Checkout Image Item Identification Using Scanner-Labeled Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Training machine learning models for object identification in images is time-consuming due to the need for manual annotation of datasets, and new items introduced into a store are not recognized until the model is updated with additional training data.
Innovation Solution
A method that automatically trains a visual identification model using information from a scanner system, correlating scan times with image capture times to annotate and train the model with item identities, leveraging existing scanner data to reduce manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation of images is used to train the model, then the model can be trained with accurate labels, but the process is very time consuming
Solution Approach 1:
The scanner system automatically performs the annotation function that previously required manual user input. When an item is scanned, the system automatically captures an image and uses the scanner's item identification to create labeled training data, allowing the system to serve itself rather than requiring human annotators.
Solution Approach 2:
The patent replaces the manual mechanical process of user annotation with an automated electronic system. The scanner system electronically identifies items and triggers automatic image capture and data processing, substituting human manual operations with automated mechanical/electronic processes.
2Productivity
If the model is trained only on currently stocked items, then the training process is efficient, but new items cannot be identified until the model is updated
Solution Approach 1:
The system enables continuous learning by automatically capturing images and creating training data in real-time as items are scanned. This continuous stream of new data allows the model to continuously update its knowledge without interruption, maintaining both efficiency and adaptability.
Solution Approach 2:
The system performs preliminary actions by automatically capturing images and preparing training data in advance as items are scanned. This preliminary data collection and processing enables the model to be updated proactively with new items before they need to be recognized, ensuring continuous adaptability.
3Loss of time
If the scanner system automatically captures images and uses scanner information for training, then the time required for model updates is reduced, but the system complexity increases
Solution Approach 1:
The patent merges multiple functions into the existing scanner system: item identification, image capture triggering, data correlation with timestamps, and training data generation. By combining these functions into a unified system, the patent reduces overall complexity compared to having separate systems for each function.
Solution Approach 2:
The scanner system is enhanced to perform multiple functions: it not only identifies items but also triggers image capture, manages timing data, and contributes to model training. This multi-functionality reduces the need for separate dedicated systems, thereby managing complexity while achieving automation.
Data Source
AI summary
The technology disclosed herein enables identification of items in an image using a machine learning model that is automatically trained using information captured by a scanner system. In a particular example, a method includes receiving an image captured at a capture time of a checkout space including a scanner system and receiving an indication that an item has been scanned by the scanner system. The indication includes an identity of the item and identifies a scan time when the item was scanned. The method also includes correlating the scan time with the capture time and providing the image and the identity of the item to a visual identification model to train the visual identification model to identify the item from other images.