Pose-Based Image Selection Using Exclusion Score Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image selection methods often include images with poses other than the desired pose, necessitating inefficient manual deletion.
Innovation Solution
An image selection apparatus and method that computes an exclusion score for each target image based on its similarity to exclusion pose information, allowing efficient selection and deletion of unwanted poses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image selection is performed based on pose information, then images with desired poses can be selected, but images with unwanted poses are also included in the selection results
Solution Approach 1:
The patent divides the image selection process into two separate stages: a positive selection stage that identifies images with desired poses, and a negative selection stage that removes images with unwanted poses. This segmentation allows each stage to focus on its specific task, improving overall selection accuracy by addressing both inclusion and exclusion criteria independently.
Solution Approach 2:
The patent introduces an intermediary exclusion image as a reference for negative selection. This exclusion image serves as a mediator that defines the characteristics of unwanted poses, allowing the system to compare target images against it and automatically identify images that should be excluded from the selection results.
2Reliability
If manual deletion of unwanted images is performed, then selection accuracy can be improved, but processing time and labor increase
Solution Approach 1:
The patent enables the system to automatically perform the deletion of unwanted images without requiring manual intervention. By computing exclusion scores and automatically removing images with high exclusion scores, the system serves itself to maintain high selection accuracy while eliminating the time loss associated with manual processing.
Solution Approach 2:
The patent transforms the selection process by introducing a new parameter - the exclusion score - that quantifies how much an image matches unwanted pose characteristics. By changing from purely positive matching to a system that also evaluates exclusion parameters, the system can automatically identify and remove unwanted images based on quantitative thresholds rather than manual review.
3Productivity
If exclusion score computation is performed for all target images, then unwanted images can be efficiently identified, but computational complexity increases
Solution Approach 1:
The patent applies partial action by computing exclusion scores selectively rather than uniformly for all images. The system focuses computational resources on images that are most likely to be unwanted based on initial positive selection results, performing detailed exclusion analysis only where necessary to maintain efficiency while managing computational complexity.
Data Source
AI summary
A search information acquisition unit acquires a plurality of pieces of search pose information generated for each of a plurality of target images and indicate a pose of a person included in the target image. An exclusion information acquisition unit acquires exclusion pose information indicating a pose of a person included in an exclusion query image. The exclusion query image is an image to be a query of an image needed to be excluded from a search result, and includes at least a person. An exclusion score computation unit computes an exclusion score for each of the plurality of pieces of search pose information. The exclusion score indicates a degree of similarity of the search pose information to the exclusion pose information. An exclusion image selection unit selects, from the plurality of target images by using the exclusion score, an image needed to be excluded from a search result.