Article Recognition Using Weight Verification for Overlapping Items
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing article recognition systems fail to recognize articles when they are disposed in an overlapping manner, leading to inefficiencies in processing and potential errors during settlement processes.
Innovation Solution
An article recognition apparatus and method that captures images of overlapping articles, uses a weight scale to measure weights, and employs a processor to extract article areas, recognize articles based on registered weights, and generate difference images to identify absent articles, thereby overcoming the challenge of overlapping articles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional image recognition is used to identify articles, then the system is simple and easy to operate, but it cannot recognize articles when they are disposed in an overlapping manner
Solution Approach 1:
A weight scale is introduced as an intermediary device to measure the weight of articles on the table. The processor compares the measured weight with the sum of registered weights of recognized articles to detect unrecognized articles, even those hidden in overlapping positions. This intermediary measurement mechanism enables reliable recognition without requiring direct visual access to all articles.
Solution Approach 2:
The system implements a feedback mechanism where the processor continuously compares the weight measured by the weight scale with the cumulative weight of articles that have been visually recognized. When a discrepancy is detected, the system identifies unrecognized articles and prompts the user to remove them, creating a closed-loop verification process that ensures complete article recognition.
2Quantity of substance
If multiple articles are disposed in an overlapping manner to save space, then the storage efficiency is improved, but the article recognition system fails to identify all articles
Solution Approach 1:
The weight scale serves as an intermediary that provides indirect information about the presence and quantity of articles. By measuring the total weight and comparing it with the sum of recognized article weights, the system can detect the presence of hidden articles in overlapping arrangements, maintaining detection accuracy regardless of article positioning.
Solution Approach 2:
The system transitions from purely two-dimensional image-based recognition to a three-dimensional approach by incorporating weight measurement as an additional dimension of information. This multidimensional verification allows the system to account for articles that are occluded in the visual field but contribute to the total weight measurement.
3Productivity
If the system processes articles one by one without weight verification, then the processing speed is fast, but errors occur during settlement processes
Solution Approach 1:
The system implements a feedback loop where the weight measurement continuously verifies the completeness of article recognition. The processor compares the measured weight with the cumulative weight of recognized articles, and when they match, the system confirms accurate recognition. This feedback mechanism ensures settlement accuracy while maintaining efficient processing through automated verification.
Solution Approach 2:
The system performs self-verification by automatically comparing weight measurements with recognized article weights without requiring manual intervention. The processor autonomously detects discrepancies and prompts users to remove unrecognized articles, enabling the system to self-correct and ensure accurate settlement processing.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
According to one embodiment, an article recognition apparatus includes an image interface, a weight interface and a processor. The image interface acquires an image captured by photographing a predetermined place where a plurality of articles are disposed. The processor acquires a first image, acquires a second image after detecting a predetermined event, recognizes an article, based on an image of an article area of an article which is absent in the second image, among article areas extracted from the first image, acquires a registered weight of the recognized article from an article database, and outputs an error if total of the registered weights disagrees with a difference weight between a first weight which a weight scale measures at a time of photographing the first image, and a second weight which the weight scale measures at a time of photographing the second image.