dmr-invoice-extractor Open-source Invoice Understanding Model - Free Deployment to Achieve Multi-language Invoice Key Field Extraction

Dmr Invoice Extractor

Developed by rhovhannisyan

Invoice understanding model fine-tuned based on the Donut base model, supporting key field extraction from multilingual invoices

Image-to-Text

Transformers

#Invoice information extraction #Multilingual document understanding #OCR-free processing

Downloads 57

Release Time : 6/28/2023

Model Overview

This model is based on the Donut architecture, specifically optimized for invoice documents, capable of extracting key fields such as amount, date, number, etc. from invoice images.

Model Features

Invoice-specific optimization

Specially fine-tuned for invoice documents, capable of accurately recognizing invoice-specific fields and formats

Multilingual support

Training data includes invoices from different countries and languages, providing certain multilingual processing capabilities

Efficient training

Training can be completed in just 4 hours on an NVIDIA RTX A4000 GPU

High-resolution processing

Input resolution set to 1280x1920 pixels, suitable for processing high-quality document images

Model Capabilities

Invoice image understanding

Key field extraction

Document classification

Multilingual processing

Use Cases

Financial automation

Invoice digitization

Automatically extract key information from scanned invoices to reduce manual entry

Improve financial processing efficiency and reduce error rates

Enterprise ERP systems

Accounts payable automation

Integration into ERP systems for automatic invoice matching and payment processing

Shorten payment cycles and improve financial operational efficiency

🚀 Donut finetuned on invoices

This project is a fine - tuned version of the Donut model specifically for invoice processing. It can effectively extract key information from invoices and other documents.

🚀 Quick Start

Refer to the documentation which includes code examples for quick usage.

✨ Features

Based on Donut Base: Built upon the Donut base model introduced in the paper OCR - free Document Understanding Transformer by Geewok et al. and first released in this repository.
Diverse Training Data: Trained with a few thousand annotated invoices and non - invoices from different countries and languages (one - page documents only).
Specific Input Resolution: Set to an input resolution of 1280x1920 pixels, samples with dpi higher than 150 have no added value.
Training Details: Trained for about 4 hours on an NVIDIA RTX A4000 for 20k steps, with a val_metric of 0.03413819904382196 at the end.
Rich Indexes: The train set includes indexes such as DocType, Currency, DocumentDate, etc.

📦 Installation

No specific installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

Refer to the documentation which includes code examples.

Advanced Usage

No advanced usage code examples are provided in the original document, so this part is skipped.

📚 Documentation

Model description

Donut consists of a vision encoder (Swin Transformer) and a text decoder (BART). Given an image, the encoder first encodes the image into a tensor of embeddings (of shape batch_size, seq_len, hidden_size), after which the decoder autoregressively generates text, conditioned on the encoding of the encoder.

model image

🔧 Technical Details

The model was trained with a few thousand of annotated invoices and non - invoices (for those the doctype will be 'Other'). They span across different countries and languages. They are always one page only. The dataset is proprietary unfortunately. Model is set to input resolution of 1280x1920 pixels. So any sample you want to try with higher dpi than 150 has no added value. It was trained for about 4 hours on a NVIDIA RTX A4000 for 20k steps with a val_metric of 0.03413819904382196 at the end. The following indexes were included in the train set:

DocType Currency DocumentDate GrossAmount InvoiceNumber NetAmount TaxAmount OrderNumber CreditorCountry

📄 License

The model is licensed under CC - BY - NC - SA - 4.0.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご