Hierarchical Lexical Database Image Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing technologies are not accurate enough for sorting and grouping images, particularly in research applications like identifying new land and marine creatures, due to low output probabilities, requiring significant human effort.
Innovation Solution
An image processing system that uses a hierarchically organized lexical database and image classification software to assign hyponym terms and probability values, building a subset hierarchical tree and calculating scores to identify hypernyms that accurately describe image content, improving classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image classification software is used to sort and group images, then the processing speed is improved, but the classification accuracy deteriorates due to low output probabilities
Solution Approach 1:
The patent introduces an intermediary hierarchical taxonomy system with pre-computed confidence scores between the image classification software and the final classification result. This intermediary layer aggregates probabilities across multiple hierarchical levels (from specific to general categories), transforming low-confidence single-level predictions into high-confidence multi-level classifications. The intermediary taxonomy structure mediates between the fast but inaccurate classification software and the requirement for accurate image sorting.
2Device complexity
If traditional image classification is used, then the device complexity is reduced, but the reliability of image identification deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing confidence scores for all taxonomic relationships in the hierarchical database before image classification occurs. This pre-computed confidence information is then rapidly retrieved and aggregated during image processing, eliminating the need for complex real-time probability calculations while significantly improving identification reliability. The preliminary preparation of confidence metrics allows the system to achieve high reliability without proportionally increasing device complexity.
Data Source
AI summary
An image processing system involves a camera, at least one processor associated with the camera, non-transitory storage, a lexical database of terms and image classification software. The image processing system uses the image classification software to assign hyponyms and associated probabilities to an image and then builds a subset hierarchical tree of hypernyms from the lexical database of terms. The processor then scores the hypernyms and identifies at least one hypernym for the image that has a score that is calculated to have a value that is greater than one of: a pre-specified threshold score, or all other calculated level scores within the subset hierarchical tree. The associated methods are also disclosed.