PatchFinder uses VLM token confidence to select patch size and the most confident patch, achieving 94% field-extraction accuracy on 190 noisy scanned well documents.
Automated structured data extraction from scanned document images,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Patchfinder: Leveraging Visual Language Models for Accurate Information Retrieval using Model Uncertainty
PatchFinder uses VLM token confidence to select patch size and the most confident patch, achieving 94% field-extraction accuracy on 190 noisy scanned well documents.