Lineless-to-Lined Table Conversion Using Conditional GANs

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing automated systems struggle to accurately detect and convert lineless tables into lined tables due to the lack of horizontal and vertical lines, leading to inefficiencies and errors in data extraction from documents like leases, invoices, and schedules.

Innovation Solution

A method using a conditional generative adversarial network (cGAN) and artificial neural network (ANN) is trained on table pairs with and without lines to predict line placement based on visual structure, enabling end-to-end conversion of lineless tables into lined tables.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If traditional automated systems are used to detect lineless tables, then the process can be automated, but the detection accuracy and conversion accuracy deteriorate due to the lack of visual line cues

Engineering Contradiction:

Improveautomation of table detectionVSAvoidtable structure detection accuracy

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent replaces traditional mechanical/optical line detection systems with a deep learning-based neural network system. The neural network learns to infer table structure and line positions from visual patterns in lineless tables, substituting physical line detection with intelligent pattern recognition and prediction algorithms.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If lineless tables are processed without line prediction, then processing speed is maintained, but data extraction accuracy deteriorates due to error propagation

Engineering Contradiction:

Improvetable processing speedVSAvoiddata extraction accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent applies preliminary action by predicting and establishing line positions before data extraction occurs. The neural network predicts horizontal and vertical line positions in advance, creating a structured framework that guides subsequent data extraction processes, thereby preventing error propagation while maintaining processing efficiency.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If manual line detection is used for lineless tables, then detection accuracy can be improved, but processing time and operational complexity increase

Engineering Contradiction:

Improveline detection accuracyVSAvoidoperational simplicity

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

The patent implements self-service by enabling the system to automatically detect and predict lines in lineless tables without human intervention. The neural network autonomously analyzes visual patterns, predicts line positions, and structures the table data, eliminating the need for manual line detection while maintaining high accuracy and operational simplicity.

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS12444163B2Apparatus and methods for converting lineless tables into lined tables using generative adversarial networks

Publication Date: 2025.10.14 EYGS LLP

AI summary

A method for converting a lineless table into a lined table includes associating a first set of tables with a second set of tables to form a set of multiple table pairs that includes tables with lines and tables without lines. A conditional generative adversarial network (cGAN) is trained, using the table pairs, to produce a trained cGAN. Using the trained cGAN, lines are identified for overlaying onto a lineless table. The lines are overlaid onto the lineless table to produce a lined table.