Merchant Logo Detection With Modular Machine-Learning Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems struggle to accurately identify and classify logos in images, particularly from merchant webpages and social media accounts, which is crucial for brand recognition and transaction verification.

Innovation Solution

A system utilizing machine-learning models, including logo detection and classification architectures, to automatically extract and verify logos from images, using ensemble decision models, semantic similarity scores, and feature extraction techniques to ensure accurate logo identification and merchant matching.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional logo identification methods are used, then the system is simpler to implement, but the accuracy of logo detection and classification deteriorates

Engineering Contradiction:
Improvelogo detection accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the logo identification task into multiple specialized machine learning models: a logo detection model that identifies whether an image contains a logo, a logo classification model that categorizes the logo type, and a merchant identification model that matches logos to merchants. This segmentation allows each model to specialize in a specific aspect, improving overall accuracy while managing complexity through modular architecture.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system transforms logo images into numerical feature vectors through machine learning models, converting visual data into a format that can be processed and compared algorithmically. This parameter transformation enables precise mathematical comparisons and classifications that significantly improve detection accuracy beyond traditional image processing methods.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If manual logo verification is used, then the system requires less computational resources, but the time and labor required increases

Engineering Contradiction:
Improvelogo processing speedVSAvoidverification time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The system implements automated self-verification through machine learning models that independently detect, classify, and verify logos without human intervention. The models automatically compare detected logos against merchant databases, perform semantic similarity analysis, and generate verification results, enabling the system to process numerous logos rapidly and eliminating manual verification bottlenecks.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system replaces manual visual inspection and verification processes with automated machine learning algorithms. The mechanical action of human eyes and brains analyzing logos is substituted with computational models that perform detection, classification, and matching operations, dramatically increasing processing speed and throughput.

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

3Measurement precision

If simple image matching is used, then the computational cost is lower, but the precision of merchant identification deteriorates

Engineering Contradiction:
Improvemerchant identification accuracyVSAvoidcomputational resources
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system applies different processing qualities to different aspects of logo analysis: lightweight feature extraction for initial filtering, medium-complexity classification for categorization, and more intensive semantic similarity analysis for final merchant matching. This localized application of computational resources optimizes the balance between accuracy and energy consumption by allocating intensive processing only where most needed.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system performs preliminary logo detection and classification before conducting detailed merchant identification. By pre-processing images to identify and categorize logos first, the system reduces the search space and enables more efficient subsequent matching operations, thereby improving final identification accuracy while managing computational costs through staged processing.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4641515A1System and method for logo detection and classification using machine- learning
Publication Date: 2025.10.29 FISERV INC
  • EP4641515A1 patent drawingFigure 1
  • EP4641515A1 patent drawingFigure 2
  • EP4641515A1 patent drawingFigure 3

AI summary

A system and method of identifying merchant logos may include one or more processors and a computer-readable, non-transitory medium including instructions which, when executed by the one or more processors, cause at least one of the one or more processors to obtain a logo image associated with a merchant, execute a logo detection machine-learning model using as input the logo image to determine whether the logo image is a logo, in response to determining that the image logo is a logo, execute a logo classification machine-learning architecture to identify the merchant.