OCR Accuracy via Object Detection and Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional optical character recognition (OCR) techniques are limited by reliance on user input and similarity-based image analysis, leading to inaccurate results, especially in environments without user input and in complex image contexts.
Innovation Solution
An online system that employs object detection and image classification models to identify objects and labels within images, using machine learning to improve OCR accuracy by analyzing discrepancies and predicting text across overlapping zones, and providing recommendations for improved image capture conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional OCR techniques are used, then the system is simple to operate, but the accuracy is poor and requires user input
Solution Approach 1:
The system segments the image processing into distinct modules: object detection module, image classification module, and OCR module. Each module handles specific tasks independently, improving overall accuracy while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent introduces object detection and image classification models as intermediary components between the input image and OCR processing. These intermediaries provide contextual information that guides the OCR process, enabling accurate text recognition without requiring user input by automatically understanding the image content and expected text locations.
2Measurement precision
If user input is required to specify text field locations, then the system can be more accurate, but the ease of operation decreases
Solution Approach 1:
The system performs self-service by automatically detecting objects, classifying images, and identifying text regions without requiring user input. The object detection and image classification models autonomously determine where text is likely to be located based on the image content, eliminating the need for users to manually specify text field locations while maintaining high accuracy.
3Reliability
If conventional similarity-based OCR is used, then the processing is fast, but the reliability is low in complex contexts
Solution Approach 1:
The system performs preliminary object detection and image classification before executing the OCR process. By pre-identifying the objects and expected text locations through machine learning models, the system can then focus the OCR processing only on relevant regions, improving reliability in complex contexts while maintaining efficient processing speed through targeted analysis rather than analyzing the entire image.
Data Source
AI summary
A system performs optical character recognition (OCR) on an image displaying a portion of an object. An image classification system identifies the object in the image, based on which one or more object detection models identify labels associated with the object within the image. The system determines text of the identified labels using OCR, and analyzes the OCR resultant text for discrepancies and/or inaccuracies. In response to identifying a discrepancy, the system provides a recommendation for improving the accuracy of the OCR resultant text.