https://github.com/thisisbhavin/graphicalForest

Using the adjacency matrix and random forest get the Name, Address, Items, Prices, Grand total from all kind of invoices.
https://github.com/thisisbhavin/graphicalForest

adjacency-matrix graph graph-convolution graph-neural-networks information-retrieval invoice-parser random-forest

Last synced: 3 months ago
JSON representation

Using the adjacency matrix and random forest get the Name, Address, Items, Prices, Grand total from all kind of invoices.

• Host: GitHub
• URL: https://github.com/thisisbhavin/graphicalForest
• Owner: thisisbhavin
• Created: 2020-03-06T19:02:32.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-03-08T03:47:45.000Z (over 4 years ago)
• Last Synced: 2024-07-14T14:33:36.853Z (4 months ago)
• Topics: adjacency-matrix, graph, graph-convolution, graph-neural-networks, information-retrieval, invoice-parser, random-forest
• Language: Python
• Homepage:
• Size: 68.1 MB
• Stars: 17
• Watchers: 1
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Graph Convolution on Structured Documents

This repo contains code to convert Structured Documents to Graphs and implement a Graph Convolution Neural Network (incomplete) for Node Classification, each node being an entity in the document.

## Code

The `grapher.py` file contains the code to convert a structured document to a graph. 


An object map made using a Commercial OCR Tool is needed as the input which provides the bounding-box coordinates of each entity in the image along with it's recognized text. The script can then be used to generate an `object_tree.png` file and a 

`connections.csv` file. The script joins each object to it's nearest object to the right and underneath thus generating a graph. 

## References

1. Riba, Dutta et al - Table Detection in Invoice Documents by Graph Neural Networks - [Link](https://priba.github.io/assets/publi/conf/2019_ICDAR_PRiba.pdf)

2. Adam W. Harley, Alex Ufkes, and Konstantinos G. Derpanis - Department of Computer Science, Ryerson University, Toronto, Ontario - Evaluation of Deep Convolutional Nets for Document Image Classification and Retrieval - [Link](https://arxiv.org/abs/1502.07058)

3. Victor Garcia, Joan Bruna - Few-Shot Learning with Graph Neural Networks - [Link](https://arxiv.org/abs/1711.04043)