Pose-Based Image Retrieval for Verifying Human Pose Estimation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing technologies face challenges in accurately estimating the pose of a capture target person in an image, as the correctness of pose estimation is not reliably verified, leading to inaccuracies in determining similar poses or actions.

Innovation Solution

An information processing apparatus and method that includes a pose acquisition unit to acquire pose information from reference and query images, a retrieval unit to find similar poses based on similarity degrees, and a display control unit to display the pose information for retrieval, enhancing accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If pose estimation is performed using existing image processing techniques, then pose information can be obtained from images, but the accuracy of pose estimation cannot be verified or improved

Engineering Contradiction:
Improvepose estimation accuracyVSAvoidverification capability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system retrieves reference images with similar poses from a database and displays them alongside the query image. This feedback mechanism allows users to verify whether the estimated pose is correct by comparing it with actual reference images, thereby improving both the verification capability and the accuracy of pose estimation through iterative refinement.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system pre-collects and stores reference images with known correct pose information in a database before the actual pose estimation task. This preliminary preparation enables subsequent verification and accuracy improvement by having ground truth data readily available for comparison with estimation results.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If pose estimation is performed without reference verification, then the processing speed is maintained, but the accuracy of pose estimation remains insufficient

Engineering Contradiction:
Improvepose estimation accuracyVSAvoidprocessing efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The system performs pose estimation on only certain selected images or in certain scenarios where verification is particularly valuable, rather than applying the full verification process to every single image. This partial application of the verification mechanism maintains overall processing efficiency while still improving accuracy in critical cases.

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If reference images are retrieved and displayed for verification, then pose estimation accuracy can be improved, but the system complexity increases

Engineering Contradiction:
Improvepose estimation accuracyVSAvoidsystem structure
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system employs a multi-functional architecture where the image database serves both as a storage repository and as an active verification tool. The retrieval unit and display unit enable the same database infrastructure to support both routine operations and accuracy verification, reducing the need for separate dedicated verification components.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS12625904B2Information processing apparatus, information processing method, and non-transitory computer readable medium for image retrieval based on pose information
Publication Date: 2026.05.12 NEC CORP
  • US12625904B2 patent drawing
  • US12625904B2 patent drawing
  • US12625904B2 patent drawing

AI summary

An information processing apparatus includes a pose acquisition unit, a retrieval unit, and a display control unit. The pose acquisition unit acquires first pose information indicating a pose of a person shown in each of a plurality of reference images associated with a predetermined pose, and second pose information indicating a pose of a capture target person shown in a query image. The retrieval unit retrieves a reference image showing a person whose pose or action is similar to that of the capture target person shown in the query image, from among the plurality of reference images, based on a similarity degree between the first pose information and the second pose information. The display control unit causes a display unit to display at least one of the first pose information and the second pose information in a display mode according to an index used for retrieving the reference image.