Visual Product Identification via AR and Machine Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current inventory management systems in retail environments face challenges in accurately identifying products using traditional methods, which can lead to inefficiencies and errors in tracking and monitoring inventory.

Innovation Solution

A visual product identification system utilizing object detection with machine learning, enhanced by image acquisition with augmented reality, allows for improved product classification and verification, with a feedback loop for continuous model improvement based on usage data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional product identification methods are used, then the system is simple to operate, but the identification accuracy is low

Engineering Contradiction:
Improveproduct identification accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces traditional mechanical/manual product identification methods with a visual-based machine learning system. The system uses image capture devices to capture product images, processes them through trained machine learning models, and automatically identifies products, substituting manual operations with automated visual recognition technology.

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

Solution Approach 2:

The patent creates visual copies (images) of physical products and processes these copies through digital machine learning models. Instead of directly manipulating or examining physical products, the system captures their visual representations and performs identification on these digital copies, enabling automated and accurate product recognition.

Inventive Principle:
Principle #26Copying

2Reliability

If traditional inventory tracking methods are used, then the process is simple, but errors in tracking and monitoring occur

Engineering Contradiction:
Improveinventory tracking accuracyVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements a feedback mechanism where the machine learning model continuously processes product images, compares them against trained data, and provides identification results. The system includes feedback loops that allow for model retraining and improvement based on actual usage data, ensuring continuous enhancement of tracking reliability.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system performs self-verification through automated machine learning models that independently identify and track products without requiring manual verification. The model serves itself by automatically processing images, making decisions, and improving through accumulated data, reducing human intervention and associated errors.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If visual product identification with machine learning is implemented, then identification accuracy is improved, but the system complexity increases

Engineering Contradiction:
Improveproduct identification accuracyVSAvoidsystem operation ease
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The machine learning system operates autonomously, automatically capturing images, processing them through trained models, and generating identification results without requiring complex manual operations. The system self-manages the identification process, making it easy to operate despite the underlying complexity of the machine learning technology.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3929832A1Visual product identification
Publication Date: 2021.12.29 MAUI JIM INC
  • EP3929832A1 patent drawingFigure 1
  • EP3929832A1 patent drawingFigure 2
  • EP3929832A1 patent drawingFigure 3

AI summary

A model for visual product identification uses object detection with machine learning. Image acquisition with augmented reality can improve the model's identification, which may include classifying those images that is further verified with machine learning. Usage of the visual product identification data can further improve the model.