AI Gaze Region Validity Evaluation for Image Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing technologies face challenges in efficiently confirming the validity of prediction results in image recognition using artificial intelligence, particularly due to the large number of images used for model learning, which makes it difficult and costly to validate the basis of these predictions.

Innovation Solution

A program and information processing apparatus that evaluate the validity of gaze regions within prediction regions in images using artificial intelligence, by calculating and visualizing the contribution degree of each image region, allowing for the determination of valid, invalid, or uncertain gaze regions, thereby reducing the cost and effort required for validation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If image recognition is performed on a large number of images using AI, then the accuracy and comprehensiveness of recognition is improved, but the cost and difficulty of validating the basis of prediction results increases

Engineering Contradiction:
Improveaccuracy of prediction resultsVSAvoidtime and cost for validation
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent divides the image into multiple regions (first region, second region, third region) and evaluates the gaze region validity separately for each region. This segmentation allows the system to validate only specific portions of the image rather than the entire image, reducing validation time and cost while maintaining accuracy in identifying valid prediction bases.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different evaluation criteria and attention levels to different regions of the image. By determining gaze region validity specifically for regions where recognition targets are predicted to exist, the system optimizes validation resources toward the most critical areas, improving overall validation efficiency without compromising accuracy.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If the number of images used for learning AI model is increased, then the model's predictive capability is improved, but the difficulty of confirming validity of prediction bases increases

Engineering Contradiction:
Improvepredictive capabilityVSAvoidcomplexity of validation process
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system automatically evaluates the validity of gaze regions using predefined criteria and computational algorithms, eliminating the need for manual validation of each prediction base. This self-service approach maintains the ability to handle large numbers of images while simplifying the validation process through automated region-based assessment.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent performs preliminary evaluation of gaze region validity before finalizing prediction results. By pre-determining which regions are likely to contain valid prediction bases, the system reduces the complexity of subsequent validation steps and maintains efficient processing even with large image datasets.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS20240428572A1Program, information processing apparatus, and information processing method
Publication Date: 2024.12.26 SONY GROUP CORP
  • US20240428572A1 patent drawing
  • US20240428572A1 patent drawing
  • US20240428572A1 patent drawing

AI summary

A program causes an arithmetic processing device to execute a validity evaluation function of evaluating validity of a gaze region on the basis of a prediction region that is an image region in which a recognition target is predicted to exist by image recognition using artificial intelligence on an input image and the gaze region that is an image region that is a basis of prediction.