Connection question correction method, storage medium and equipment

Abstract

The invention discloses a connection question correcting method, a storage medium and equipment. The method comprises the following steps: acquiring an image of a to-be-corrected connection question; identifying the image of the to-be-corrected line connection question by adopting a target detection and identification model, and obtaining line connection classification and a connection relationship in the image of the to-be-corrected line connection question; according to the connection line classification and the connection relation obtained through recognition, the connection line classification and the connection relation are compared with the connection classification and the connection relation in a preset standard answer, if the connection line classification and the connection relation are completely matched, the correction result is correct, and otherwise, the correction result is wrong. Compared with the prior art, the connection questions are used as a whole to be compared and recognized to obtain the connection line classification and the connection line relation of the connection questions, each connection line and the connection object connected with the connection line do not need to be specifically recognized, the accuracy rate is high, and implementation is easy.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent marking, and in particular to a connecting question marking method, storage medium and device. Background Art

[0002] At present, photo-taking and question-judgment has been widely used in various electronic educational devices. Users can take photos of test questions through terminal devices and upload them to the application. The application uses the question type judgment model to perform layout analysis and text analysis on the image to obtain the layout frame and text frame. At the same time, according to the characteristics of the question type, the layout frame and text frame corresponding to each test question are analyzed, and then the layout frame and text frame are cropped. The cropped image is used to recognize the question content through the recognition model, and the question is judged according to the question content.

[0003] However, the above method can only handle limited types of questions, such as arithmetic questions (horizontal, vertical, freehand, etc.), word problems, etc., but there are still quite a few types of questions that cannot be effectively graded using the above method, such as connecting questions. Since connecting questions come in many forms and rich varieties, it is generally difficult to grade them using conventional logical rules. Therefore, how to accurately and efficiently grade connecting questions has become a problem that needs to be solved in the marking process. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention aims to provide a connecting question grading method, storage medium and device.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] A method for correcting a connecting question comprises the following steps:

[0007] S1, obtaining an image of the connecting-the-dots question to be corrected;

[0008] S2, using a target detection and recognition model to recognize the image of the connect-the-wire problem to be corrected, and obtaining the connection classification and connection relationship in the image of the connect-the-wire problem to be corrected;

[0009] The target detection and recognition model is trained and learned in advance using sample images of connect-the-line questions of different connection classifications. A connection relationship is made in each sample image of connect-the-line questions, and the connection classification and connection relationship are marked. The connect-the-line question sample image data belonging to the same connection classification includes all connect-the-line question sample images of different connection modes.

[0010] The representation of the connection classification includes information on the number of levels in the connection problem and the number of connection objects in each level; each connection object at each level is named according to a set naming rule, and the representation of the connection relationship includes the names of the corresponding two connection objects;

[0011] The object detection and recognition model trained by the above-mentioned connect-the-dots problem sample images can recognize the connection samples and connection relationships of the images of the connect-the-dots problem to be corrected;

[0012] S3. Compare the connection classification and connection relationship obtained according to step S2 with the connection classification and connection relationship in the preset standard answer. If the connection classification and connection relationship are completely matched, the correction result is correct, otherwise the correction result is wrong.

[0013] Furthermore, in step S2, the sample image of the connect-the-line question discards the question stem information, the connect-the-line object is virtually constructed, and the target detection and recognition model only detects the connect-the-line classification and connect-the-line relationship in the sample image of the connect-the-line question.

[0014] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0015] The present invention also provides an electronic device, comprising a processor and a memory, wherein the memory is used to store a computer program; and when the processor is used to execute the computer program, the above method is implemented.

[0016] The beneficial effect of the present invention is that the present invention compares and identifies the connection questions themselves as a whole to obtain the connection classification and connection relationship of the connection questions, without specifically identifying each connection and the connection object connected by the connection, with high accuracy and simple implementation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is an example diagram of the classification and connection method of the connection questions in the embodiment of the present invention. DETAILED DESCRIPTION

[0018] The present invention will be further described below in conjunction with the accompanying drawings. It should be noted that this embodiment is based on the technical solution and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to this embodiment.

[0019] This embodiment provides a method for correcting a connecting question, comprising the following steps:

[0020] S1, obtaining an image of the connecting-the-dots question to be corrected;

[0021] S2, using a target detection and recognition model to recognize the image of the connect-the-wire problem to be corrected, and obtaining the connection classification and connection relationship in the image of the connect-the-wire problem to be corrected;

[0022] The target detection and recognition model is pre-trained and learned using a large number of sample images of connection questions with different connection classifications. A connection relationship is made in each sample image of the connection question, and the connection classification and connection relationship are marked; the connection classification is represented by AXB, B represents the number of levels contained in the connection question, and A represents the number of connection objects contained in each layer; each connection object at each level is named according to the set naming rules, and the connection relationship is represented by a combination of the names of the corresponding two connection objects.

[0023] For example, in a connection problem with a connection category of 3X2, there are two levels, with three connection objects in each level. In this connection category, there are 6 connection methods, such as Figure 1 When naming, capital letters are used to name the first-layer connection objects, and numbers are used to name the second-layer connection objects. The connection relationship between connection object A and connection object 1 is represented by A1.

[0024] The connect-the-line problem sample image data belonging to the same connection classification includes all connect-the-line problem sample images of different connection modes. For example, the connect-the-line problem sample images include all connection classifications such as 2X2, 3X2, and 4X2, and the connect-the-line problem sample images with the connection classification of 3X2 also include all 6 connection modes of connect-the-line problem sample images, and the same is true for the connect-the-line problem sample images of other connection classifications.

[0025] The object detection and recognition model trained with the above-mentioned connect-the-dots problem sample images can identify connection samples and connection relationships in the images of the connect-the-dots problems to be corrected.

[0026] It should be noted that the target detection and recognition model can adopt the YOLOv model based on deep learning.

[0027] S3. Compare the connection classification and connection relationship obtained according to step S2 with the connection classification and connection relationship in the preset standard answer. If the connection classification and connection relationship are completely matched, the correction result is correct, otherwise the correction result is wrong.

[0028] By comparing the connection relationship identified by the target detection and recognition model with the connection relationship of the standard answer, we can also find out which connection relationships in the student's answer are wrong. For example, the standard answer is a 3X2 structure of A1, B2, C3, and the connection relationship identified by the target detection and recognition model is A2, B1, C3, so we can know that A2 and B1 are connected incorrectly.

[0029] In this embodiment, the sample image of the connect-the-line question discards the question stem information, the connect-the-line object is virtually constructed, and the target detection and recognition model only detects the connect-the-line classification and connect-the-line relationship in the sample image of the connect-the-line question.

[0030] For those skilled in the art, various corresponding changes and modifications can be made according to the above technical solutions and concepts, and all of these changes and modifications should be included in the protection scope of the claims of the present invention.

Claims

1. A method for correcting connecting questions, characterized in that: The steps include: S1, obtaining an image of the connecting-the-dots question to be corrected; S2, using a target detection and recognition model to recognize the image of the connect-the-wire problem to be corrected, and obtaining the connection classification and connection relationship in the image of the connect-the-wire problem to be corrected; The target detection and recognition model is trained and learned in advance using sample images of connect-the-line questions of different connection classifications. A connection relationship is made in each sample image of connect-the-line questions, and the connection classification and connection relationship are marked. The connect-the-line question sample image data belonging to the same connection classification includes all connect-the-line question sample images of different connection modes. The representation of the connection classification includes information on the number of levels in the connection problem and the number of connection objects in each level; each connection object at each level is named according to a set naming rule, and the representation of the connection relationship includes the names of the corresponding two connection objects; The object detection and recognition model trained by the above-mentioned connect-the-dots problem sample images can recognize the connection samples and connection relationships of the images of the connect-the-dots problem to be corrected; S3. Compare the connection classification and connection relationship obtained according to step S2 with the connection classification and connection relationship in the preset standard answer. If the connection classification and connection relationship are completely matched, the correction result is correct, otherwise the correction result is wrong.

2. The method according to claim 1, characterized in that In step S2, the sample image of the connect-the-line question discards the question stem information, the connect-the-line object is virtually constructed, and the target detection and recognition model only detects the connect-the-line classification and connect-the-line relationship in the sample image of the connect-the-line question.

3. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 2 is implemented.

4. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory is used to store a computer program; when the processor is used to execute the computer program, the method described in any one of claims 1 to 2 is implemented.

Images