System and method for collecting and annotating cooking images for training smart cooking appliances, locating food items in smart cooking appliances, and determining cooking progress of food items in smart cooking appliances

Abstract

A food preparation system provides a user-friendly data collection and data annotation process that leverages crowd-sourced data to reduce the prohibitive costs associated with developing an extensive training set using dedicated cooking experiments and specialized manual annotation efforts. A method and system for locating food items and determining their outlines in the images captured by a food preparation system is disclosed herein. A method and system for assessing individual and / or overall cooking progress of food items in a food preparation system is disclosed herein.

Description

technical field [0001] The present disclosure relates to the field of cooking appliances, and more particularly to systems and methods for capturing and annotating cooking images to train an intelligent cooking appliance, to a system and method for training a cooking appliance and / or an associated server to determine a level of cooking progress of food in the cooking appliance Systems and methods, and related to systems and methods for training an intelligent cooking appliance and / or an associated server to determine the location and outline of food in the cooking appliance. Background technique [0002] Traditional food preparation systems, such as microwave ovens, stoves, ovens, electric cookers, ranges, and steamers, rely on manual input of cooking temperature and cooking duration parameters. These traditional systems require the user to have extensive knowledge and experience on how different foods should be heated and cooked to the proper doneness without compromising t...

AI Technical Summary

Problems solved by technology

However, due to the large amount of food available, different forms of food are available for different recipes, and images are captured under different conditions, resulting in very poor recognition results

Furthermore, the number of food ingredients and cooking recipes in the hundreds, and the training of AI and deep learning models requires large amounts of annotated training data, making these solutions difficult to deploy outside the confines of purely academic research settings

[0004] Furthermore, even when training images are available, the problem of designing an appropriate image processing method or data processing pipeline to determine the identity and / or cooking state of food inside a smart cooking appliance is critical in terms of prediction accuracy, model size, training efficiency, and / or runtime Very challenging in terms of speed

Images