Image Recognition Verification Engine for False Positive Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition technologies face challenges in reliability due to poor image capture conditions and lack of image references, leading to failures in recognizing subjects or generating false positives.
Innovation Solution
A verification engine is introduced to select appropriate image processing and matching techniques based on various factors, including characteristics of query and candidate images, to verify the results of image recognition processes, using techniques such as region-of-interest analysis and signature generation to enhance matching accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image recognition is performed under poor image capture conditions or with limited image references, then the recognition process can still be executed, but the reliability of recognition results deteriorates due to false positives and failures to recognize subjects
Solution Approach 1:
The patent introduces a verification engine as an intermediary component between the image recognition process and the final result. This verification engine receives candidate images from the image recognition process, applies additional verification techniques (such as comparing multiple candidate images against each other and against the query image), and determines whether to accept or reject the recognition results. This intermediary verification step resolves the contradiction by maintaining productivity while significantly improving reliability through secondary validation.
2Reliability
If verification techniques are applied to increase recognition reliability, then false positives are reduced, but the computational complexity and processing time increase
Solution Approach 1:
The verification engine applies verification techniques selectively rather than uniformly to all image recognition cases. It assesses the quality and characteristics of candidate images and applies appropriate verification methods based on local conditions. For example, it may apply more rigorous verification only when candidate images show ambiguous characteristics or when the initial recognition confidence is low, while skipping verification for clearly identifiable subjects. This resolves the contradiction by improving reliability where needed while minimizing unnecessary computational complexity.
Solution Approach 2:
The verification engine performs partial verification rather than complete verification in all cases. It may verify only the top candidate images, or apply lighter verification techniques when the initial recognition is already highly confident. This partial action approach maintains reliability improvements while avoiding the full computational overhead of exhaustive verification, thus resolving the contradiction between reliability and system complexity.
3Measurement precision
If multiple verification techniques are applied to reduce false positives, then recognition accuracy improves, but the processing time and computational resources increase
Solution Approach 1:
The verification engine implements a multi-stage verification process where different verification techniques are applied in sequence rather than all at once. The first stage applies quick, computationally light verification methods to eliminate obviously incorrect candidates. Subsequent stages apply more rigorous verification techniques only to the remaining candidates. This periodic, staged approach resolves the contradiction by achieving high recognition accuracy through multiple verification techniques while minimizing total processing time by applying techniques in an optimized sequence.
Data Source
AI summary
Systems and methods of verifying the results of an initial image recognition process are presented. A verification engine can receive a set of candidate images corresponding to the results of an image recognition process performed on a captured query image. The verification engine can determine an appropriate verification technique to apply to the images of the candidate set, and classify, re-rank or otherwise re-organize the candidate set such that the best match from the candidate set is confirmed as a proper match.