Multi-classifier MMR System for Image Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing image recognition process for printed media is computationally expensive and time-consuming, especially when dealing with large datasets, such as millions of pages, leading to inefficiencies in accurately recognizing images and locating them within input queries.
Innovation Solution
A Mixed Media Reality (MMR) system utilizing multiple classifiers to predict, monitor, and adjust index tables for improved image recognition, comprising mobile devices, a pre-processing server, and an MMR matching unit, which processes image queries and returns accurate document and location information, while also allowing for online performance monitoring and offline classifier prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image recognition processes are used to handle large datasets, then comprehensive image analysis is achieved, but computational cost and processing time increase significantly
Solution Approach 1:
The patent divides the large dataset into multiple smaller batches or segments that can be processed independently and in parallel. This segmentation allows the system to maintain recognition accuracy while reducing the computational burden on any single processing unit, thereby improving overall processing speed and productivity.
Solution Approach 2:
The patent implements preprocessing steps such as image normalization, feature extraction, and indexing before the main recognition process. By performing these preliminary actions, the system reduces the complexity of subsequent analysis, enabling faster processing without compromising recognition accuracy.
2Measurement precision
If multiple classifiers are used to improve recognition accuracy, then measurement precision increases, but device complexity increases
Solution Approach 1:
The patent divides the classification task into multiple specialized classifiers, each handling specific types of images or features. This segmentation allows the system to achieve high recognition accuracy through specialized processing while managing complexity by organizing classifiers in a modular, hierarchical structure that can be independently configured and maintained.
Solution Approach 2:
The patent designs a universal classifier framework that can handle multiple types of images and recognition tasks through a common architecture. This multi-functionality allows the system to achieve high accuracy across diverse applications while reducing overall complexity by avoiding the need for completely separate systems for each task type.
3Measurement precision
If index tables are continuously updated to maintain accuracy, then measurement precision is maintained, but processing time and computational resources increase
Solution Approach 1:
The patent implements periodic updates of index tables at scheduled intervals rather than continuous updates. This periodic action maintains search accuracy by ensuring index tables are refreshed regularly while minimizing processing time and computational resource usage by concentrating updates into discrete, manageable events rather than continuous operations.
Solution Approach 2:
The patent incorporates feedback mechanisms that monitor the quality and relevance of index tables, triggering updates only when necessary based on actual performance degradation or changes in data distribution. This feedback-driven approach maintains search accuracy while avoiding unnecessary updates that would waste processing time and computational resources.
Data Source
AI summary
A MMR system that uses multiple classifiers for predicting, monitoring, and adjusting index tables for image recognition comprises a plurality of mobile devices, a pre-processing server or MMR gateway, and an MMR matching unit, and may include an MMR publisher. The MMR matching unit includes a plurality of recognition unit and index table pairs corresponding to classifiers to be applied to received image queries, as well as an image registration unit for storing and monitoring performance data for the classifiers. The MMR matching unit receives the image query and identifies, using a classifier set, a result including a document, the page, and the location on the page corresponding to the image query. The present invention also includes methods for monitoring online performance of a multiple classifier image recognition system, for classifier selection and comparison, and for offline classifier prediction.