Skip to content

Latest commit

 

History

History
35 lines (24 loc) · 1.29 KB

README.md

File metadata and controls

35 lines (24 loc) · 1.29 KB

Donut-Docs-Implementation

This repository utilizes Donut (Document Understanding Transformer) for information extraction from scanned images TNI-AU dataset. Document Types are:

• Decree_On_Implementation.

• Military_Education_Certificate.

• Rank_Promotion_Degree.


[UPDATE]

Accuracy of Model:

  • Rank_Promotion_Degree: - Total number of samples: 4, Tree Edit Distance (TED) based accuracy score: 0.7174657534246576, F1 accuracy score: 0.8181818181818182

  • Decree_on_implementation: - Total number of samples: 9, Tree Edit Distance (TED) based accuracy score: 0.7151542241291141, F1 accuracy score: 0.4727272727272727

  • Military_Education_Certificate: - Total number of samples: 3, Tree Edit Distance (TED) based accuracy score: 0.23286052009456262, F1 accuracy score: 0.24


[UPDATE]

All Data Files: