Masked OCR Region Detection for Banner Image Text Categorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition techniques struggle to efficiently categorize and annotate large volumes of banner images due to challenges in temporal sequencing, product variety, and human error, particularly in distinguishing text within banner images from other product text.
Innovation Solution
Employing Artificial Intelligence (AI) with Computer Vision (CV) and Natural Language Processing (NLP) techniques, utilizing region-based convolutional neural networks (R-CNN) and Optical Character Recognition (OCR) to detect and classify text regions within banner images, applying masks to reduce interference from other text, and employing neural networks for accurate categorization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If existing image recognition techniques are used to categorize banner images, then the process can be automated, but the accuracy and reliability of text categorization deteriorates due to difficulty in distinguishing text within banner images from other product text
Solution Approach 1:
The patent segments the banner image into multiple regions of interest (ROIs) using object detection algorithms. Each ROI is then processed separately through OCR to extract text, allowing the system to focus on specific areas rather than the entire image. This segmentation improves the reliability of text categorization by reducing interference from other text elements while maintaining automation.
Solution Approach 2:
The patent introduces an intermediary processing layer between image detection and text categorization. This intermediary uses region-based convolutional neural networks (R-CNN) to identify and isolate text regions, then feeds only the relevant text data to the categorization model. This intermediary step filters out irrelevant information and improves the accuracy of text categorization while preserving automation.
2Reliability
If manual annotation and categorization of banner images is performed, then the accuracy of text categorization is high, but the time consumption and labor requirements increase significantly
Solution Approach 1:
The patent implements a self-service system where the automated pipeline performs image detection, region identification, text extraction, and categorization without human intervention. The system uses trained neural networks to automatically classify banner images into product categories, eliminating the need for manual annotation while maintaining high accuracy through sophisticated algorithms.
Solution Approach 2:
The patent replaces manual mechanical processes of image annotation and text identification with automated computational systems. Object detection algorithms, OCR technology, and machine learning models work together to automatically categorize banner images, substituting human labor with intelligent systems that process images rapidly and accurately.
3Loss of information
If the entire banner image is processed for text extraction, then all text information is captured, but the computational complexity and processing time increase due to large image data volume
Solution Approach 1:
The patent divides the banner image into multiple smaller regions of interest (ROIs) using object detection. Instead of processing the entire large image, the system extracts and processes only the relevant text-containing regions. This segmentation reduces computational complexity while maintaining information completeness, as all text regions are captured through the segmented approach.
Solution Approach 2:
The patent extracts only the necessary text regions from the banner image using region-based detection algorithms. By taking out and processing only the relevant portions (ROIs) rather than the entire image, the system reduces computational complexity and processing time while preserving all important text information through targeted extraction.
Data Source
AI summary
Methods, apparatus, systems, and articles of manufacture are disclosed to categorize image text. An example apparatus includes region detection model training circuitry to identify candidate regions in an input image that include text, and generate bounding boxes around respective ones of the identified candidate regions. The example apparatus also includes mask application circuitry to improve optical character recognition (OCR) by applying a mask to the input image, wherein the mask removes content of the input image except for portions of the input image within the bounding boxes, and OCR circuitry to perform OCR on the masked input image to obtain text data within the bounding boxes.