Method and device for remotely and automatically judging whether laser removal can be carried out on nevus on face or not

Abstract

The invention relates to the technical field of removing face nevi through laser, in particular to a method and device for remotely and automatically judging whether laser removal can be conducted on the face nevi or not, and the method comprises the steps that S1, a face image remotely input by a user is obtained; s2, using a trained nevus detection model to detect position information and size information of each nevus on the face image, the nevus detection model being trained by using a model automatic optimization mode; s3, judging whether each nevus can be removed by laser or not according to the position information and the size information of each nevus; and S4, returning a judgment result to the user. According to the invention, whether laser removal can be carried out on the nevus on the face can be remotely and automatically judged.

Description

Technical Field

[0001] The present invention relates to the technical field of laser removal of facial moles, and in particular to a method and device for remotely and automatically judging whether a facial mole can be removed by laser. Background Art

[0002] Many hospitals that can perform laser removal of facial moles are located in big cities, but users who want to have laser removal of facial moles are distributed in various places, and many are in small towns far away from big cities. If these hospitals can automatically determine whether facial moles can be removed by laser through facial images input remotely by users, it will greatly save these users the time and energy of going to big cities. Therefore, there is an urgent need for a method to remotely and automatically determine whether facial moles can be removed by laser. Summary of the invention

[0003] The present invention provides a method and device for remotely and automatically determining whether a facial mole can be removed by laser, so as to solve the problems existing in the above-mentioned prior art. The technical solution is as follows:

[0004] On the one hand, a method for remotely and automatically determining whether a facial mole can be removed by laser is provided, comprising:

[0005] S1. Obtaining a face image input remotely by a user;

[0006] S2. Using the trained facial mole detection model, detecting the position information and size information of each facial mole on the face image, wherein the facial mole detection model is trained using a model automatic optimization method;

[0007] S3, judging whether each facial mole can be removed by laser according to the position information and size information of each facial mole;

[0008] S4. Return the judgment result to the user.

[0009] Optionally, the facial mole detection model is trained using a model automatic optimization method, specifically including:

[0010] S21, obtaining historical manually marked data, analyzing the historical manually marked data based on object features, and determining data collection type information;

[0011] S22, determining a data search range according to the data collection type information, and collecting data within the data search range to obtain original learning data;

[0012] S23, dividing and extracting the original learning data to form a training set and a test set;

[0013] S24, automatically learning the facial mole detection model according to the training set to form an optimized model;

[0014] S25. Perform a target test on the optimization model according to the test set, and control the automatic learning process of the model according to the result of the target test.

[0015] Optionally, the S21 specifically includes:

[0016] S211, obtaining historical manually labeled data, analyzing and extracting the same objects, and determining that the automatic learning object is face image data;

[0017] S212, based on the automatic learning object, extracting the relevant range of the historical manually labeled data to form an object feature whose automatic learning object range is facial mole;

[0018] S213, extracting object features from historical manually labeled data according to the automatic learning object range, and forming automatic learning object feature information as location information and size information of facial moles;

[0019] S214: Combine the automatic learning object, the automatic learning object range, and the automatic learning object feature information to form data collection type information.

[0020] Optionally, the S22 specifically includes:

[0021] S221, defining a scope according to the automatic learning object to determine a data search scope;

[0022] S222, collecting data according to the automatic learning object range within the data search range to obtain initial collected data;

[0023] S223: Analyze the collection amount of the initial collected data based on the characteristic information of the automatic learning object to determine the original learning data.

[0024] Optionally, the S223 specifically includes:

[0025] S2231, determining the feature proportions of all objects in the initial collected data based on the automatically learned object feature information, and judging the sufficiency of the collected data according to the feature proportions to obtain a sufficiency judgment result;

[0026] S2232, determining the proportion of difference features of objects having automatic learning object feature information in the initial collected data, and judging the richness of the collected data according to the proportion of difference features to obtain a richness judgment result;

[0027] S2233. Process the initial collected data according to the sufficiency judgment result and the richness judgment result to form original learning data.

[0028] Optionally, the S2231 specifically includes:

[0029] Setting the adequacy threshold , obtain the number of objects with automatically learned object feature information in the initial collected data , make the following adequacy judgment:

[0030] If exists , it is judged that the initial collected data is sufficient;

[0031] If exists , it is determined that the initial collected data is insufficient;

[0032] in, Indicates the total number of objects in the initial collection data;

[0033] The S2232 specifically includes:

[0034] Setting the richness threshold , obtain the number of objects with different automatic learning object feature information in the initial collection data , make the following richness judgment:

[0035] If exists , it is judged that the initial collected data is rich;

[0036] If exists , it is judged that the initial collected data is not rich;

[0037] in, Represents the number of objects with automatically learned object feature information in the initial collected data.

[0038] Optionally, the S2233 specifically includes:

[0039] Judging the sufficiency judgment result and the richness judgment result;

[0040] If the sufficiency is judged as sufficient and the richness is judged as rich, the currently collected initial collection data is determined as the original learning data;

[0041] If the sufficiency is judged as sufficient and the richness is judged as not rich, then objects with different automatic learning object feature information are obtained to expand the initial collected data until the richness is judged as rich, and the expanded initial collected data is determined as the original learning data;

[0042] If the sufficiency is judged as insufficient and the richness is judged as rich, objects without automatic learning object characteristic information in the initial collected data are screened out until the sufficiency is judged as sufficient, and the screened out initial collected data are determined as original learning data;

[0043] If the sufficiency is judged as insufficient and the richness is judged as not rich, data collection is performed according to the scope of the automatic learning object to expand the initial collected data until the sufficiency is judged as sufficient and the richness is judged as rich, and the expanded initial collected data is determined as the original learning data.

[0044] Optionally, the S23 specifically includes:

[0045] Set the training collection threshold , performing object-based random collection on the original learning data to obtain an object set; wherein the number of objects in the object set is not less than the training collection amount threshold ;

[0046] forming a training set according to the object set;

[0047] The data other than the object set in the original learning data is calibrated based on the automatic learning object feature information to form a test set.

[0048] Optionally, the S25 specifically includes:

[0049] Get the number of objects in the test set that have automatically learned object feature information ;

[0050] Obtaining the number of objects having automatic learning object feature information determined by analyzing the test set by the optimization model ;

[0051] Setting target learning threshold , and perform the following judgment processing:

[0052] If exists , then continue to obtain the original learning data to train the optimization model until it meets ;

[0053] If exists , then the optimization model is determined as the trained facial mole detection model.

[0054] On the other hand, a device for remotely and automatically determining whether a facial mole can be removed by laser is provided, the device comprising:

[0055] An acquisition module is used to acquire a face image input remotely by a user;

[0056] A detection module, used to detect the position information and size information of each facial mole on the face image using a trained facial mole detection model, wherein the facial mole detection model is trained using a model automatic optimization method;

[0057] A judgment module, used for judging whether each facial mole can be removed by laser according to the position information and size information of each facial mole;

[0058] The return module is used to return the judgment result to the user.

[0059] On the other hand, an electronic device is provided, comprising a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the above-mentioned method of remotely and automatically determining whether a facial mole can be removed by laser.

[0060] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-mentioned method of remotely and automatically determining whether a facial mole can be removed by laser.

[0061] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0062] The present invention uses a trained facial mole detection model to detect the position information and size information of each facial mole on a face image input remotely by a user, and judges whether each facial mole can be removed by laser (for example, facial moles on the forehead, cheeks, and chin within a certain size range can be removed, facial moles around the eyes cannot be removed, and facial moles outside a certain size range cannot be removed), and returns the judgment result to the user. The user decides whether to go to a big city where a hospital is located based on the judgment result. If the facial mole cannot be removed, the user does not need to go to a big city, thereby greatly saving the user's time and energy in going to the big city where the hospital is located. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0064] Figure 1 This is a flow chart of a method for remotely and automatically determining whether a facial mole can be removed by laser, provided by an embodiment of the present invention;

[0065] Figure 2 This is a block diagram of a device for remotely and automatically determining whether a facial mole can be removed by laser, provided by an embodiment of the present invention;

[0066] Figure 3 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0068] The embodiment of the present invention provides a method for remotely and automatically determining whether a facial mole can be removed by laser. The method can be implemented by an electronic device, which can be a terminal or a server. Figure 1 The flowchart of a method for remotely and automatically determining whether a facial mole can be removed by laser is shown. The processing flow of the method may include the following steps:

[0069] S1. Obtaining a face image input remotely by a user;

[0070] S2. Using the trained facial mole detection model, detecting the position information and size information of each facial mole on the face image, wherein the facial mole detection model is trained using a model automatic optimization method;

[0071] S3, judging whether each facial mole can be removed by laser according to the position information and size information of each facial mole;

[0072] S4. Return the judgment result to the user.

[0073] Optionally, the facial mole detection model is trained using a model automatic optimization method, specifically including:

[0074] S21, obtaining historical manually marked data, analyzing the historical manually marked data based on object features, and determining data collection type information;

[0075] Optionally, the S21 specifically includes:

[0076] S211, obtaining historical manually labeled data, analyzing and extracting the same objects, and determining that the automatic learning object is face image data;

[0077] S212, based on the automatic learning object, extracting the relevant range of the historical manually labeled data to form an object feature whose automatic learning object range is facial mole;

[0078] S213, extracting object features from historical manually labeled data according to the automatic learning object range, and forming automatic learning object feature information as location information and size information of facial moles;

[0079] S214: Combine the automatic learning object, the automatic learning object range, and the automatic learning object feature information to form data collection type information.

[0080] S22, determining a data search range according to the data collection type information, and collecting data within the data search range to obtain original learning data;

[0081] Optionally, the S22 specifically includes:

[0082] S221, defining a scope according to the automatic learning object to determine a data search scope;

[0083] S222, collecting data according to the automatic learning object range within the data search range to obtain initial collected data;

[0084] S223: Analyze the collection amount of the initial collected data based on the characteristic information of the automatic learning object to determine the original learning data.

[0085] Optionally, the S223 specifically includes:

[0086] S2231, determining the feature proportions of all objects in the initial collected data based on the automatically learned object feature information, and judging the sufficiency of the collected data according to the feature proportions to obtain a sufficiency judgment result;

[0087] Optionally, the S2231 specifically includes:

[0088] Setting the adequacy threshold , obtain the number of objects with automatically learned object feature information in the initial collected data , make the following adequacy judgment:

[0089] If exists , it is judged that the initial collected data is sufficient;

[0090] If exists , it is determined that the initial collected data is insufficient;

[0091] in, Indicates the total number of objects in the initial collection data;

[0092] The initial collected data of the embodiment of the present invention is the facial image data with facial moles, the number of which is M, wherein the number of objects with automatically learned object feature information (facial mole location information and size information) is N, and the N / M and the preset sufficiency judgment threshold are used to calculate the sufficiency of the facial mole. Compare and get the adequacy judgment result.

[0093] S2232, determining the proportion of difference features of objects having automatic learning object feature information in the initial collected data, and judging the richness of the collected data according to the proportion of difference features to obtain a richness judgment result;

[0094] Optionally, the S2232 specifically includes:

[0095] Setting the richness threshold , obtain the number of objects with different automatic learning object feature information in the initial collection data , make the following richness judgment:

[0096] If exists , it is judged that the initial collected data is rich;

[0097] If exists , it is judged that the initial collected data is not rich;

[0098] in, Represents the number of objects with automatically learned object feature information in the initial collected data.

[0099] The number of objects with different automatic learning object feature information (different position information and size information of facial moles, such as different positions of forehead, cheek, chin, eye area, etc., facial moles in the same position are taken as one object, such as several facial moles on the forehead are taken as one object, and those of the same size are taken as one object) in the initial collected data of the embodiment of the present invention is U, and the number of objects with automatic learning object feature information (position information and size information of facial moles) is N. By comparing U / N with a preset richness judgment threshold Compare and obtain the richness judgment result.

[0100] S2233. Process the initial collected data according to the sufficiency judgment result and the richness judgment result to form original learning data.

[0101] Optionally, the S2233 specifically includes:

[0102] Judging the sufficiency judgment result and the richness judgment result;

[0103] If the sufficiency is judged as sufficient and the richness is judged as rich, the currently collected initial collection data is determined as the original learning data;

[0104] If the sufficiency is judged as sufficient and the richness is judged as not rich, then objects with different automatic learning object feature information (different position information and size information of facial moles) are obtained to expand the initial collected data until the richness is judged as rich, and the expanded initial collected data is determined as the original learning data;

[0105] If the sufficiency is judged as insufficient and the richness is judged as rich, the objects (location information and size information of facial moles) that do not have the characteristic information of the automatic learning object in the initial collected data are screened out until the sufficiency is judged as sufficient, and the initial collected data after the screening is determined as the original learning data;

[0106] If the sufficiency is judged as insufficient and the richness is judged as not rich, data collection is performed according to the scope of the automatic learning object to expand the initial collected data until the sufficiency is judged as sufficient and the richness is judged as rich, and the expanded initial collected data is determined as the original learning data.

[0107] S23, dividing and extracting the original learning data to form a training set and a test set;

[0108] Optionally, the S23 specifically includes:

[0109] Set the training collection threshold , performing object-based random collection on the original learning data to obtain an object set; wherein the number of objects in the object set is not less than the training collection amount threshold ;

[0110] forming a training set according to the object set;

[0111] The data other than the object set in the original learning data is calibrated based on the automatic learning object feature information to form a test set.

[0112] S24, automatically learning the facial mole detection model according to the training set to form an optimized model;

[0113] S25. Perform a target test on the optimization model according to the test set, and control the automatic learning process of the model according to the result of the target test.

[0114] Optionally, the S25 specifically includes:

[0115] Get the number of objects in the test set that have automatically learned object feature information ;

[0116] Obtaining the number of objects having automatic learning object feature information determined by analyzing the test set by the optimization model ;

[0117] Setting target learning threshold , and perform the following judgment processing:

[0118] If exists , then continue to obtain the original learning data to train the optimization model until it meets ;

[0119] If exists , then the optimization model is determined as the trained facial mole detection model.

[0120] In the embodiment of the present invention, in the process of training the facial mole detection model, the model automatic optimization method is used for training, and the direction of the model automatic optimization is determined based on the historical manually marked data, and the object data to be collected for the optimization direction is determined, so as to ensure that the learning data collected subsequently has training feature information that can obviously achieve model optimization. At the same time, when obtaining the original learning data, reasonable data collection, analysis and judgment are performed based on the feature information to form basic training data and test data that can effectively achieve model optimization. In addition, after each optimization of the model, the processing effect of the optimized model is judged by the proposed final optimization target, and then the automated optimization process is obtained in combination with the judgment result, so as to fully and automatically achieve reasonable optimization of the model efficiently.

[0121] like Figure 2 As shown, an embodiment of the present invention provides a device for remotely and automatically determining whether a facial mole can be removed by laser, the device comprising:

[0122] An acquisition module 210 is used to acquire a face image inputted remotely by a user;

[0123] A detection module 220, configured to detect the position information and size information of each facial mole on the face image using a trained facial mole detection model, wherein the facial mole detection model is trained using a model automatic optimization method;

[0124] A determination module 230, for determining whether each facial mole can be removed by laser according to the location information and size information of each facial mole;

[0125] The return module 240 is used to return the judgment result to the user.

[0126] An embodiment of the present invention provides a device for remotely and automatically determining whether a facial mole can be removed by laser. Its functional structure corresponds to a method for remotely and automatically determining whether a facial mole can be removed by laser provided by an embodiment of the present invention, and will not be repeated here.

[0127] Figure 3 It is a structural diagram of an electronic device 300 provided in an embodiment of the present invention. The electronic device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 301 and one or more memories 302, wherein the memory 302 stores at least one instruction, and the at least one instruction is loaded and executed by the processor 301 to implement the steps of the above-mentioned method for remotely and automatically determining whether a facial mole can be removed by laser.

[0128] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions, which can be executed by a processor in a terminal to complete the above-mentioned method of remotely and automatically determining whether a facial mole can be removed by laser. For example, the computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0129] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for remotely and automatically determining whether facial moles can be removed by laser. It is characterized in that The method comprises: S1. Obtaining a face image input remotely by a user; S2. Using the trained facial mole detection model, detecting the position information and size information of each facial mole on the face image, wherein the facial mole detection model is trained using a model automatic optimization method; S3, judging whether each facial mole can be removed by laser according to the position information and size information of each facial mole; S4. Return the judgment result to the user.

2. The method according to claim 1, It is characterized in that The facial mole detection model is trained using a model automatic optimization method, specifically including: S21, obtaining historical manually marked data, analyzing the historical manually marked data based on object features, and determining data collection type information; S22, determining a data search range according to the data collection type information, and collecting data within the data search range to obtain original learning data; S23, dividing and extracting the original learning data to form a training set and a test set; S24, automatically learning the facial mole detection model according to the training set to form an optimized model; S25. Perform a target test on the optimization model according to the test set, and control the automatic learning process of the model according to the result of the target test.

3. The method according to claim 2, It is characterized in that The S21 specifically includes: S211, obtaining historical manually labeled data, analyzing and extracting the same objects, and determining that the automatic learning object is face image data; S212, based on the automatic learning object, extracting the relevant range of the historical manually labeled data to form an object feature whose automatic learning object range is facial mole; S213, extracting object features from historical manually labeled data according to the automatic learning object range, and forming automatic learning object feature information as location information and size information of facial moles; S214: Combine the automatic learning object, the automatic learning object range, and the automatic learning object feature information to form data collection type information.

4. The method according to claim 3, It is characterized in that The S22 specifically includes: S221, defining a scope according to the automatic learning object to determine a data search scope; S222, collecting data according to the automatic learning object range within the data search range to obtain initial collected data; S223: Analyze the collection amount of the initial collected data based on the characteristic information of the automatic learning object to determine the original learning data.

5. The method according to claim 4, It is characterized in that The S223 specifically includes: S2231, determining the feature proportions of all objects in the initial collected data based on the automatically learned object feature information, and judging the sufficiency of the collected data according to the feature proportions to obtain a sufficiency judgment result; S2232, determining the proportion of difference features of objects having automatic learning object feature information in the initial collected data, and judging the richness of the collected data according to the proportion of difference features to obtain a richness judgment result; S2233. Process the initial collected data according to the sufficiency judgment result and the richness judgment result to form original learning data.

6. The method according to claim 5, It is characterized in that The S2231 specifically includes: Setting the adequacy threshold , obtain the number of objects with automatically learned object feature information in the initial collected data , make the following adequacy judgment: If exists , it is judged that the initial collected data is sufficient; If exists , it is determined that the initial collected data is insufficient; in, Indicates the total number of objects in the initial collection data; The S2232 specifically includes: Setting the richness threshold , obtain the number of objects with different automatic learning object feature information in the initial collection data , make the following richness judgment: If exists , it is judged that the initial collected data is rich; If exists , it is judged that the initial collected data is not rich; in, Represents the number of objects with automatically learned object feature information in the initial collected data.

7. The method according to claim 5, It is characterized in that The S2233 specifically includes: Judging the sufficiency judgment result and the richness judgment result; If the sufficiency is judged as sufficient and the richness is judged as rich, the currently collected initial collection data is determined as the original learning data; If the sufficiency is judged as sufficient and the richness is judged as not rich, then objects with different automatic learning object feature information are obtained to expand the initial collected data until the richness is judged as rich, and the expanded initial collected data is determined as the original learning data; If the sufficiency is judged as insufficient and the richness is judged as rich, objects without automatic learning object characteristic information in the initial collected data are screened out until the sufficiency is judged as sufficient, and the screened out initial collected data are determined as original learning data; If the sufficiency is judged as insufficient and the richness is judged as not rich, data collection is performed according to the scope of the automatic learning object to expand the initial collected data until the sufficiency is judged as sufficient and the richness is judged as rich, and the expanded initial collected data is determined as the original learning data.

8. The method according to claim 2, It is characterized in that The S23 specifically includes: Set the training collection threshold , performing object-based random collection on the original learning data to obtain an object set; wherein the number of objects in the object set is not less than the training collection amount threshold ; forming a training set according to the object set; The data other than the object set in the original learning data is calibrated based on the automatic learning object feature information to form a test set.

9. The method according to claim 2, It is characterized in that The S25 specifically includes: Get the number of objects in the test set that have automatically learned object feature information ; Obtaining the number of objects having automatic learning object feature information determined by analyzing the test set by the optimization model ; Setting target learning threshold , and perform the following judgment processing: If exists , then continue to obtain the original learning data to train the optimization model until it meets ; If exists , then the optimization model is determined as the trained facial mole detection model.

10. A device for remotely and automatically determining whether facial moles can be removed by laser. It is characterized in that The device comprises: An acquisition module, used to acquire a face image input remotely by a user; A detection module, used to detect the position information and size information of each facial mole on the face image using a trained facial mole detection model, wherein the facial mole detection model is trained using a model automatic optimization method; A judgment module, used for judging whether each facial mole can be removed by laser according to the position information and size information of each facial mole; The return module is used to return the judgment result to the user.

Images