A data-driven food safety remote monitoring method and system

Abstract

The present invention relates to the technical field of remote supervision of food safety, and in particular to a data-driven remote supervision method and system for food safety. The present invention can detect the upper opening of a food storage box placed in a food storage box each time through a food storage box, a picture acquisition module, an analysis and judgment module, and a food supervision module, and screen out the food storage boxes whose upper openings are not completely covered by the plastic wrap, and can quickly mark the food storage boxes on the corresponding material placement positions, and directly remind employees on site that the food ingredients cannot be used and need to be discarded. The remote monitoring end can remind employees in real time that the food ingredients cannot be used and need to be discarded, and can replace manual supervision of the food ingredients in the food storage box. The "out-only but not in" management of the food ingredients in a food storage box is realized, and dual-channel management of the food ingredients by the on-site end and the supervision end is achieved, which improves the supervision intensity and precision of the remote management of food during the food processing process, and greatly improves food safety during the food processing process.

Description

Technical Field

[0001] The present invention relates to the technical field of food safety remote supervision, and in particular to a data-driven food safety remote supervision method and system. Background Art

[0002] Data-driven means collecting massive amounts of data through mobile Internet or other related software, organizing the data into information, and then integrating and refining the relevant information, and forming automated decisions through training and fitting based on the data.

[0003] The large-scale promotion of "Internet +, open kitchen and bright stove" has brought new supervision methods to the food safety in the kitchen, making the kitchen hygiene management and food processing operation management more standardized. In the process of food cooking and processing, a lot of ingredients are needed. In order to facilitate taking and using, some ingredients will be placed in food storage boxes. The upper opening of the food storage box is covered with plastic wrap. The plastic wrap plays the role of isolation and preservation. If the plastic wrap does not cover the entire upper opening of the food storage box, the ingredients in the food storage box will be exposed and contaminated. In addition, a special person is required to check regularly whether the food storage time has exceeded the limit. If it exceeds the limit, it cannot be used anymore. If the expired ingredients are not taken out in time, the camera in the kitchen cannot identify the expired ingredients in time, which will cause food safety hazards and fail to ensure food safety. Summary of the invention

[0004] The purpose of the present invention is to provide a data-driven remote food safety supervision method and system to solve the existing problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a data-driven food safety remote monitoring system, the system comprising:

[0006] The food storage box is used to store food ingredients for food processing, seal and preserve the food ingredients, obtain the weight value and placement time of the food ingredients in real time, and use the time when the reset switch is pressed as the starting point for the accumulation of the food ingredient retention time. Each time the food ingredients are placed on the placement position, the measurement value of the gravity scale is the weight value;

[0007] The model training module is used to obtain training samples and build a classification model based on a convolutional neural network. It is also used to optimize the parameters of the built classification model using the collected training samples as input, output the trained classification model, and classify the data collected on-site through the classification model.

[0008] The image acquisition module is used to obtain the video image of the door of the food storage box, and then obtain the image of each time the food enters and exits the food storage box, and also obtain the space monitoring video of the kitchen;

[0009] The analysis and judgment module is used to analyze the acquired weight value, retention time and video footage at the box door to determine whether the food in the food storage box meets the food processing standards.

[0010] The food supervision module is used to issue rectification instructions to the kitchen according to the judgment results of the analysis and judgment module, name each material placement position, and record the name of the food corresponding to each material placement position.

[0011] A foreground target extraction module, wherein the foreground target extraction module is used to extract a foreground target from the kitchen monitoring video by background subtraction based on an initial background template;

[0012] The classification model includes a staff classification model for judging whether the staff's clothing is compliant, a kitchen waste classification model for judging whether there is garbage in the kitchen, and a live animal classification model for judging whether there are live animals in the kitchen; the staff classification model is used to judge whether the staff's clothing, masks and hats are compliant; the kitchen waste classification model is used to judge whether there is garbage on the kitchen floor and tabletop; the live animal classification model is used to judge

[0013] The food storage box includes a horizontal plate, a material placement position, a gravity scale, a camera and a food storage box. The inner wall of the food storage box is provided with a plurality of horizontal plates, and the surface of the horizontal plate is provided with a plurality of material placement positions. The inner wall of the horizontal plate is provided with a gravity scale on the material placement position, and the side wall of the horizontal plate is provided with a reset switch and a prompt light. A reset switch, a prompt light, a gravity scale and a material placement position form a group. A food storage box and a food storage bag are placed on the material placement position. The upper opening of the food storage box is covered with plastic wrap, and a camera is provided on the top of the food storage box.

[0014] A data-driven remote food safety supervision method, the method comprising:

[0015] Step S100: The model training module obtains training samples and constructs a classification model based on a convolutional neural network, and is used to optimize parameters of the constructed classification model using the collected training samples as input, output the trained classification model, and classify the data collected on the field end through the classification model;

[0016] Step S200: The image acquisition module obtains a video image of the door of the food storage box, and then obtains images of food entering and exiting the food storage box each time, thereby obtaining a space monitoring video of the kitchen;

[0017] Step S300: the analysis and judgment module analyzes the acquired weight value, retention time and the video image at the box door to determine whether the food in the food storage box meets the food processing standards;

[0018] Step S400: extracting an initial background template from the kitchen's spatial monitoring video by using a target detection algorithm, extracting a foreground target from the kitchen's spatial monitoring video by background subtraction based on the initial background template; and discriminating the foreground target by using a trained classification model;

[0019] Step S500, the food supervision module is used to issue corresponding rectification instructions to the kitchen according to the judgment results of the analysis and judgment module. The corresponding prompt light on the material placement position that needs to be rectified lights up red, each material placement position is named, and the name of the food corresponding to each material placement position is recorded.

[0020] The step S200 includes:

[0021] Step S201: The employee puts one of the ingredients into the food storage box, covers the upper opening of the food storage box with a layer of plastic wrap, opens the box door, and presses the reset switch;

[0022] Step S202: Place the food storage box through the opened door and into the corresponding material placement position. The image acquisition module obtains the video image of the door of the food storage box through the camera, captures several images of the food storage box passing through the opened door, and selects the image with the highest definition.

[0023] The step S300 includes:

[0024] Step S301: the analysis and judgment module selects the outer edge of the cling film on the image, and then applies a colored translucent layer, marks the outer edge line of the upper opening of the food storage box with a line as a detection circle, and fills the part of the detection circle that is not coated with the translucent layer with a red shadow; if the red shadow area is greater than 0, it is determined that the sealing of the upper opening of the food storage box meets the standard; if the red shadow area is less than 0, it is determined that the sealing of the upper opening of the food storage box does not meet the standard;

[0025] Step S302: After the food storage box is placed at the corresponding material placement position, the gravity scale corresponding to the material placement position weighs the food storage box containing the food, and records the weight value and placement time point of the weighing as the initial weight value and initial placement time point respectively;

[0026] Step S303: Take the food storage box placed in step S202 and put it back into the food storage box again, take it out and put it back n times, repeat the process of step S202 each time, and record the weight value and placement time of each weighing.

[0027] The step S303 further includes:

[0028] Calculate the time interval H1 between the last placement time point and the initial weight value. If the time interval H1 is greater than the set preservation time, it is determined that the food quality does not meet the standard; if the time interval H1 is less than the set preservation time, it is determined that the food quality meets the standard;

[0029] Calculate the time interval H2 between the last placement time point and the last weight value obtained. If the interval H2 is greater than the set food storage time, it is determined that the food has been out of the fresh-keeping environment for a long time and should be stopped from use. If the interval H2 is less than the set food storage time, it is determined that the food is reasonably fresh and should continue to be used.

[0030] The weight value of the last measurement is compared with the weight value of the last measurement. If the weight value of the last measurement is greater than the weight value of the last measurement, it is determined that foreign objects are mixed into the food; if the weight value of the last measurement is less than the weight value of the last measurement, it is determined that no foreign objects are mixed into the food.

[0031] The step S400 includes:

[0032] The analysis and judgment module uses the staff classification model to determine whether the staff's attire is in compliance with regulations, including whether the staff's clothing, masks and hats are in compliance with regulations. It uses the kitchen waste classification model to determine whether there is waste in the kitchen, including whether there is waste on the kitchen floor and table. It uses the live animal classification model to determine whether there are live animals in the kitchen.

[0033] The step S500 includes:

[0034] If the judgment results of the analysis and judgment module include that the seal of the opening on the food storage box does not meet the standards, the quality of the food does not meet the standards, the food has been out of the fresh-keeping environment for a long time and should be stopped from use, the staff's attire is not in compliance with regulations, there is garbage in the kitchen, there are live animals in the kitchen, and foreign objects are mixed in the food, then a corresponding rectification instruction will be issued, and the corresponding prompt light on the material placement position that needs rectification will light up red; if the judgment results of the analysis and judgment module include that the seal of the opening on the food storage box meets the standards, the quality of the food meets the standards, the food preservation is reasonable and should continue to be used, the staff's attire is in compliance with regulations, there is no garbage in the kitchen, there are no live animals in the kitchen, and no foreign objects are mixed in the food, then no rectification instruction will be issued.

[0035] The food storage box includes a horizontal plate, a material placement position, a gravity scale, a camera and a food storage box. The inner wall of the food storage box is provided with a plurality of horizontal plates, and the surface of the horizontal plate is provided with a plurality of material placement positions. The inner wall of the horizontal plate is provided with a gravity scale on the material placement position, and the side wall of the horizontal plate is provided with a reset switch and a prompt light. A reset switch, a prompt light, a gravity scale and a material placement position form a group. A food storage box and a food storage bag are placed on the material placement position. The upper opening of the food storage box is covered with plastic wrap, and a camera is provided on the top of the food storage box.

[0036] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0037] The present invention can detect the upper opening of the food storage box that is put into the food storage box each time, screen out the food storage boxes whose upper openings are not completely covered by the plastic wrap, quickly mark the food storage boxes on the corresponding material placement positions, and directly remind employees on site that the food ingredients cannot be used and need to be discarded. The remote monitoring end can remind employees in real time that the food ingredients cannot be used and need to be discarded, and can replace manual supervision of the food ingredients in the food storage box. The "out-only but not in" management of the food ingredients in a food storage box is realized, and the dual-channel management of the food ingredients by the on-site end and the supervision end is achieved, which improves the supervision intensity and precision of the remote management of food during the food processing process, and greatly improves food safety during the food processing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0039] Figure 1 is a plan view of the food storage box of the present invention;

[0040] Figure 2 is a flow chart of the supervision method of the present invention;

[0041] In the figure: 1. storage box; 2. horizontal board; 3. material placement; 4. gravity scale; 5. camera; 6. food storage box. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] See also Figure 1-Figure 2 The present invention provides a technical solution: a data-driven food safety remote monitoring system, the system comprising:

[0044] The food storage box 1 is used to place food materials for food processing, seal and preserve the food materials, obtain the weight value and placement time of the food materials in real time, and use the time when the reset switch is pressed as the starting point of the accumulation of the retention time of the food materials. Each time the food materials are placed on the material placement position 3, the measurement value of the gravity scale 4 is the weight value;

[0045] The model training module is used to obtain training samples and build a classification model based on a convolutional neural network. It is also used to optimize the parameters of the built classification model using the collected training samples as input, output the trained classification model, and classify the data collected on-site through the classification model.

[0046] The image acquisition module is used to obtain the video image of the door of the food storage box 1, and then obtain the image of each time when the food enters and exits the food storage box 1, and also obtain the space monitoring video of the kitchen;

[0047] The analysis and judgment module is used to analyze the acquired weight value, retention time and the video image at the box door to determine whether the food in the food storage box 1 meets the food processing standards

[0048] The food supervision module is used to issue rectification instructions to the kitchen according to the judgment results of the analysis and judgment module, name each material placement position 3, and record the name of the food corresponding to each material placement position 3.

[0049] The food storage box 1 includes a horizontal plate 2, a material placement position 3, a gravity scale 4, a camera 5 and a food storage box 6. The inner wall of the food storage box 1 is provided with a plurality of horizontal plates 2, and the surface of the horizontal plate 2 is provided with a plurality of material placement positions 3. The inner wall of the horizontal plate 2 is provided with a gravity scale 4 on the material placement position 3, and the side wall of the horizontal plate 2 is provided with a reset switch and a prompt light. A reset switch, a prompt light, a gravity scale 4 and a material placement position 3 form a group. A food storage box 6 and a food storage bag are placed on the material placement position 3. The upper opening of the food storage box 6 is covered with a plastic wrap, and a camera 5 is provided on the top of the food storage box 1.

[0050] The system also includes:

[0051] A foreground target extraction module, wherein the foreground target extraction module is used to extract a foreground target from the kitchen monitoring video by background subtraction based on an initial background template;

[0052] The classification model includes a staff classification model for determining whether the staff's clothing is in compliance with regulations, a kitchen garbage classification model for determining whether there is garbage in the kitchen, and a live animal classification model for determining whether there are live animals in the kitchen; the staff classification model is used to determine whether the staff's clothing, masks and hats are in compliance with regulations; the kitchen garbage classification model is used to determine whether there is garbage on the kitchen floor and tabletop; and the live animal classification model is used for judgment.

[0053] A data-driven remote food safety supervision method, the method comprising:

[0054] Step S100: The model training module obtains training samples and constructs a classification model based on a convolutional neural network, and is used to optimize parameters of the constructed classification model using the collected training samples as input, output the trained classification model, and classify the data collected on the field end through the classification model;

[0055] Step S200: the image acquisition module obtains a video image of the door of the food storage box 1, and then obtains an image of each time when food enters and exits the food storage box 1, and obtains a space monitoring video of the kitchen;

[0056] Step S300: the analysis and judgment module analyzes the acquired weight value, retention time and the video image at the box door to determine whether the food in the food storage box 1 meets the food processing standard;

[0057] Step S400: extracting an initial background template from the kitchen's spatial monitoring video by using a target detection algorithm, extracting a foreground target from the kitchen's spatial monitoring video by background subtraction based on the initial background template; and discriminating the foreground target by using a trained classification model;

[0058] Step S500, the food supervision module is used to issue corresponding rectification instructions to the kitchen according to the judgment result of the analysis and judgment module. The corresponding prompt light on the material placement position 3 that needs to be rectified lights up red, each material placement position 3 is named, and the name of the food corresponding to each material placement position 3 is recorded.

[0059] The step S200 includes:

[0060] Step S201: The employee puts one of the ingredients into the food storage box 6, covers the upper opening of the food storage box 6 with a layer of plastic wrap, opens the box door, and presses the reset switch;

[0061] Step S202: Place the food storage box 6 through the opened door and into the corresponding material placement position 3. The image acquisition module obtains the video image of the door of the food storage box 1 through the camera 5, captures several images of the food storage box 6 passing through the opened door, and selects the image with the highest definition.

[0062] The step S300 includes:

[0063] Step S301: The analysis and judgment module selects the outer edge of the plastic wrap on the image, then applies a colored translucent layer, marks the outer edge line of the opening on the food storage box 6 with a line as a detection circle, and fills the part of the detection circle that is not coated with the translucent layer with a red shadow; if the red shadow area is greater than 0, it is determined that the sealing of the upper opening of the food storage box 6 meets the standard; if the red shadow area is less than 0, it is determined that the sealing of the upper opening of the food storage box 6 does not meet the standard; the upper opening of the food storage box 6 placed in the food storage box 1 can be detected each time, and the food storage box 6 whose upper opening is not completely covered by the plastic wrap can be screened out, and the food storage box 6 on the corresponding material placement position 3 can be quickly marked, and employees can be directly reminded on site that the food cannot be used and needs to be discarded. The remote monitoring end can remind employees in real time that the food cannot be used and needs to be discarded, and the food in the food storage box 1 can be supervised instead of manual labor.

[0064] Step S302: After the food storage box 6 is placed at the corresponding material placement position 3, the gravity scale 4 corresponding to the material placement position 3 weighs the food storage box 6 containing the food, and records the weight value and the placement time point of the weighing as the initial weight value and the initial placement time point respectively;

[0065] Step S303: Take the food storage box 6 placed in step S202 and put it back into the food storage box 1 again, take it out and put it back n times, repeat the process of step S202 each time, and record the weight value and placement time point of each weighing.

[0066] The step S303 further includes:

[0067] Calculate the time interval H1 between the last placement time point and the initial weight value. If the time interval H1 is greater than the set preservation time, it is determined that the food quality does not meet the standard; if the time interval H1 is less than the set preservation time, it is determined that the food quality meets the standard;

[0068] Calculate the time interval H2 between the last placement time point and the last weight value obtained. If the interval H2 is greater than the set food storage time, it is determined that the food has been out of the fresh-keeping environment for a long time and should be stopped from use. If the interval H2 is less than the set food storage time, it is determined that the food is reasonably fresh and should continue to be used.

[0069] Compare the weight value of the last measurement with the weight value of the last measurement. If the weight value of the last measurement is greater than the weight value of the last measurement, it is determined that foreign objects are mixed into the food; if the weight value of the last measurement is less than the weight value of the last measurement, it is determined that no foreign objects are mixed into the food. The "out-only but not in" management of food in a food storage box 1 is realized, and dual-channel management of food at the on-site end and the supervision end is achieved, which improves the supervision intensity and precision of remote food management during food processing, and greatly improves food safety during food processing.

[0070] The step S400 includes:

[0071] The analysis and judgment module uses the staff classification model to determine whether the staff's attire is in compliance with regulations, including whether the staff's clothing, masks and hats are in compliance with regulations. It uses the kitchen waste classification model to determine whether there is waste in the kitchen, including whether there is waste on the kitchen floor and table. It uses the live animal classification model to determine whether there are live animals in the kitchen.

[0072] The step S500 includes:

[0073] If the judgment results of the analysis and judgment module include that the seal of the opening on the food storage box 6 does not meet the standards, the quality of the food does not meet the standards, the food has been out of the fresh-keeping environment for a long time and should be stopped from use, the staff's attire is not in compliance with regulations, there is garbage in the kitchen, there are live animals in the kitchen, and foreign objects are mixed in the food, then a corresponding rectification instruction is issued, and the corresponding prompt light on the storage position 3 that needs rectification lights up red; if the judgment results of the analysis and judgment module include that the seal of the opening on the food storage box 6 meets the standards, the quality of the food meets the standards, the food preservation is reasonable and should continue to be used, the staff's attire is in compliance with regulations, there is no garbage in the kitchen, there are no live animals in the kitchen, and no foreign objects are mixed in the food, then no rectification instruction is issued.

[0074] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data-driven remote food safety supervision method, characterized in that: The method includes: Step S100: The model training module obtains training samples and constructs a classification model based on a convolutional neural network, and is used to optimize parameters of the constructed classification model using the collected training samples as input, output the trained classification model, and classify the data collected on the field end through the classification model; Step S200, the image acquisition module obtains a video image of the door of the food storage box (1), and then obtains an image of each time when food enters and exits the food storage box (1), thereby obtaining a space monitoring video of the kitchen; Step S300: the analysis and judgment module analyzes the acquired weight value, retention time and the video image at the box door to determine whether the food in the food storage box (1) meets the food processing standards; Step S400: extracting an initial background template from the kitchen's spatial monitoring video using a target detection algorithm, extracting a foreground target from the kitchen's spatial monitoring video using background subtraction based on the initial background template; and discriminating the foreground target using a trained classification model; Step S500, the food supervision module is used to issue corresponding rectification instructions to the kitchen according to the judgment result of the analysis and judgment module, the corresponding prompt light on the material placement position (3) that needs to be rectified lights up red, each material placement position (3) is named, and the name of the food corresponding to each material placement position (3) is recorded; The step S200 includes: Step S201: the employee puts one of the ingredients into the food storage box (6), covers the upper opening of the food storage box (6) with a layer of plastic wrap, opens the box door, and presses the reset switch; Step S202: the food storage box (6) is placed on the corresponding material placement position (3) after passing through the opened door, the image acquisition module obtains the video image of the door of the food storage box (1) through the camera (5), captures a number of images of the food storage box (6) passing through the opened door, and selects the image with the highest definition; The step S300 includes: Step S301: the analysis and judgment module selects the outer edge of the cling film on the image, then applies a colored translucent layer, marks the outer edge line of the upper opening of the food storage box (6) with a line as a detection circle, and fills the portion of the detection circle that is not coated with the translucent layer with a red shadow; if the red shadow area is greater than 0, it is determined that the sealing of the upper opening of the food storage box (6) meets the standard; if the red shadow area is less than 0, it is determined that the sealing of the upper opening of the food storage box (6) does not meet the standard; Step S302: After the food storage box (6) is placed at the corresponding material placement position (3), the gravity scale (4) corresponding to the material placement position (3) weighs the food storage box (6) containing the food, and records the weight value and the placement time point of the weighing as the initial weight value and the initial placement time point respectively; Step S303: taking the food storage box (6) placed in step S202 and putting it back into the food storage box (1), taking it out and putting it back n times, repeating the process of step S202 each time, and recording the weight value and placement time of each weighing; The step S303 further includes: Calculate the time interval H1 between the last placement time point and the initial weight value. If the time interval H1 is greater than the set preservation time, it is determined that the food quality does not meet the standard; if the time interval H1 is less than the set preservation time, it is determined that the food quality meets the standard; Calculate the time interval H2 between the last placement time point and the last weight value obtained. If the interval H2 is greater than the set food storage time, it is determined that the food has been out of the fresh-keeping environment for a long time and should be stopped from use. If the interval H2 is less than the set food storage time, it is determined that the food is reasonably fresh and should continue to be used. The weight value of the last measurement is compared with the weight value of the last measurement. If the weight value of the last measurement is greater than the weight value of the last measurement, it is determined that foreign objects are mixed into the food; if the weight value of the last measurement is less than the weight value of the last measurement, it is determined that no foreign objects are mixed into the food.

2. A data-driven remote food safety supervision method according to claim 1, characterized in that: The step S400 includes: The analysis and judgment module uses the staff classification model to determine whether the staff's attire is in compliance with regulations, including whether the staff's clothing, masks and hats are in compliance with regulations. It uses the kitchen waste classification model to determine whether there is waste in the kitchen, including whether there is waste on the kitchen floor and table. It uses the live animal classification model to determine whether there are live animals in the kitchen.

3. A data-driven remote food safety supervision method according to claim 2, characterized in that: The step S500 includes: If the judgment result of the analysis and judgment module includes that the seal of the upper opening of the food storage box (6) does not meet the standard, the quality of the food does not meet the standard, the food has been out of the fresh-keeping environment for a period of time and should be stopped from use, the staff's attire is not in compliance with regulations, there is garbage in the kitchen, there are live animals in the kitchen, and foreign objects are mixed in the food, then a corresponding rectification instruction is issued, and the corresponding prompt light on the storage position (3) that needs rectification lights up red; if the judgment result of the analysis and judgment module includes that the seal of the upper opening of the food storage box (6) meets the standard, the quality of the food meets the standard, the food preservation is reasonable and should continue to be used, the staff's attire is in compliance with regulations, there is no garbage in the kitchen, there are no live animals in the kitchen, and foreign objects are not mixed in the food, then no rectification instruction is issued.

4. A data-driven remote food safety supervision method according to claim 3, characterized in that: The food storage box (1) comprises a transverse plate (2), a material placement position (3), a gravity scale (4), a camera (5) and a food storage box (6). The inner wall of the food storage box (1) is provided with a plurality of transverse plates (2), a surface of the transverse plate (2) is provided with a plurality of material placement positions (3), a gravity scale (4) is placed on the inner wall of the transverse plate (2) on the material placement positions (3), a reset switch and a prompt light are provided on the side wall of the transverse plate (2), a reset switch, a prompt light, a gravity scale (4) and a material placement position (3) form a group, a food storage box (6) and a food storage bag are placed on the material placement position (3), an upper opening of the food storage box (6) is covered with a plastic wrap, and a camera (5) is provided on the top of the food storage box (1).

5. A data-driven food safety remote monitoring system, applied to a data-driven food safety remote monitoring method according to any one of claims 1 to 4, characterized in that: The system includes: A food storage box (1) is used to store food materials for food processing, seal and preserve the food materials, obtain the weight value and placement time of the food materials in real time, and use the time when the reset switch is pressed as the starting point for the accumulation of the retention time of the food materials. Each time the food materials are placed on the placement position (3), the measurement value of the gravity scale (4) is the weight value; The model training module is used to obtain training samples and build a classification model based on a convolutional neural network. It is also used to optimize the parameters of the built classification model using the collected training samples as input, output the trained classification model, and classify the data collected on-site through the classification model. A picture acquisition module, used to obtain a video picture of the door of the food storage box (1), and then obtain a picture of each time food enters and leaves the food storage box (1), and also obtain a space monitoring video of the kitchen; An analysis and judgment module is used to analyze the acquired weight value, retention time and video footage at the box door to determine whether the food in the food storage box (1) meets the food processing standards. The food material supervision module is used to issue rectification instructions to the kitchen according to the judgment result of the analysis and judgment module, name each material placement position (3), and record the name of the food material corresponding to each material placement position (3).

6. A data-driven food safety remote monitoring system according to claim 5, characterized in that: The food storage box (1) comprises a transverse plate (2), a material placement position (3), a gravity scale (4), a camera (5) and a food storage box (6). The inner wall of the food storage box (1) is provided with a plurality of transverse plates (2), a surface of the transverse plate (2) is provided with a plurality of material placement positions (3), a gravity scale (4) is placed on the inner wall of the transverse plate (2) on the material placement positions (3), a reset switch and a prompt light are provided on the side wall of the transverse plate (2), a reset switch, a prompt light, a gravity scale (4) and a material placement position (3) form a group, a food storage box (6) and a food storage bag are placed on the material placement position (3), an upper opening of the food storage box (6) is covered with a plastic wrap, and a camera (5) is provided on the top of the food storage box (1).

7. A data-driven food safety remote monitoring system according to claim 6, characterized in that: The system also includes: A foreground target extraction module, wherein the foreground target extraction module is used to extract a foreground target from the kitchen monitoring video by background subtraction based on an initial background template; The classification model includes a staff classification model for determining whether the staff's clothing is in compliance with regulations, a kitchen garbage classification model for determining whether there is garbage in the kitchen, and a live animal classification model for determining whether there are live animals in the kitchen; the staff classification model is used to determine whether the staff's clothing, masks and hats are in compliance with regulations; the kitchen garbage classification model is used to determine whether there is garbage on the kitchen floor and tabletop; and the live animal classification model is used for judgment.

Images