Does this repository include the scripts for OCR-based visual block extraction and the subsequent metric calculation used for the Design2Code benchmark? However, I noticed that the visual block extraction method seems to differ from the original Design2Code approach (which uses source-code modification to assign unique text colors for precise pixel tracking). Could you please clarify if the OCR-based evaluation code is available here?
Does this repository include the scripts for OCR-based visual block extraction and the subsequent metric calculation used for the Design2Code benchmark? However, I noticed that the visual block extraction method seems to differ from the original Design2Code approach (which uses source-code modification to assign unique text colors for precise pixel tracking). Could you please clarify if the OCR-based evaluation code is available here?