Automatic labeling of products via expedited checkout system
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional checkout systems in stores are inefficient, leading to long wait times and high costs due to the need for multiple cashiers, while automated systems face challenges in accurately identifying products due to occlusion and lack of trained data, and require expensive human-labeled images for training.
Innovation Solution
A portable checkout unit that automatically generates training data by scanning items and capturing images, using cameras and sensors to create labeled image data, which can be used to train automated checkout systems, reducing development costs and improving product identification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional cashiers manually scan each item, then product identification accuracy is maintained, but checkout time and operational costs increase significantly
Solution Approach 1:
The patent replaces the mechanical manual scanning process with an automated computer vision system using cameras and machine learning models to detect and identify products in shopping carts, eliminating the need for manual barcode scanning by cashiers
Solution Approach 2:
The system enables self-service checkout by automatically tracking products as customers place them in shopping carts through camera detection, allowing customers to checkout without waiting for cashier intervention while maintaining accurate product identification
2Productivity
If automated POS systems are deployed, then checkout efficiency improves, but user interface complexity and difficulty of use increase
Solution Approach 1:
The system performs automated product detection and tracking without requiring customer interaction with complex interfaces, making the checkout process as simple as placing items in the cart while maintaining high efficiency
3Speed
If automated checkout systems use machine-learned models, then product identification speed improves, but training data generation costs increase due to requirement for human-labeled images
Solution Approach 1:
The system performs preliminary product identification during the checkout process itself, capturing images and generating training data in advance before the automated system is fully deployed, eliminating the need for expensive post-deployment data collection
Solution Approach 2:
The automated checkout system simultaneously performs its primary function of product identification while also generating training data for itself, creating a self-improving system that reduces external dependency for training data generation
4Extent of automation
If cameras capture images of products in shopping carts, then automated product detection is enabled, but image quality sufficiency decreases due to occlusion and edge cases
Solution Approach 1:
The system uses feedback from the checkout process where known product identities are confirmed through scanning, and uses this feedback to improve and refine the machine learning models for handling occlusion and edge cases
Solution Approach 2:
The system captures and processes images during the checkout process before deployment, using these preliminary captures to train and improve the automated detection system's ability to handle challenging scenarios
Data Source
AI summary
A portable checkout unit automatically generates training data for an automatic checkout system as a customer collects items in a store. A customer uses an item scanner of portable checkout unit to generate a virtual shopping list of items collected in the shopping cart. When the customer adds a new item to the shopping cart or on some regular interval, the portable checkout unit captures images of the items contained by the shopping cart and can generate bounding boxes for each product in each image. The bounding boxes can be associated with item identifiers from previously-generated bounding boxes to identify the items captured by the bounding boxes. Each bounding box paired with an item identifier can then be used as training data for an automated checkout system.