Intelligent cookware bottom pasting prevention intermittent heating method and system based on image recognition

Abstract

The invention relates to the field of intelligent kitchens, in particular to an intelligent pot bottom-pasting-preventing intermittent heating method and system based on image recognition. The method comprises the following steps that a camera is used for continuously shooting kitchen photos, a pre-trained machine learning model is used for recognizing the kitchen photos, and a recognition result is obtained and comprises the position of an intelligent cooker and whether the intelligent cooker boils or not; and the intelligent cookware at the corresponding position carries out continuous heating and stopping circulation according to the duty ratio set by the user, namely the time ratio of the boiling state to the boiling stopping state, and the state data about whether boiling is carried out or not given by the machine learning model, so that the duty ratio set by the user is met. The boiling and non-boiling states of the intelligent cooker are detected by using the camera and the machine learning model, efficient intermittent heating can be realized without adding hardware, and the meal delivery time is shortened as much as possible while bottom pasting is prevented.

Description

Technical Field

[0001] The present invention relates to the field of intelligent kitchens, and particularly to an intelligent anti-scorching bottom intermittent heating method and system for cookware based on image recognition. Background Art

[0002] In modern cooking processes, especially in the automated preparation of Chinese dishes, heat conduction control technology faces unique challenges. Taking the typical stir-frying and reducing sauce process as an example, this process requires maintaining a specific boiling intensity to ensure the water evaporation rate. However, the traditional continuous heating method has significant technical defects: when a stable bubble layer forms at the bottom of the pot, the liquid-solid contact area decreases sharply, resulting in a decrease in the local heat transfer coefficient. At this time, the ingredients are in direct contact with the metal bottom of the pot for a long time locally, causing the bottom of the pot to burn. Traditional manual cooking realizes material turning through periodic pot-tossing operations, effectively controlling the interface contact time. However, the engineering implementation of this mechanical action in large-scale automated cooking equipment faces multiple obstacles, including but not limited to the complication of the driving mechanism (requiring a multi-axis linkage device), the risk of dynamic seal failure, and the increase in maintenance costs.

[0003] In existing automated solutions, the intermittent heating strategy can be adopted, and the thermodynamic effect of manual pot-tossing can be simulated by setting a power output mode with an adjustable duty cycle. However, the common practice is to directly set the heating and non-heating times and then perform a cycle. However, heating and non-heating do not exactly correspond to boiling and non-boiling. Therefore, there are significant limitations in using the above method to control the boiling and non-boiling processes: 1) The setting of the empirical threshold depends on manual experience and cannot be adaptively adjusted when dealing with different materials or liquid level changes; 2) The existing control system lacks a real-time closed-loop feedback mechanism and is difficult to cope with the dynamic changes in boiling intensity (such as affected by factors such as ambient air pressure and initial water temperature).

[0004] Precisely because in the process of intermittent heating, the time of boiling and non-boiling states cannot be accurately controlled. In the face of various situation changes, such as too high a proportion of boiling time, the ingredients are prone to burning at the bottom of the pot. On the contrary, if the proportion of boiling time is too low, it will affect the meal serving time. Detecting the boiling and non-boiling states and forming feedback can better control. However, traditional detection methods such as temperature sensors are generally used in the field of hot water kettles. If used in intelligent kitchenware, additional hardware needs to be added, and temperature sensors are extremely easy to be damaged in intelligent kitchenware that often needs to stir ingredients, which is not suitable.

[0005] It should be particularly noted that this technical contradiction is particularly prominent in the industrialized production scenario of group meals that pursue efficient continuous production, and has become a key technical bottleneck restricting the popularization of automated cooking equipment. Therefore, on the premise of minimizing the difficulty of equipment transformation, how to precisely control and achieve efficient intermittent boiling has become an important research direction for improving the reliability and adaptability of automated cooking equipment. Summary of the Invention

[0006] In view of the above problems, the present invention proposes the following technical solutions:

[0007] An intelligent anti-scorching bottom intermittent heating method for cookware based on image recognition, comprising the following steps

[0008] S1: Use a camera to continuously take pictures of the kitchen, and use a pre-trained machine learning model to recognize the kitchen pictures to obtain recognition results, where the recognition results include the position of the intelligent cookware and whether it is boiling;

[0009] The intelligent cookware at the corresponding position performs continuous heating and stopping heating cycles according to the duty cycle set by the user, that is, the cycle period of the boiling state and the stopped boiling state, and continuously corrects the heating and stopping heating times according to the state data of whether it is boiling given by the machine learning model to meet the duty cycle set by the user.

[0010] Further, in the recognition result, when the state is boiling, it also includes a boiling degree index;

[0011] During the same cooking process of the intelligent cookware at the corresponding position, each time the boiling state appears, calculate the ratio of the boiling degree index of this time to the maximum boiling degree index during this cooking process as the water volume change parameter. When the water volume change parameter is lower than the threshold, stop the intermittent heating process.

[0012] Further, each time boiling occurs, compare the boiling degree index of this time with the boiling degree indexes of the previous n times. When the deviation exceeds the deviation threshold, discard the boiling degree index of this time; where n is a preset parameter.

[0013] Further, the frequency of the camera taking pictures of the kitchen is 0.5Hz.

[0014] Further, the camera is placed in a closed container with a transparent front. The front of the container is automatically cleaned intermittently by a squeegee. At the same time, the frequency and start time of the automatic cleaning of the squeegee are such that the automatic cleaning of the squeegee does not affect the camera's taking pictures of the kitchen.

[0015] Further, the frequency of the camera taking pictures of the kitchen is 0.5Hz, the frequency of the automatic cleaning of the squeegee is 0.25Hz, and the start time is 0.5 seconds after the camera starts taking pictures.

[0016] Further, the frequency of the camera taking kitchen photos is dynamically adjusted as needed.

[0017] The present invention also provides an intelligent cookware system that controls intermittent heating using the above method.

[0018] Beneficial effects: Through intermittent heating, the present invention solves the problem that when an intelligent cookware continuously heats and boils, it is easy to burn the bottom without stirring or tossing the pot. At the same time, a camera and a machine learning model are used to detect the boiling and non-boiling states of the intelligent cookware. Without installing additional hardware (since there is inevitably a camera in a general kitchen), efficient intermittent heating can be achieved, preventing the bottom from burning while minimizing the meal preparation time. Specific embodiments

[0019] To make the objectives, features, and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the embodiments described below are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0020] An intelligent cookware anti-burning bottom intermittent heating method based on image recognition includes the following steps

[0021] S1: Use a camera to continuously take kitchen photos, and use a pre-trained machine learning model to identify the kitchen photos to obtain an identification result, where the identification result includes the position of the intelligent cookware and whether it is boiling;

[0022] S2: The intelligent cookware at the corresponding position performs continuous heating and stopping heating cycles according to the duty cycle set by the user, that is, the cycle period of the boiling state and the stopped boiling state, and continuously corrects the heating and stopping heating times according to the state data of whether it is boiling given by the machine learning model to meet the duty cycle set by the user. For example, the user needs to perform continuous intermittent heating with a cycle of boiling for 60 seconds and stopping boiling for 60 seconds. Initially, heat for 60 seconds and then stop heating for 60 seconds. According to the feedback of the machine learning model, it is detected that it is actually boiling for 40 seconds and stopping boiling for 80 seconds. At this time, increase the heating time and decrease the stopping heating time in the next cycle for correction until the feedback of the machine learning model meets the user's duty cycle expectation.

[0023] Regarding how long the intermittent heating lasts, it can be set manually. However, in a further embodiment, the duration can also be automatically controlled. Specifically, in the recognition result, when the state is boiling, it also includes an index of the boiling degree. In the same cooking process of the intelligent cookware at the corresponding position, each time the boiling state appears, the ratio of the boiling degree index of this time to the maximum boiling degree index in this cooking process is calculated as the water volume change parameter. When the water volume change parameter is lower than the threshold, the intermittent heating process is stopped. At the same time, to prevent errors in the recognition results of individual frames of the machine learning model caused by reasons such as artificial occlusion, each time boiling occurs, the boiling degree index of this time is compared with the boiling degree indexes of the previous n times. When the deviation exceeds the deviation threshold, the boiling degree index of this time is discarded; where n is a preset parameter.

[0024] In a further embodiment, the frequency of the camera taking pictures of the kitchen is 0.5 Hz. Using this frequency to take pictures can achieve the technical effect while reducing the computational amount. Especially when there are many intelligent cookware in a kitchen, appropriately reducing the frequency of the camera taking pictures of the kitchen can greatly reduce the requirements for hardware. In some embodiments, the frequency of the camera taking pictures of the kitchen is dynamically adjusted as needed. For example, when there is no boiling and no heating, it is impossible to boil at this time, so the shooting frequency can be reduced. When there is no boiling but heating has started, boiling may occur suddenly at any time, so the shooting frequency should be increased.

[0025] In some kitchens, the environment is relatively harsh, with a lot of water vapor, oil fumes, etc. in the kitchen. The camera lens is easily soiled, affecting the image effect. In this case, during the implementation of this solution, the camera is placed in a closed container with a transparent front. The front of the container is automatically cleaned intermittently by a squeegee. At the same time, the frequency and start time of the automatic cleaning of the squeegee are such that the automatic cleaning of the squeegee does not affect the camera's taking pictures of the kitchen. In this embodiment, the frequency of the camera taking pictures of the kitchen is 0.5 Hz, the frequency of the automatic cleaning of the squeegee is 0.25 Hz, and the start time is 0.5 seconds after the camera starts taking pictures.

[0026] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent intermittent heating method for preventing the bottom of a cooker from getting burnt based on image recognition, characterized in that: The following steps are involved: S1: Use a camera to continuously take pictures of the kitchen, and use a pre-trained machine learning model to recognize the kitchen pictures to obtain recognition results, where the recognition results include the location of the smart pot and whether it is boiling; S2: The smart cooker at the corresponding position performs a continuous heating and stopping cycle according to the empty ratio set by the user, that is, the cycle of boiling state and stopping boiling state, and then continuously corrects the heating and stopping time according to the boiling state data given by the machine learning model to meet the empty ratio set by the user.

2. According to claim 1, a method for intermittent heating of intelligent pots to prevent the bottom from burning based on image recognition, characterized in that: The identification result also includes a boiling degree indicator when the state is boiling; During the same cooking process, each time the smart cooker at the corresponding position reaches a boiling state, the ratio of the boiling degree index of this time to the maximum boiling degree index of this cooking process is calculated as the water volume change parameter. When the water volume change parameter is lower than the threshold, the intermittent heating process is stopped.

3. The method for intermittent heating of an intelligent pot to prevent the bottom from burning based on image recognition according to claim 2, characterized in that: Each time boiling occurs, the boiling degree index is compared with the boiling degree index of the previous n times. When the deviation exceeds the deviation threshold, the boiling degree index of this time is discarded; wherein n is a preset parameter.

4. The method of intermittent heating of intelligent pots to prevent the bottom from burning based on image recognition according to claim 1, characterized in that: The camera takes pictures of the kitchen at a frequency of 0.5Hz.

5. The method of intermittent heating of intelligent pots to prevent the bottom from burning based on image recognition according to claim 1, characterized in that: The camera is placed in a sealed container with a transparent front. The front of the container is intermittently and automatically cleaned by a scraper. At the same time, the frequency and start time of the automatic cleaning of the scraper are set so that the automatic cleaning of the scraper does not affect the camera's shooting of kitchen photos.

6. The method of intermittent heating of intelligent pots to prevent the bottom from burning based on image recognition according to claim 5, characterized in that: The camera takes pictures of the kitchen at a frequency of 0.5 Hz, and the scraper is automatically cleaned at a frequency of 0.25 Hz, and the start time is 0.5 seconds after the camera starts taking images.

7. The method of intermittent heating of intelligent cookware to prevent the bottom from burning based on image recognition according to claim 1, characterized in that: The frequency with which the camera takes pictures of the kitchen is adjusted dynamically as needed.

8. An intelligent cookware system, characterized in that: The intermittent heating is controlled using the method described in any one of claims 1 to 7.