InvoiceLens – Document Extraction Model

Fine-tuned vision-language model that extracts line items, GSTIN, totals and taxes from Indian invoices with 97% field accuracy.

by Sofia MartinsVerified Author(5) 84 sales 4.5K viewsUpdated 17 Aug 2026
InvoiceLens – Document Extraction Model — preview 1
1 / 3

InvoiceLens automates accounts-payable data entry with a compact model you can run on your own infrastructure.

Key features

  • Line items, taxes and totals extraction
  • GSTIN and PAN validation
  • Handles scanned, photographed and digital PDFs
  • REST API with confidence scores
  • Human review UI for low-confidence fields

Tech stack

  • PyTorch
  • FastAPI
  • React review UI

Requirements

  • GPU with 12 GB VRAM recommended
  • Python 3.11+

Installation

  1. Install requirements and download weights.
  2. Start the API with uvicorn app:api.
  3. Open the review UI at /review.

What's included

  • Full source code with comments
  • Step-by-step documentation
  • Demo data and seed scripts where applicable
  • 12 months of author support and all future updates

More from this author

View all

Related in AI & Machine Learning Tools

View all