Food Recognition Using Image Segmentation and Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In dining facilities like cafeterias, the existing commodity recognition systems are inefficient as they primarily recognize containers rather than the cooked food itself, limiting the freedom of container selection and prolonging checkout processing.
Innovation Solution
A commodity recognition apparatus using object recognition technology captures images of cooked food, extracts feature data, and compares it to a recognition dictionary to identify the food items, allowing users to select and recognize items directly, eliminating the need for container-specific recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If container recognition is used to identify cooked food, then the recognition process is simplified, but the container selection freedom is limited and recognition accuracy decreases
Solution Approach 1:
The patent extracts the food item from the container for independent recognition. Instead of recognizing the container as a whole, the system extracts and identifies individual food items within the container, allowing any container type to be used while maintaining accurate food recognition
Solution Approach 2:
The patent segments the recognition process into two stages: first capturing the overall container image, then dividing and recognizing individual food items within the container. This segmentation allows the system to handle various container types while accurately identifying specific food items
2Measurement precision
If manual registration of each cooked food item is performed, then recognition accuracy is high, but checkout processing time increases
Solution Approach 1:
The system performs automatic food item recognition without requiring manual registration by cashiers. The recognition apparatus autonomously captures images, identifies food items, and processes checkout information, eliminating manual labor while maintaining high recognition accuracy
Solution Approach 2:
The patent replaces the mechanical manual registration process with an automated image recognition system. The recognition apparatus uses optical imaging and pattern recognition algorithms to automatically identify food items, substituting human operation with automated technological processes
3Extent of automation
If conventional recognition apparatus is used, then checkout processing is automated, but container selection freedom is limited
Solution Approach 1:
The recognition apparatus is designed with universal functionality to handle various container types. The system can recognize food items regardless of the container material, shape, or size, making it adaptable to different cafeteria operations while maintaining full automation
Data Source
AI summary
The commodity recognition apparatus displays a frame for surrounding a commodity in an image captured by an image capturing module. Then the commodity recognition apparatus recognizes a candidate of the commodity imaged in the frame according to the feature amount of the image in the area surrounded by the frame, and outputs information of a highest ranked candidate commodity. If a change instruction for the candidate is received, the commodity recognition apparatus outputs information of the commodity other than the highest ranked candidate selected from the candidates.