Automatic Cook Program Determination Using Adaptive Food Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated appliances struggle to recognize personalized and custom food items using generic computer-vision models, limiting their ability to determine appropriate cook programs for unique meals.
Innovation Solution
A method and system that includes a food identification module capable of determining new food classes through image recognition, using a combination of neural networks and clustering techniques, allowing for personalized cook program generation and storage, and enabling appliances to automatically cook unknown foods by training auxiliary sub-modules with minimal training data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a generic computer-vision model is used for food recognition, then the system can recognize common food items, but it cannot recognize personalized or custom food items
Solution Approach 1:
The food identification module is segmented into multiple specialized sub-modules, each trained to recognize specific food classes. This segmentation allows the system to handle diverse food items with specialized recognition capabilities rather than using a single generic model.
Solution Approach 2:
The system dynamically adapts by training new sub-modules on-demand when unrecognized food items are detected. The architecture evolves from a static generic model to a dynamic system that expands its recognition capabilities based on actual usage patterns and user needs.
2Adaptability or versatility
If the system trains auxiliary sub-modules with minimal training data, then personalized food recognition is achieved, but the training process requires user input and time
Solution Approach 1:
The system performs preliminary actions by pre-training auxiliary sub-modules with minimal training data before actual cooking operations. This preparation work is done in advance, so that when personalized food items are encountered, the recognition is already optimized without requiring time during the cooking process.
Solution Approach 2:
The system serves itself by automatically initiating the training of auxiliary sub-modules when unrecognized food items are detected. The training process is triggered autonomously based on system needs rather than requiring manual user initiation, reducing the perceived time loss for users.
3Adaptability or versatility
If the food identification module is updated with new food classes, then the system can recognize more personalized foods, but the device complexity increases
Solution Approach 1:
The food identification module is divided into multiple independent sub-modules, each responsible for specific food classes. This segmentation allows the system to expand food class coverage by adding or updating individual sub-modules rather than redesigning the entire system, thereby managing complexity.
Solution Approach 2:
The auxiliary sub-modules are designed to be universal and interchangeable. Each sub-module follows the same architecture and can be trained for different food classes, allowing the system to expand capabilities without increasing structural complexity. The same template serves multiple functions across different food recognition tasks.
Data Source
AI summary
In variants, the method can include: sampling cavity measurements of the cook cavity; determining an image representation using the cavity measurements; determining a food class based on the image representation; optionally comparing the image representation to prior image representations; optionally determining a new food class based on the image representation; optionally updating a food identification module with the new food class; and optionally determining a cook program associated with the new food class.