Model processing method and device, and job correction method and device

Abstract

The application provides a model processing method and device and a homework correction method and device, and relates to the technical field of information processing. The model processing method provided by the application first acquires at least one new symbol sample image corresponding to a new symbol in response to a new symbol identification request for a symbol identification model. Then, the symbol identification model is updated based on the at least one new symbol sample image, so that the updated symbol identification model can identify the new symbol. The model processing method provided by the embodiment of the application updates the symbol identification model, so that the updated symbol identification model can identify the new symbol, thereby achieving the purpose of increasing the symbols that can be identified by the symbol identification model according to the actual needs of the user in the application stage of the symbol identification model, meeting the personalized needs of the user and improving the user experience.

Description

Technical Field

[0001] This application relates to the field of information processing technology, specifically to a model processing method and apparatus, and a job correction method and apparatus. Background Technology

[0002] With the rapid development of information technology, homework machines that can recognize teachers' correction marks are increasingly being incorporated into teaching support systems.

[0003] However, current homework assignment machines can only recognize fixed categories of correction symbols and cannot recognize teachers' custom correction symbols, thus failing to meet teachers' personalized correction needs. Therefore, how to enable homework assignment machines to recognize teachers' custom correction symbols has become an urgent problem to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a model processing method and apparatus, and a job correction method and apparatus.

[0005] In a first aspect, one embodiment of this application provides a model processing method, which includes: in response to a request for adding a new symbol to a symbol recognition model, acquiring at least one sample image of the new symbol corresponding to the new symbol; and updating the symbol recognition model based on the at least one sample image of the new symbol so that the updated symbol recognition model can recognize the new symbol.

[0006] In conjunction with the first aspect, in some implementations of the first aspect, updating the symbol recognition model based on at least one newly added symbol sample image includes: extracting features from at least one newly added symbol sample image to obtain symbol features corresponding to each of the at least one newly added symbol sample image; determining the Gaussian mean corresponding to the newly added symbol based on the symbol features corresponding to each of the at least one newly added symbol sample image; and updating the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol.

[0007] In conjunction with the first aspect, in some implementations of the first aspect, the symbol recognition model is updated based on the Gaussian mean corresponding to the newly added symbol, including: determining the character variance corresponding to the newly added symbol based on the Gaussian mean corresponding to the newly added symbol and the symbol features corresponding to at least one sample image of the newly added symbol; determining the Gaussian variance corresponding to the newly added symbol based on the character variance corresponding to the newly added symbol; and updating the symbol recognition model based on the Gaussian mean and Gaussian variance corresponding to the newly added symbol.

[0008] In conjunction with the first aspect, in some implementations of the first aspect, determining the Gaussian variance corresponding to the newly added symbol based on the character variance corresponding to the newly added symbol includes: determining auxiliary variance calculation data corresponding to the newly added symbol, wherein the auxiliary variance calculation data includes at least one of global variance, local variance, and identity matrix corresponding to the newly added symbol, and the auxiliary variance calculation data is used to combine with the character variance corresponding to the newly added symbol to determine the Gaussian variance corresponding to the newly added symbol; and determining the mean of the auxiliary variance calculation data and the character variance corresponding to the newly added symbol as the Gaussian variance corresponding to the newly added symbol.

[0009] In conjunction with the first aspect, in some implementations of the first aspect, the method for determining the global variance includes: determining the symbol features of each of the at least one original symbol that the symbol recognition model can recognize; and determining the global variance based on the symbol features corresponding to each of at least one newly added symbol sample image and the symbol features of each of the at least one original symbol.

[0010] In conjunction with the first aspect, in some implementations of the first aspect, the method for determining the local variance includes: determining the top N symbol features that are closest to the Gaussian mean corresponding to the newly added symbol, the symbol features corresponding to at least one sample image of the newly added symbol, and the symbol features of at least one original symbol; and determining the local variance based on the top N symbol features.

[0011] In conjunction with the first aspect, in some implementations of the first aspect, the symbol recognition model is updated based on the Gaussian mean corresponding to the newly added symbol, including: determining the identity matrix corresponding to the Gaussian mean corresponding to the newly added symbol; and updating the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol and the identity matrix corresponding to the Gaussian mean.

[0012] Secondly, one embodiment of this application provides a method for grading assignments, which includes: acquiring a target assignment image with grading symbol traces; processing the target assignment image using a symbol recognition model to obtain a symbol recognition result corresponding to the target assignment image, wherein the symbol recognition model is obtained by the model processing method mentioned in any of the above embodiments.

[0013] Thirdly, one embodiment of this application provides a model processing apparatus, which includes: an acquisition module, configured to acquire at least one sample image of a newly added symbol corresponding to the newly added symbol in response to a request for adding a new symbol to a symbol recognition model; and an update module, configured to update the symbol recognition model based on the at least one sample image of the newly added symbol, so that the updated symbol recognition model can recognize the newly added symbol.

[0014] Fourthly, one embodiment of this application provides a job correction device, which includes: an acquisition module for acquiring a target job image with correction symbol traces; and a processing module for processing the target job image using a symbol recognition model to obtain a symbol recognition result corresponding to the target job image, wherein the symbol recognition model is obtained by processing the model processing method mentioned in any of the above embodiments.

[0015] Fifthly, one embodiment of this application provides a computer-readable storage medium storing a computer program for performing the methods mentioned in any of the above embodiments.

[0016] In a sixth aspect, one embodiment of this application provides an electronic device, the electronic device comprising: a processor; a memory for storing processor-executable instructions; the processor being configured to perform the methods mentioned in any of the foregoing embodiments.

[0017] The model processing method provided in this application first responds to a request for adding a new symbol to the symbol recognition model by acquiring at least one sample image of the new symbol; then, based on the at least one sample image of the new symbol, it updates the symbol recognition model so that the updated symbol recognition model can recognize the new symbol. The model processing method provided in this application update the symbol recognition model so that the updated symbol recognition model can recognize the new symbol, thereby achieving the goal of increasing the number of symbols that the symbol recognition model can recognize according to the actual needs of the user during the application stage of the symbol recognition model, meeting the user's personalized needs, and improving the user experience. Attached Figure Description

[0018] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0019] Figure 1 The diagram shown is a structural schematic of a work machine in an implementation scenario provided by an embodiment of this application.

[0020] Figure 2 The diagram shown is a flowchart illustrating a model processing method provided in an exemplary embodiment of this application.

[0021] Figure 3 The diagram shown is a schematic diagram of the newly added symbols provided in an exemplary embodiment of this application.

[0022] Figure 4The diagram shown is a schematic representation of an exemplary embodiment of this application, illustrating the process of updating a symbol recognition model based on at least one newly added symbol sample image.

[0023] Figure 5 The diagram shown is a schematic representation of an exemplary embodiment of this application, illustrating the process of updating a symbol recognition model based on the Gaussian mean corresponding to a newly added symbol.

[0024] Figure 6 The diagram shown is a flowchart illustrating how a newly added symbol is determined based on the character variance corresponding to the newly added symbol, according to an exemplary embodiment of this application.

[0025] Figure 7 The diagram shown is a flowchart illustrating the method for determining global variance provided in an exemplary embodiment of this application.

[0026] Figure 8 The diagram shown is a flowchart illustrating the method for determining local variance provided in an exemplary embodiment of this application.

[0027] Figure 9 The diagram shown is a flowchart illustrating the process of updating the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol, as provided in another exemplary embodiment of this application.

[0028] Figure 10 The diagram shown is a flowchart illustrating a job grading method provided in an exemplary embodiment of this application.

[0029] Figure 11 The diagram shown is a schematic representation of the structure of a model processing apparatus provided in an exemplary embodiment of this application.

[0030] Figure 12 The diagram shown is a structural schematic of a job correction device provided in an exemplary embodiment of this application.

[0031] Figure 13 The diagram shown is a structural schematic of an electronic device provided in an exemplary embodiment of this application. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0033] In their daily studies, students typically complete numerous exercises and tests. These paper materials, accumulated over time, eventually become a valuable review resource. Students can review previously completed exercises to identify and address their weaknesses, especially in areas where they are weak. However, a major problem students face in managing traditional paper learning materials is the sheer volume of accumulated material, making targeted searching, retrieval, and review extremely difficult. While some common methods can improve efficiency, such as categorizing and sorting materials by subject and time, and some students maintain a dedicated error notebook to copy down incorrect answers, many others resort to a tedious, exhaustive search through a sea of ​​homework and tests. Furthermore, maintaining an error notebook typically requires significant additional effort, placing a considerable burden on students.

[0034] For teachers, there are also shortcomings in making full use of resources such as students' daily homework and exam papers. Because the content of paper materials is difficult to analyze automatically, teachers struggle to obtain quantitative information about students' learning quality. Relying solely on their memory, teachers find it difficult to accurately understand each student's strengths and weaknesses, let alone grasp their learning progress at each stage.

[0035] The emergence of homework machines has solved the problem of automating the statistical analysis of paper materials. Specifically, for a set of learning materials that have already been answered by students and graded by teachers, the homework machine can photograph the materials and scan them to obtain the questions, student answers, and teacher grading results (i.e., grading symbols) for each question. In one possible implementation, the accuracy rate of each question can be calculated based on the scan results, and personalized teaching plans can be generated for teachers based on the statistical results. For example, questions with lower accuracy rates may require focused explanation, while questions with higher accuracy rates may not need explanation. Alternatively, for the same student, statistical results for different questions can be generated. These results can indicate the student's mastery of various knowledge points, and personalized practice questions can be recommended based on these results to help the student better master the knowledge points.

[0036] However, current homework machines can only recognize fixed categories of marking symbols (such as checkmarks and crosses) and cannot recognize teachers' custom marking symbols, thus failing to meet teachers' personalized marking needs. In one possible marking scenario, a teacher might find a student's answer particularly excellent and want to use it as a model for other students to learn from. They might want to mark the answer with a triangle in the vicinity as a marking result, but current learning machines cannot recognize custom marking symbols like triangles, failing to meet users' personalized needs. Therefore, enabling homework machines to recognize teachers' custom marking symbols has become an urgent problem to be solved.

[0037] Figure 1 The diagram shown is a structural schematic of a work machine in an implementation scenario provided by an embodiment of this application. Figure 1 As shown, the workstation 10 provided in this embodiment includes: a motherboard 11, a camera assembly 12, a bracket 13, a base 14, and a server 15. The camera assembly 12 can be built into a recess in the motherboard 11 for capturing learning materials; one end of the bracket 13 is connected to the motherboard 11 to support it; the other end of the bracket 13 is connected to the base 14, and the projection of the base 14 onto a plane parallel to the motherboard 11 is smaller than the projection of the motherboard 11 onto the same plane.

[0038] In practical applications, teachers can place the assignments to be graded within the camera's field of view (area 12) while the assignment is being graded, and the machine simultaneously captures and recognizes the grading marks. Alternatively, teachers can grade the assignments first, then place the graded assignments within the camera's field of view (area 12) for capture and recognition of the grading marks.

[0039] It should be noted that server 15 is used to update the symbol recognition model so that the updated symbol recognition model can recognize newly added symbols. Server 15 can be configured on the main unit of the work machine, or it can be set up independently of the work machine and connected to the work machine's signal.

[0040] Figure 2 The diagram shown is a flowchart illustrating a model processing method provided in an exemplary embodiment of this application. Figure 2 As shown in the embodiments of this application, the model processing method includes the following steps.

[0041] Step S201: In response to the request for adding a new symbol for the symbol recognition model, obtain at least one sample image of the new symbol corresponding to the new symbol.

[0042] In one embodiment, when a teacher needs to add a new marking symbol (such as a triangle), they can write down at least one sample image of the new symbol and issue a request to add the new symbol by operating the marking machine.

[0043] Figure 3 The diagram shown is a schematic representation of a newly added symbol provided in an exemplary embodiment of this application. For example... Figure 3 As shown, 10 sample images of the newly added symbols are presented.

[0044] Step S202: Update the symbol recognition model based on at least one newly added symbol sample image.

[0045] In practical applications, the method first responds to a request for a new symbol to be recognized by the symbol recognition model by acquiring at least one sample image of the new symbol. Then, based on this sample image, the symbol recognition model is updated so that it can recognize the new symbol. The model processing method provided in this application can recognize newly added custom symbols, thereby meeting users' personalized correction needs and improving user experience.

[0046] Figure 4 The diagram illustrates a process for updating a symbol recognition model based on at least one newly added symbol sample image, as provided in an exemplary embodiment of this application. Figure 2 This application extends from the embodiments shown. Figure 4 The illustrated embodiment will be described in detail below. Figure 4 The illustrated embodiments and Figure 2 The differences between the embodiments shown are not repeated here, and the similarities are not repeated here.

[0047] like Figure 4 As shown, in the model processing method provided in this application embodiment, the step of updating the symbol recognition model based on at least one newly added symbol sample image includes the following steps.

[0048] Step S401: Extract features from at least one newly added symbol sample image to obtain the symbol features corresponding to each of the at least one newly added symbol sample image.

[0049] In one embodiment, a feature extraction module extracts features from at least one newly added symbol sample image to obtain the symbol features corresponding to each of the at least one newly added symbol sample image. The symbol features mentioned in this application are illustrated below using a three-dimensional row vector as an example.

[0050] In one embodiment, with Figure 3 The following explanation will be based on three images out of the ten newly added symbol sample images. The symbol features corresponding to the three newly added symbol (such as triangle) sample images are as follows: a = (a1, a2, a3), b = (b1, b2, b3), c = (c1, c2, c3).

[0051] Step S402: Based on the symbol features corresponding to at least one newly added symbol sample image, determine the Gaussian mean corresponding to the newly added symbol;

[0052] In one embodiment, the Gaussian mean corresponding to the newly added symbol refers to the average value of the symbol features corresponding to at least one newly added symbol sample image. Specifically, the first dimension of the Gaussian mean m is the average value of the first dimension data a1 of a, the first dimension data b1 of b, and the first dimension data c1 of c; the second and third dimensions of the Gaussian mean are obtained based on the same calculation method.

[0053] The formula is expressed as follows:

[0054]

[0055] Step S403: Update the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol.

[0056] In one embodiment, the Gaussian mean m corresponding to the newly added symbol is the recognition information corresponding to the newly added symbol. By storing the Gaussian mean corresponding to the newly added symbol, the symbol recognition model is updated, so that the updated symbol recognition model can recognize the newly added symbol, and can obtain the symbol recognition result of the symbol to be recognized and corrected based on the Gaussian mean of the symbol to be recognized and corrected.

[0057] In practical applications, feature extraction is first performed on at least one newly added symbol sample image to obtain the symbol features corresponding to each of the at least one newly added symbol sample image. Then, based on the symbol features corresponding to each of the at least one newly added symbol sample image, the Gaussian mean corresponding to the newly added symbol is determined. Finally, based on the Gaussian mean corresponding to the newly added symbol, the symbol recognition model is updated. The model processing method provided in this application embodiment enables the updated symbol recognition model to recognize newly added symbols, and can obtain the symbol recognition result of the symbol to be recognized and corrected based on the Gaussian mean of the symbol to be recognized and corrected.

[0058] Figure 5 The diagram shown is a schematic representation of an exemplary embodiment of this application, illustrating the process of updating a symbol recognition model based on the Gaussian mean corresponding to a newly added symbol. In this application... Figure 4 This application extends from the embodiments shown. Figure 5 The illustrated embodiment will be described in detail below. Figure 5 The illustrated embodiments and Figure 4 The differences between the embodiments shown are not repeated here, and the similarities are not repeated here.

[0059] like Figure 5 As shown, in the model processing method provided in this application embodiment, the step of updating the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol includes the following steps.

[0060] Step S501: Determine the character variance corresponding to the newly added symbol based on the Gaussian mean corresponding to the newly added symbol and the symbol features corresponding to at least one sample image of the newly added symbol.

[0061] In one embodiment, reference Figure 4 The example in this embodiment further illustrates the character variance corresponding to the newly added symbol. The character variance ∑1 corresponding to the newly added symbol is a three-row, three-column matrix, and its specific formula is as follows:

[0062]

[0063] in,

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072] Step S502: Determine the Gaussian variance corresponding to the newly added symbol based on the character variance corresponding to the newly added symbol.

[0073] In one embodiment, the Gaussian variance includes the character variance, which can be considered as one representation of the Gaussian variance.

[0074] Step S503: Update the symbol recognition model based on the Gaussian mean and Gaussian variance corresponding to the newly added symbol.

[0075] In one embodiment, the Gaussian mean m and character variance ∑1 corresponding to the newly added symbol are the recognition information corresponding to the newly added symbol. By storing the Gaussian mean and character variance corresponding to the newly added symbol, the symbol recognition model is updated, enabling the updated symbol recognition model to recognize the newly added symbol. Furthermore, based on the Gaussian mean and character variance of the symbol to be recognized and corrected, the symbol recognition result of the symbol to be recognized and corrected can be obtained, thus improving the recognition accuracy.

[0076] In practical applications, the character variance of the newly added symbol is first determined based on the Gaussian mean and symbol features of at least one sample image of the newly added symbol. Then, the Gaussian variance of the newly added symbol is determined based on the character variance. Finally, the symbol recognition model is updated based on the Gaussian mean and Gaussian variance of the newly added symbol. The model processing method provided in this application enables the updated symbol recognition model to recognize newly added symbols and obtains the symbol recognition result of the symbol to be recognized and corrected based on the Gaussian mean and character variance of the symbol to be recognized and corrected, thus improving recognition accuracy.

[0077] In reality, teachers can provide dozens or even hundreds of new symbol sample images. However, in order to improve the recognition accuracy of the symbol recognition model, dozens or even hundreds of samples are obviously far from enough. Therefore, the model processing method provided in this application effectively overcomes the defect of insufficient sample quantity through the following embodiments, thereby improving the recognition accuracy of the symbol recognition model.

[0078] Figure 6 The diagram shown is a schematic representation of an exemplary embodiment of this application, illustrating the process of determining the Gaussian variance of a newly added symbol based on the character variance corresponding to that symbol. In this application... Figure 5 This application extends from the embodiments shown. Figure 6 The illustrated embodiment will be described in detail below. Figure 6 The illustrated embodiments and Figure 5 The differences between the embodiments shown are not repeated here, and the similarities are not repeated here.

[0079] like Figure 6 As shown, in the model processing method provided in this application embodiment, the step of determining the Gaussian variance corresponding to the newly added symbol based on the character variance corresponding to the newly added symbol includes the following steps.

[0080] Step S601: Determine the variance auxiliary calculation data corresponding to the newly added symbol.

[0081] In one embodiment, the variance-aided calculation data includes at least one of the following: global variance, local variance, and identity matrix corresponding to the newly added symbol. The variance-aided calculation data is used to combine the character variance corresponding to the newly added symbol to determine the Gaussian variance corresponding to the newly added symbol. Specifically, the global variance measures the dispersion of the symbol features of all symbols that the symbol recognition model can recognize; the local variance measures the dispersion of the top N symbol features that are closest to the Gaussian mean corresponding to the newly added symbol; and the identity matrix I1 corresponding to the newly added symbol has the same dimension as the Gaussian mean corresponding to the newly added symbol, where N is a positive integer.

[0082] For example, with Figure 4 Taking the Gaussian mean m as an example in the illustrated embodiment, the dimension of the identity matrix I1 corresponding to the newly added symbol is the same as the dimension of the Gaussian mean corresponding to the newly added symbol. The identity matrix I1 corresponding to the newly added symbol can be expressed by the following formula:

[0083]

[0084] Step S602: The mean of the variance auxiliary calculation data and the character variance corresponding to the newly added symbol is determined as the Gaussian variance corresponding to the newly added symbol.

[0085] In one embodiment, the mean of at least one variance-aided calculation data and the character variance corresponding to the newly added symbol can be regarded as different representations of Gaussian variance.

[0086] In practical applications, the variance-aided calculation data corresponding to the newly added symbol is first determined. Then, the mean of the variance-aided calculation data and the character variance corresponding to the newly added symbol is determined as the Gaussian variance corresponding to the newly added symbol. The model processing method provided in this application introduces variance-aided calculation data, which participates in the calculation of the Gaussian variance corresponding to the new symbol, thereby making up for the deficiency of insufficient sample size and improving the recognition accuracy of the symbol recognition model.

[0087] Figure 7 The diagram shown is a flowchart illustrating the method for determining global variance provided in an exemplary embodiment of this application. Figure 6 This application extends from the embodiments shown. Figure 7 The illustrated embodiment will be described in detail below. Figure 7 The illustrated embodiments and Figure 6 The differences between the embodiments shown are not repeated here, and the similarities are not repeated here.

[0088] like Figure 7 As shown, the global variance determination method in the model processing method provided in this application embodiment includes the following steps.

[0089] Step S701: Determine the symbol features of at least one original symbol that the symbol recognition model can recognize.

[0090] In one embodiment, the original symbols refer to the correction symbols that the current symbol recognition model can already recognize, such as checkmarks and crosses. The following explanation uses the symbol features corresponding to the three sample images of checkmarks and the three sample images of crosses as examples.

[0091] The symbol features of the three sample images with checkmarks are as follows: d = (d1, d2, d3), e = (e1, e2, e3), f = (f1, f2, f3). The symbol features of the three sample images with crosses are as follows: g = (g1, g2, g3), h = (h1, h2, h3), i = (i1, i2, i3).

[0092] Step S702: Determine the global variance based on the symbol features corresponding to at least one newly added symbol sample image and the symbol features of at least one existing symbol.

[0093] In one embodiment, the mean p of the symbol features corresponding to at least one newly added symbol sample image and the symbol features of at least one existing symbol is first determined. Then, based on the mean p, the symbol features corresponding to at least one newly added symbol sample image and the symbol features of at least one existing symbol, the global variance ∑2 is determined.

[0094] The formula for the mean p is as follows:

[0095] p = (p1, p2, p3),

[0096] in,

[0097]

[0098]

[0099] The determination of the global variance ∑2 can be referenced from the determination of the character variance ∑1 in formula (2). The difference is that the global variance ∑2 is a nine-row, nine-column matrix, which will not be listed here. The following unified determination method of Gaussian variance ∑ is used as an example for explanation. Gaussian variance ∑ is an n-row, n-column matrix, where n can be any positive integer.

[0100]

[0101] Where E represents the mean, E(X) represents the mean of sample X, T represents the transpose of the original matrix, X1 represents the first dimension of each sample, ..., X n Let μn represent the nth dimension of the data for each sample, μ1 represent the first dimension of the sample mean, ..., μn represent the nth dimension of the data for each sample. n The nth dimension represents the sample mean.

[0102] In practical applications, the symbol features of at least one original symbol that the symbol recognition model can recognize are first determined. Then, based on the symbol features corresponding to at least one newly added symbol sample image and the symbol features of at least one original symbol, the global variance is determined. This embodiment obtains the global variance, which affects the determination of the Gaussian variance corresponding to the newly added symbol from the perspective of all symbols that the symbol recognition model can recognize. This compensates for the significant deficiency in sample size and improves the recognition accuracy of the symbol recognition model.

[0103] Figure 8 The diagram shown is a flowchart illustrating a method for determining local variance provided in an exemplary embodiment of this application. Figure 6 This application extends from the embodiments shown. Figure 8 The illustrated embodiment will be described in detail below. Figure 8 The illustrated embodiments and Figure 6The differences between the embodiments shown are not repeated here, and the similarities are not repeated here.

[0104] like Figure 8 As shown, the method for determining local variance in the model processing method provided in this application embodiment includes the following steps.

[0105] Step S801: Based on the Gaussian mean corresponding to the newly added symbol, the symbol features corresponding to at least one sample image of the newly added symbol, and the symbol features of at least one original symbol, determine the top N symbol features that are closest to the Gaussian mean corresponding to the newly added symbol.

[0106] In one embodiment, the Euclidean distances between the Gaussian mean corresponding to the newly added symbol and the symbol features corresponding to at least one newly added symbol sample image, as well as the symbol features of at least one existing symbol, are determined. Specifically, the Euclidean distances between the Gaussian mean m and a, b, c, d, e, f, g, h, and i are determined, and the top N symbol features closest to the Gaussian mean m are identified. N can be selected according to actual needs, and this embodiment does not further limit this selection.

[0107] Step S802: Determine the local variance based on the first N symbol features.

[0108] In one embodiment, firstly, the mean of the first N symbol features is determined; then, the local variance is determined based on the first N symbol features and their corresponding mean. The method for determining the mean can refer to... Figure 4 The method for determining the Gaussian mean m and the method for determining the local variance in the illustrated embodiment can be found in [reference needed]. Figure 5 The methods for determining the character variance ∑1 in the illustrated embodiments will not be listed one by one here.

[0109] In practical applications, the first step is to determine the top N symbol features that are closest to the Gaussian mean of the newly added symbol, the symbol features of at least one sample image of the newly added symbol, and the symbol features of at least one existing symbol. Then, based on these top N symbol features, the local variance is determined. This embodiment of the application obtains the local variance, influencing the determination of the Gaussian variance of the newly added symbol from the perspective of the local symbol features closest to the Gaussian mean of the newly added symbol. This compensates for the significant deficiency in sample size and further improves the recognition accuracy of the symbol recognition model.

[0110] Figure 9 The diagram shown is a schematic representation of a process for updating a symbol recognition model based on the Gaussian mean corresponding to a newly added symbol, provided in another exemplary embodiment of this application. Figure 4 This application extends from the embodiments shown. Figure 9The illustrated embodiment will be described in detail below. Figure 9 The illustrated embodiments and Figure 4 The differences between the embodiments shown are not repeated here, and the similarities are not repeated here.

[0111] like Figure 9 As shown, in the model processing method provided in this application embodiment, the step of updating the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol includes the following steps.

[0112] Step S901: Determine the identity matrix corresponding to the Gaussian mean of the newly added symbol.

[0113] In one embodiment, with Figure 4 Taking the Gaussian mean m in the illustrated embodiment as an example, the identity matrix I2 corresponding to the Gaussian mean of the newly added symbol can be expressed by the following formula:

[0114]

[0115] Step S902: Update the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol and the identity matrix corresponding to the Gaussian mean.

[0116] In one embodiment, the Gaussian mean *m* corresponding to the newly added symbol and the identity matrix *I2* corresponding to *m* are the recognition information corresponding to the newly added symbol. By storing the Gaussian mean and the identity matrix corresponding to the newly added symbol, the symbol recognition model is updated, enabling the updated symbol recognition model to recognize the newly added symbol. Furthermore, based on the Gaussian mean and the identity matrix corresponding to the symbol to be recognized and corrected, the symbol recognition result of the symbol to be recognized and corrected can be obtained.

[0117] In practical applications, the identity matrix corresponding to the Gaussian mean of the newly added symbol is first determined. Then, based on the Gaussian mean of the newly added symbol and the identity matrix corresponding to the Gaussian mean, the symbol recognition model is updated. The model processing method provided in this application embodiment enables the updated symbol recognition model to recognize the newly added symbol, and can obtain the symbol recognition result of the symbol to be recognized and corrected based on the Gaussian mean of the symbol to be recognized and corrected and the identity matrix corresponding to the Gaussian mean. The model processing method provided in this application embodiment simplifies the update process of the symbol recognition model, requiring only the determination of the Gaussian mean corresponding to the newly added symbol, thus improving the calculation speed of the symbol recognition model.

[0118] Figure 10 The diagram shown is a flowchart illustrating a job grading method provided in an exemplary embodiment of this application. Figure 10 As shown in the embodiments of this application, the job correction method includes the following steps.

[0119] Step S1001: Obtain the target job image with correction mark traces.

[0120] In one embodiment, when grading students' assignments, teachers typically leave grading marks on the assignments. For example, a correct answer is marked with a checkmark in the corresponding answer area, and an incorrect answer is marked with an X. Thus, obtaining an image of the target assignment with these grading marks allows teachers or students to clearly see the grading results.

[0121] In one embodiment, the target job image can be a pre-stored electronic image of the job to be corrected in the target area, or it can be an image obtained by taking real-time photos of the job to be corrected in the target area.

[0122] When photographing the work to be corrected in the target area, the shooting area corresponding to the camera component 12 can be set as the target area, or the camera component 12 can be aimed at the target area after the target area is determined. In this way, the work to be corrected can be placed in the target area, and then the target area can be photographed to obtain the image of the target work.

[0123] Step S1002: Use the symbol recognition model to process the target operation image and obtain the symbol recognition result corresponding to the target operation image.

[0124] In one embodiment, using a symbol recognition model, the symbol features x of the symbol to be identified and corrected are first determined; then, based on the symbol features x of the symbol to be identified and corrected, the likelihood s between the symbol to be identified and corrected and each symbol that the symbol recognition model can recognize is determined; then, the symbol with the highest likelihood with the symbol to be identified and corrected is determined as the symbol recognition result corresponding to the symbol to be identified and corrected.

[0125] In one embodiment, based on the symbolic feature x of the symbol to be identified and corrected, the likelihood s between the symbol to be identified and all symbols that the symbol recognition model can recognize is determined. The specific formula is as follows:

[0126]

[0127] Where n represents the dimension of the symbol feature x of the markup symbol to be identified.

[0128] It should be noted that, using the above formula (7), the likelihood s between the symbol to be identified and each symbol that the symbol recognition model can recognize is determined based on the symbol feature x of the symbol to be identified and the Gaussian mean m and Gaussian variance ∑ of each symbol to be identified and corrected, respectively. The symbol recognition model is obtained by processing the model using the model processing method mentioned in any of the above embodiments.

[0129] Figure 11The diagram shown is a schematic representation of the structure of a model processing apparatus provided in an exemplary embodiment of this application. Figure 11 As shown, the model processing apparatus provided in this application embodiment includes:

[0130] The acquisition module 201 is used to acquire at least one sample image of the newly added symbol corresponding to the newly added symbol in response to the request for adding a new symbol for the symbol recognition model;

[0131] The update module 202 is used to update the symbol recognition model based on at least one newly added symbol sample image, so that the updated symbol recognition model can recognize the newly added symbol.

[0132] In one embodiment of this application, the update module 202 is further configured to extract features from at least one newly added symbol sample image to obtain symbol features corresponding to each of the at least one newly added symbol sample image; determine the Gaussian mean corresponding to the newly added symbol based on the symbol features corresponding to each of the at least one newly added symbol sample image; and update the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol.

[0133] In one embodiment of this application, the update module 202 is further configured to determine the character variance corresponding to the newly added symbol based on the Gaussian mean corresponding to the newly added symbol and the symbol features corresponding to at least one sample image of the newly added symbol; determine the Gaussian variance corresponding to the newly added symbol based on the character variance corresponding to the newly added symbol; and update the symbol recognition model based on the Gaussian mean and Gaussian variance corresponding to the newly added symbol.

[0134] In one embodiment of this application, the update module 202 is further configured to determine at least one variance auxiliary calculation data corresponding to the newly added symbol. The variance auxiliary calculation data is one of the global variance, local variance, and the identity matrix corresponding to the newly added symbol. The global variance is used to measure the dispersion of the symbol features of all symbols that the symbol recognition model can recognize. The local variance is used to measure the dispersion of the top N symbol features that are closest to the Gaussian mean corresponding to the newly added symbol. The dimension of the identity matrix corresponding to the newly added symbol is the same as the dimension of the Gaussian mean corresponding to the newly added symbol, and N is a positive integer. The mean of at least one variance auxiliary calculation data and the character variance corresponding to the newly added symbol is determined as the Gaussian variance corresponding to the newly added symbol.

[0135] In one embodiment of this application, the update module 202 is further configured to determine the symbol features of at least one original symbol that the symbol recognition model can recognize; and to determine the global variance based on the symbol features corresponding to at least one newly added symbol sample image and the symbol features of at least one original symbol.

[0136] In one embodiment of this application, the update module 202 is further configured to determine the top N symbol features that are closest to the Gaussian mean corresponding to the newly added symbol, the symbol features corresponding to at least one sample image of the newly added symbol, and the symbol features of at least one original symbol; and determine the local variance based on the top N symbol features.

[0137] In one embodiment of this application, the update module 202 is further configured to determine the identity matrix corresponding to the Gaussian mean of the newly added symbol; and update the symbol recognition model based on the Gaussian mean of the newly added symbol and the identity matrix corresponding to the Gaussian mean.

[0138] Figure 12 The diagram shown is a structural schematic of a job correction device provided in an exemplary embodiment of this application. Figure 12 As shown, the job correction device provided in this application embodiment includes:

[0139] The acquisition module 1001 is used to acquire target job images with correction mark traces;

[0140] The processing module 1002 is used to process the target operation image using a symbol recognition model to obtain the symbol recognition result corresponding to the target operation image, wherein the symbol recognition model is obtained by the model processing method mentioned in any of the above embodiments.

[0141] It should be understood that Figure 11 The provided model processing device includes an acquisition module 201, an update module 202, and... Figure 12 The operation and functions of the acquisition module 1001 and processing module 1002 in the provided work correction device can be referred to the above. Figures 1 to 10 The methods provided will not be repeated here to avoid duplication.

[0142] Below, for reference Figure 13 This describes an electronic device according to embodiments of the present application. Figure 13 The diagram shown is a structural schematic of an electronic device provided in an exemplary embodiment of this application.

[0143] like Figure 13 As shown, the electronic device 30 includes one or more processors 301 and memory 302.

[0144] The processor 301 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 30 to perform desired functions.

[0145] The memory 302 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 301 may execute the program instructions to implement the methods of the various embodiments of this application described above and / or other desired functions. Various contents, such as newly added symbol sample images, Gaussian mean, and Gaussian variance, may also be stored in the computer-readable storage medium.

[0146] In one example, the electronic device 30 may also include an input device 303 and an output device 304, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0147] The input device 303 may include, for example, a keyboard, a mouse, etc.

[0148] The output device 304 can output various information to the outside, including Gaussian mean and Gaussian variance. The output device 304 may include, for example, a display, a communication network, and remote output devices connected thereto.

[0149] Of course, for the sake of simplicity, Figure 13 Only some of the components of the electronic device 30 relevant to this application are shown in this illustration; components such as buses, input / output interfaces, etc., are omitted. In addition, the electronic device 30 may include any other suitable components depending on the specific application.

[0150] In addition to the methods and apparatus described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps of the methods described above according to various embodiments of this application.

[0151] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0152] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions that, when executed by a processor, cause the processor to perform the steps of the methods described above according to various embodiments of this application.

[0153] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0154] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0155] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0156] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0157] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0158] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A model processing method, characterized in that, include: In response to a request to add a new symbol for the symbol recognition model, obtain at least one sample image of the new symbol corresponding to the new symbol; Feature extraction is performed on each of the at least one newly added symbol sample image to obtain the symbol features corresponding to each of the at least one newly added symbol sample image; Based on the symbol features corresponding to each of the at least one newly added symbol sample image, determine the Gaussian mean corresponding to the newly added symbol; Based on the Gaussian mean corresponding to the newly added symbol and the symbol features corresponding to each of the at least one newly added symbol sample image, the character variance corresponding to the newly added symbol is determined. The character variance is determined by calculating the mean of the product between matrix A and the transpose of A. Matrix A is the difference matrix between the symbol feature vector of the newly added symbol sample image and the mean vector of the symbol feature vector of the newly added symbol sample image. Based on the character variance corresponding to the newly added symbol, determine the Gaussian variance corresponding to the newly added symbol; Based on the Gaussian mean and Gaussian variance corresponding to the newly added symbol, the symbol recognition model is updated so that the updated symbol recognition model can recognize the newly added symbol.

2. The method according to claim 1, characterized in that, The step of determining the Gaussian variance corresponding to the newly added symbol based on the character variance corresponding to the newly added symbol includes: Determine the variance auxiliary calculation data corresponding to the newly added symbol. The variance auxiliary calculation data includes at least one of global variance, local variance, and identity matrix corresponding to the newly added symbol. The variance auxiliary calculation data is used to combine the character variance corresponding to the newly added symbol to determine the Gaussian variance corresponding to the newly added symbol. The mean of the variance-aided calculation data and the character variance corresponding to the newly added symbol is determined as the Gaussian variance corresponding to the newly added symbol.

3. The method according to claim 2, characterized in that, The method for determining the global variance includes: Determine the symbolic features of at least one original symbol that the symbol recognition model can recognize; The global variance is determined based on the symbol features corresponding to each of the at least one newly added symbol sample image and the symbol features of each of the at least one original symbol.

4. The method according to claim 2, characterized in that, The method for determining the local variance includes: Based on the Gaussian mean corresponding to the newly added symbol, the symbol features corresponding to each of the at least one newly added symbol sample image, and the symbol features of each of the at least one original symbol, determine the top N symbol features that are closest to the Gaussian mean corresponding to the newly added symbol; The local variance is determined based on the first N symbolic features.

5. The method according to claim 1, characterized in that, The step of updating the symbol recognition model based on the Gaussian mean corresponding to the newly added symbol includes: Determine the identity matrix corresponding to the Gaussian mean of the newly added symbol, wherein the identity matrix is ​​an identity matrix with the same dimension as the Gaussian mean; The symbol recognition model is updated based on the Gaussian mean corresponding to the newly added symbol and the identity matrix corresponding to the Gaussian mean.

6. A method for grading homework, characterized in that, include: Acquire images of target jobs that show correction marks; The target task image is processed using a symbol recognition model to obtain the symbol recognition result corresponding to the target task image, wherein the symbol recognition model is obtained by the method described in any one of claims 1 to 5.

7. A model processing device, characterized in that, include: The acquisition module is used to acquire at least one sample image of the newly added symbol in response to a request for adding a new symbol to the symbol recognition model. The update module is used to extract features from each of the at least one newly added symbol sample image to obtain the symbol features corresponding to each of the at least one newly added symbol sample image; Based on the symbol features corresponding to each of the at least one newly added symbol sample image, determine the Gaussian mean corresponding to the newly added symbol; Based on the Gaussian mean corresponding to the newly added symbol and the symbol features corresponding to each of the at least one newly added symbol sample image, the character variance corresponding to the newly added symbol is determined. The character variance is determined by calculating the mean of the product between matrix A and the transpose of A. Matrix A is the difference matrix between the symbol feature vector of the newly added symbol sample image and the mean vector of the symbol feature vector of the newly added symbol sample image. Based on the character variance corresponding to the newly added symbol, determine the Gaussian variance corresponding to the newly added symbol; Based on the Gaussian mean and Gaussian variance corresponding to the newly added symbol, the symbol recognition model is updated so that the updated symbol recognition model can recognize the newly added symbol.

8. A homework correction device, characterized in that, include: The acquisition module is used to acquire images of target jobs that have correction marks. The processing module is used to process the target operation image using a symbol recognition model to obtain the symbol recognition result corresponding to the target operation image, wherein the symbol recognition model is obtained by the method described in any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program for performing the method described in any one of claims 1 to 6.

10. An electronic device, characterized in that, The electronic device includes: processor; A memory for storing processor-executable instructions; the processor for performing the method according to any one of claims 1 to 6.

Images