Document Information Extraction Using Segmented Object Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Manual data entry for extracting specific information from documents is inefficient, and conventional OCR methods struggle with language independence, low-resolution images, and limited computational resources.

Innovation Solution

An automated method using object detection algorithms based on convolutional neural networks identifies segments of interest in documents, builds bounding boxes, and extracts relevant information with high accuracy, even in low-resolution images, by leveraging visual document structure and customizing OCR solutions for specific use cases.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional OCR methods are used for data extraction, then character recognition can be performed, but processing time is long and memory requirements are high

Engineering Contradiction:

Improvecharacter recognition accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the document processing task into two distinct phases: first using object detection to identify and locate segments containing information of the predefined type, then applying specialized OCR only to those identified segments. This segmentation avoids processing the entire document with full OCR, thereby reducing processing time and memory requirements while maintaining character recognition accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary object detection and segment identification before applying OCR. By pre-identifying which segments contain the target information and building bounding boxes around them, the system prepares the document in advance, allowing subsequent OCR to be applied only to relevant portions, thus reducing overall processing time and resource consumption.

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If conventional OCR methods are used for data extraction, then character recognition can be performed, but memory requirements are high

Engineering Contradiction:

Improvecharacter recognition accuracyVSAvoidmemory requirements

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

Inventive Principle:

Principle #1Segmentation

3Measurement precision

If manual data entry is used for extracting information, then accuracy can be maintained, but productivity is low

Engineering Contradiction:

Improveextraction accuracyVSAvoiddata extraction efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent implements an automated system that performs information extraction without manual intervention. The object detection algorithm automatically identifies segments containing information of the predefined type, builds bounding boxes around them, and extracts the information using specialized OCR. This self-service automation maintains extraction accuracy while dramatically improving productivity compared to manual data entry.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical process of manual data entry with an automated computational system. The object detection algorithm and specialized OCR work together to automatically identify, locate, and extract information, substituting human manual operations with automated image processing and pattern recognition techniques.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

4Measurement precision

If specialized OCR is applied to the entire document, then information extraction accuracy is improved, but processing time increases

Engineering Contradiction:

Improveinformation extraction accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent extracts and isolates only the relevant segments from the document that are likely to contain information of the predefined type. By taking out these specific segments and applying specialized OCR only to them, the system avoids the overhead of processing the entire document, thus improving processing efficiency while maintaining extraction accuracy.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS11783572B2Method of automatically extracting information of a predefined type from a document

Publication Date: 2023.10.10 AMADEUS SAS

AI summary

Method and system of automatically extracting information of a predefined type from a document is provided. The method includes identifying a location and classification of a segment of interest of a document that includes information associated with a predefined type. The method further includes identifying a location and classification of characters from the segment of interest based on characteristics associated with the predefined type. The method further includes extracting the identified characters from the segment of interested associated with the predefined type.