Synthetic-data-trained Tesseract ensemble with ROVER voting and post-processing cuts Maltese OCR CER from 0.0234 to 0.00700 on a competition dev set.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
LV-ROVER-MLT: Low-Resource Maltese OCR by Multi-Stream Voting
Synthetic-data-trained Tesseract ensemble with ROVER voting and post-processing cuts Maltese OCR CER from 0.0234 to 0.00700 on a competition dev set.