Wardrobe Outfit Recommendation Using Neural Coordination Scoring

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems lack the ability to effectively recommend clothing items based on a user's existing wardrobe and provide coordinated outfit suggestions that are both stylish and appropriate for the user's preferences and circumstances.

Innovation Solution

An electronic device equipped with neural networks processes images of clothing items to extract feature information, generates candidate coordination sets, evaluates their appropriateness, and recommends items based on similarity and expert/trendy sets, distinguishing selected items within recommended outfits.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the electronic device provides multiple recommended coordination sets with multiple clothing items, then the recommendation comprehensiveness is improved, but the user's ability to identify recommended items becomes difficult

Engineering Contradiction:
Improverecommendation comprehensivenessVSAvoiditem identification difficulty
Core Design Contradiction:
Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The patent applies visual differentiation techniques where recommended clothing items are displayed with distinctive markers, borders, or highlighting effects that change their visual appearance from non-recommended items. This allows users to easily identify which items are being recommended within complex coordination sets containing multiple clothing pieces.

Inventive Principle:
Principle #32Color changes

2Measurement precision

If the electronic device extracts detailed feature information from clothing images using neural networks, then the recommendation accuracy is improved, but the processing complexity increases

Engineering Contradiction:
Improverecommendation accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the complex image processing task into multiple stages: first extracting basic features (color, pattern, style) using neural networks, then using these features to generate coordination sets, and finally evaluating appropriateness separately. This segmentation reduces the complexity of any single processing step while maintaining overall accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces feature vectors as an intermediary representation between raw clothing images and coordination recommendations. The neural network converts images into compressed feature vectors that capture essential characteristics, which then serve as the basis for generating and evaluating coordination sets without requiring direct complex image processing at each stage.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If the electronic device generates and evaluates multiple candidate coordination sets, then the recommendation quality is improved, but the processing time increases

Engineering Contradiction:
Improverecommendation qualityVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent performs preliminary generation of multiple candidate coordination sets before final evaluation and recommendation. By pre-generating diverse coordination options and then efficiently evaluating their appropriateness based on extracted features, the system ensures high recommendation quality while managing processing time through structured evaluation criteria.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3905181B1Electronic device and method for operating same
Publication Date: 2026.03.04 SAMSUNG ELECTRONICS CO LTD
  • EP3905181B1 patent drawingFigure 1
  • EP3905181B1 patent drawingFigure 2
  • EP3905181B1 patent drawingFigure 3

AI summary

An electronic device includes a display, a memory storing one or more instructions, and a processor configured to execute the one or more instructions stored in the memory, wherein the processor is further configured to obtain a plurality of clothing images corresponding to a plurality of clothing items, and extract feature information corresponding to each of the plurality of clothing items by inputting the plurality of clothing images to a first neural network, generate candidate coordination sets by combining one or more clothing items among the plurality of clothing items, based on the feature information corresponding to each of the plurality of clothing items, obtain score information about each of the candidate coordination sets by inputting the candidate coordination sets to a second neural network, and control the display to display the candidate coordination sets based on the score information.