Image Spatial Relationship Evaluation Using VISOR Metrics

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing techniques for automated evaluation of image quality fail to effectively characterize how well spatial relationships between objects in an image match those expressed by associated text.

Innovation Solution

Developed metrics, referred to as VISOR, to evaluate whether spatial relationships between objects in an image match corresponding relationships expressed by text, utilizing object detection and spatial relationship analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing automated evaluation techniques are used, then image quality can be evaluated, but spatial relationships between objects cannot be effectively characterized

Engineering Contradiction:
Improvespatial relationship accuracyVSAvoidevaluation system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The evaluation is segmented into distinct components: object detection module, spatial relationship extraction module, and matching module. Each component handles a specific aspect of the evaluation independently, making the complex task manageable while improving precision in spatial relationship assessment

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces intermediate representations such as bounding boxes and spatial relationship predicates that mediate between image data and text descriptions. These intermediaries enable precise comparison of spatial relationships without directly comparing the entire image-text pair, reducing computational complexity

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If manual evaluation is used, then spatial relationships can be accurately assessed, but evaluation time and labor cost increase

Engineering Contradiction:
Improvespatial relationship assessment accuracyVSAvoidevaluation time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-service evaluation by automatically detecting objects, extracting spatial relationships, and comparing them with text descriptions without requiring human annotators. The automated pipeline handles the entire evaluation process independently, eliminating time-consuming manual assessment while maintaining accuracy

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces manual visual inspection and judgment with automated computer vision and natural language processing systems. Object detection algorithms and spatial relationship extraction mechanisms substitute for human evaluators, enabling rapid automated assessment of spatial relationships in images

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20260017862A1Automated evaluation of spatial relationships in images
Publication Date: 2026.01.15 MICROSOFT TECHNOLOGY LICENSING LLC
  • US20260017862A1 patent drawing
  • US20260017862A1 patent drawing
  • US20260017862A1 patent drawing

AI summary

This document relates to automated analysis of images. One example method involves obtaining an image and text associated with the image, detecting two or more objects in the image, and determining respective locations of the two or more detected objects in the image. The example method also involves determining whether a spatial relationship between the two or more detected objects matches a corresponding spatial relationship expressed by the text based at least on the respective locations of the two or more detected objects. The example method also involves outputting a value reflecting whether the spatial relationship between the two or more detected objects matches the corresponding spatial relationship expressed by the text.