA model links
  5. Fix PUNC mappings to MML
  6. Add custom Edmond's class
  7. Remove dead comments

Changes to root scripts

  1. Update install script to point to new model: outputs/run/infty_contour/net54.pt to be uploaded in the dprl server.
  2. Ignore generated lgeval and mprun scripts in git.
Edited by Ayush Kumar Shah

Merge request reports