A multistage extraction pipeline with page-level retrieval improves field-level accuracy by up to 31.9 percentage points over direct VLM application on 3000 pages of real multilingual KYC documents, reaching 87.27% with PaddleOCR and MiniCPM2.6.
Shaheera Saba Mohd Naseem Akhter and Priti P Rege
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Multistage Extraction Pipeline for Long Scanned Financial Documents: An Empirical Study in Industrial KYC Workflows
A multistage extraction pipeline with page-level retrieval improves field-level accuracy by up to 31.9 percentage points over direct VLM application on 3000 pages of real multilingual KYC documents, reaching 87.27% with PaddleOCR and MiniCPM2.6.