Pose-Based Image Selection Using Weighted Human Keypoints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image selection methods using pose information may include images of individuals with poses different from the desired pose, leading to reduced selection accuracy.
Innovation Solution
An image selection apparatus and method that utilizes query information including relative positions of keypoints on a human body, determining weighting based on differences between reference and query pose information to select target images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If image selection is performed using basic pose information, then the selection process is simple, but the selection accuracy deteriorates
Solution Approach 1:
The patent segments the human body into multiple keypoints (e.g., head, shoulders, elbows, wrists, hips, knees, ankles) and processes each keypoint's pose information separately. This segmentation allows the system to capture detailed pose characteristics while maintaining a structured approach to image selection, resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The patent applies different weighting coefficients to different keypoints based on their importance for pose matching. By assigning higher weights to critical keypoints and lower weights to less important ones, the system achieves accurate pose-based selection without uniformly complicating the entire selection process, thus balancing simplicity and precision.
2Measurement precision
If weighting is applied to keypoints based on pose differences, then selection accuracy improves, but computational complexity increases
Solution Approach 1:
The patent pre-calculates and stores weighting coefficients for each keypoint before the actual image selection process. By preparing these weights in advance based on pose difference analysis, the system reduces computational complexity during real-time selection while maintaining high accuracy through differentiated keypoint weighting.
3Measurement precision
If relative positions of multiple keypoints are used, then pose matching precision improves, but information processing requirements increase
Solution Approach 1:
The patent extracts only the essential relative position information between keypoints needed for pose matching, rather than processing all possible pose parameters. By selectively extracting and using only the critical relative position data, the system achieves high pose matching precision while minimizing information processing requirements.
Data Source
AI summary
An image selection apparatus includes a query acquisition unit (610) and an image selection unit (620). The query acquisition unit (610) acquires query information indicating a pose of a person. The query information includes relative positions of a plurality of keypoints indicating different portions of a human body from each other. The image selection unit (620) selects at least one target image from a plurality of images subject to selection by using the query information. Specifically, the image selection unit (620) determines weighting for at least one of the keypoints by using a difference between reference pose information indicating reference relative positions of the plurality of keypoints and the query information. Then, the image selection unit (620) selects a target image by using relative positions of the plurality of keypoints of a person included in the image subject to selection, the query information, and the weighting.