Unmanned Store Checkout Using Multi-Sensor Product Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing checkout systems in stores require manual scanning of barcodes, leading to increased customer waiting times and higher operational costs due to the need for multiple staff, and there is a need for accurate unmanned product recognition.
Innovation Solution
A method and device utilizing a product recognition unit, cameras, weight sensors, thermographic cameras, and trained models to identify products by size, shape, weight, and temperature, with image matching to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual barcode scanning is used for checkout, then product identification can be performed, but customer waiting time increases and store congestion occurs
Solution Approach 1:
The patent replaces the mechanical manual barcode scanning process with an automated image recognition system using deep learning models. The system captures images of products on the conveyor belt and automatically identifies them through trained neural networks, eliminating the need for manual scanning and significantly reducing customer waiting time while maintaining accurate product identification
Solution Approach 2:
The patent employs pre-trained deep learning models that have been trained in advance on large datasets of product images. This preliminary training allows the system to rapidly identify products during checkout without requiring real-time processing delays, as the recognition capabilities are already established through prior training
2Reliability
If multiple staff members are deployed for checkout, then checkout accuracy can be maintained, but operational costs increase
Solution Approach 1:
The patent implements an automated checkout system that performs product identification and checkout processing without human intervention. The deep learning-based image recognition system autonomously captures product images, identifies items, and processes checkout, replacing multiple staff members while maintaining or improving checkout accuracy through consistent automated recognition
Solution Approach 2:
The patent introduces an intermediate image recognition system that acts as a mediator between the physical products and the checkout processing. This intermediary system captures images, processes them through trained models, and translates them into actionable checkout data, eliminating the need for human staff to directly handle product identification
3Productivity
If simple product recognition is used, then checkout speed increases, but recognition accuracy decreases
Solution Approach 1:
The patent employs a dynamic multi-stage recognition process that adapts to different product types and recognition scenarios. The system uses an initial quick recognition stage for fast processing, followed by a verification stage using additional trained models when needed, allowing the checkout speed to remain high while maintaining accuracy through conditional deeper analysis
Solution Approach 2:
The patent combines multiple deep learning models with different specialized capabilities into a composite recognition system. Each model is trained on specific product categories or characteristics, and their results are integrated to achieve both high speed and high accuracy recognition, similar to how composite materials combine different properties
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate and efficient product recognition at unmanned stores, reducing staff requirements and operational costs while improving checkout speed and accuracy.
Implementation Method 1
measuring the weight of the first product by means of a weight sensor
Implementation Method 2
measuring the temperature of the first product by means of a thermographic camera
Implementation Method 3
identifying the size and shape of the first product by means of cameras
Data Source
AI summary
Disclosed are a method and device for product checkout in an unmanned store. The method for product checkout in an unmanned store comprises the steps of: recognizing, by means of a product recognition unit, a first product being introduced to a checkout counter; selecting, on the basis of the recognition result for the first product, at least one trained model for identifying the first product; acquiring a first image of the first product by means of a product identification unit; and identifying the first product on the basis of the at least one trained model and the first image.