Art design ability training information management method and system
Through deep learning artificial intelligence detection technology and convolutional neural network model with dual attention mechanism, feature mining is carried out on art design assessment painting images, solving the accuracy of art design ability evaluation in the existing technology, and realizing art design ability accurate evaluation and optimization of the cultivation system.
Patent Information
- Application Number
- CN202310229345.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-03-07
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2043-03-07
AI Technical Summary
The existing technology has errors and subjectivity in the evaluation of art design ability, making it difficult to accurately evaluate students' art design ability, resulting in difficulty in optimizing the training system.
Using deep learning artificial intelligence detection technology, a convolutional neural network model with dual attention mechanism is used to mine the painting images of art design assessment in the first and second academic years, capture the information of implicit differential feature distribution, and evaluate it through classifiers. Whether the artistic design capabilities are improved can optimize the training system.
It realizes accurate assessment of art design capabilities, reduces the errors and subjectivity of manual evaluation, and improves the optimization effect of the training system.
Smart Images

Figure CN116451836B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent management of talent training, and more specifically, to an art design ability training information management method and system thereof. Background Art
[0002] In recent years, with the annual expansion of enrollment in colleges and universities, the training of college students has become more and more complicated, especially the training of innovation and entrepreneurship. Art design ability is part of the comprehensive ability assessment of students, and it is of great significance in the talent training plan.
[0003] Art design refers to design based on engineering, aesthetics and economics. The practicality, beauty and environment of the product are the main contents of design research. In the teaching of art design, students need to be trained in design, modeling and manufacturing capabilities, so that students can have independent innovation capabilities from conception to design to manufacturing.
[0004] At present, in the process of cultivating students' art design ability, since students' art design ability is gradually improving, it is necessary to continuously update students' art design ability information. That is to say, it is necessary to record and compare students' art design assessment information every once in a while to determine whether the changes in their art design ability meet the predetermined standards, so as to optimize the art design ability training system.
[0005] Most of the existing solutions rely on manual statistical comparison of students' art design assessment information. Not only will manual statistical evaluation be incorrect, but it is also highly subjective, making it difficult to accurately evaluate students' art design abilities.
[0006] Therefore, an optimized information management solution for cultivating art design capabilities is desired. Summary of the invention
[0007] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides an art design ability training information management method and system, which is based on deep learning artificial intelligence detection technology, to mine features of first-year art design assessment painting images and second-year art design assessment painting images respectively by using a convolutional neural network model with a dual attention mechanism, and capture the implicit difference feature distribution information of the two in high-dimensional space, so as to accurately evaluate the art design ability of students, make corresponding adjustments to the art design ability training program, and optimize the art design ability training system.
[0008] According to one aspect of the present application, there is provided an art design ability training information management method, which includes:
[0009] Obtaining images of the first-year art design assessment paintings and second-year art design assessment paintings of the students to be assessed;
[0010] The first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively passed through a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map;
[0011] Performing feature distribution correction along the channel dimension on the first assessment painting feature map and the second assessment painting feature map to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map;
[0012] The first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively passed through a convolutional neural network model using a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map;
[0013] The corrected first assessment painting feature map and the third assessment painting feature map are merged to obtain a first multi-dimensional aggregated assessment painting feature map, and the corrected second assessment painting feature map and the fourth assessment painting feature map are merged to obtain a second multi-dimensional aggregated assessment painting feature map;
[0014] Calculating a differential feature map between the first multi-dimensional aggregated assessment painting feature map and the second multi-dimensional aggregated assessment painting feature map; and
[0015] The differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has improved.
[0016] In the above-mentioned art design ability training information management method, the first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively passed through a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map, including: using the convolutional coding part of the convolutional neural network model using the spatial attention mechanism to perform deep convolution coding on the first academic year art design assessment painting image and the second academic year art design assessment painting image to obtain a first academic year initial convolutional feature map and a second academic year initial convolutional feature map; the first academic year initial convolutional feature map and the second academic year initial convolutional feature map are respectively passed through a convolutional neural network model using a spatial attention mechanism to obtain a first academic year initial convolutional feature map and a second academic year initial convolutional feature map. Respectively input the spatial attention part of the convolutional neural network model using the spatial attention mechanism to obtain the first academic year spatial attention map and the second academic year spatial attention map; respectively pass the first academic year spatial attention map and the second academic year spatial attention map through the Softmax activation function to obtain the first academic year spatial attention feature map and the second academic year spatial attention feature map; and calculate the point-by-point multiplication of the first academic year spatial attention feature map and the first academic year initial convolution feature map to obtain the first assessment painting feature map, and calculate the point-by-point multiplication of the second academic year spatial attention feature map and the second academic year initial convolution feature map to obtain the second assessment painting feature map.
[0017] In the above-mentioned art design ability training information management method, the feature distribution of the first assessment painting feature map and the second assessment painting feature map is corrected along the channel dimension to obtain the corrected first assessment painting feature map and the corrected second assessment painting feature map, including: respectively modeling the feature correlation accumulation distinction mechanism of the first assessment painting feature map and the second assessment painting feature map to obtain the first weighted feature vector and the second weighted feature vector; and, respectively weighting the first assessment painting feature map and the second assessment painting feature map in the channel dimension with the first weighted feature vector and the second weighted feature vector to obtain the corrected first assessment painting feature map and the corrected second assessment painting feature map.
[0018] In the above-mentioned art design ability training information management method, the first assessment painting feature graph and the second assessment painting feature graph are respectively modeled with a feature correlation accumulation and differentiation mechanism to obtain a first weighted feature vector and a second weighted feature vector, including: modeling the first assessment painting feature graph with a feature correlation accumulation and differentiation mechanism to obtain the first weighted feature vector according to the following formula; wherein the formula is:
[0019]
[0020] Where F1 represents the feature map of the first test painting, cov1(·) and cov2(·) represent single-layer convolution operations based on different convolution kernels on the feature map, ReLU(·) represents the ReLU activation function, Sigmoid(·) represents the Sigmoid activation function, and GAP represents the global pooling operation on each feature matrix of the feature map. represents positional addition, V1 represents the first weighted feature vector; the second assessment painting feature graph is modeled with a feature correlation accumulation and differentiation mechanism using the following formula to obtain the second weighted feature vector; wherein the formula is:
[0021]
[0022] Where F2 represents the feature map of the second test painting, cov1(·) and cov2(·) represent single-layer convolution operations based on different convolution kernels on the feature map, ReLU(·) represents the ReLU activation function, Sigmoid(·) represents the Sigmoid activation function, and GAP represents the global pooling operation on each feature matrix of the feature map. represents positional addition, and V2 represents the first weighted feature vector.
[0023] In the above-mentioned art design ability training information management method, the first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively passed through a convolutional neural network model using a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map, including: using each layer of the convolutional neural network model using a channel attention mechanism to perform the following on the input data in the forward pass of the layer: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution feature map; performing pooling processing on the convolution feature map to generate a pooling feature map; performing activation processing on the pooling feature map to generate an activation feature map ; Calculate the global mean of each feature matrix along the channel dimension of the activation feature map to obtain a channel feature vector; Calculate the ratio of the eigenvalue of each position in the channel feature vector to the weighted sum of the eigenvalues of all positions of the channel feature vector to obtain a channel weighted feature vector; and, perform dot multiplication on the feature matrix along the channel dimension of the activation feature map using the eigenvalue of each position of the channel weighted feature vector as a weight to obtain a generated feature map; wherein the generated feature map output by the last layer of the convolutional neural network model using the channel attention mechanism is the third assessment painting feature map or the fourth assessment painting feature map.
[0024] In the above-mentioned art design ability training information management method, the first assessment painting feature map after correction and the third assessment painting feature map are integrated to obtain a first multi-dimensional aggregated assessment painting feature map, and the second assessment painting feature map after correction and the fourth assessment painting feature map are integrated to obtain a second multi-dimensional aggregated assessment painting feature map, including: integrating the first assessment painting feature map after correction and the third assessment painting feature map to obtain a first multi-dimensional aggregated assessment painting feature map according to the following formula; wherein the formula is:
[0025] X=Concat[F1 ′ ,F3]
[0026] Among them, F1 ′ represents the corrected first assessment painting feature map, F3 represents the third assessment painting feature map, Concat[·,·] represents a cascade function, and X represents the first multi-dimensional aggregated assessment painting feature map; the corrected second assessment painting feature map and the fourth assessment painting feature map are fused by the following formula to obtain a second multi-dimensional aggregated assessment painting feature map; wherein the formula is:
[0027] Y=Concat[F2 ′ ,F4]
[0028] Among them, F2 ′ represents the corrected second assessment painting feature map, F4 represents the four assessment painting feature maps, Concat[·,·] represents the cascade function, and Y represents the second multi-dimensional aggregated assessment painting feature map.
[0029] In the above-mentioned art design ability training information management method, calculating the difference feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map includes: calculating the difference feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map using the following formula; wherein the formula is: Among them, X represents the first multi-dimensional aggregate assessment painting feature map, Y represents the second multi-dimensional aggregate assessment painting feature map, and Z represents the difference feature map. Indicates difference by position.
[0030] In the above-mentioned art design ability training information management method, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has been improved, including: expanding each differential feature matrix in the differential feature map into a one-dimensional feature vector according to a row vector or a column vector and then cascading to obtain a classification feature vector; inputting the classification feature vector into the Softmax classification function of the classifier to obtain the probability value of the classification feature vector belonging to each classification label; and determining the classification label corresponding to the maximum probability value as the classification result.
[0031] According to another aspect of the present application, there is provided an art design capability training information management system, comprising:
[0032] An assessment painting acquisition module is used to acquire the first-year art design assessment painting images and the second-year art design assessment painting images of the students to be assessed;
[0033] A spatial enhancement module, for respectively applying the first academic year art design assessment painting image and the second academic year art design assessment painting image to a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map;
[0034] a feature distribution correction module, configured to perform feature distribution correction on the first assessment painting feature map and the second assessment painting feature map along a channel dimension to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map;
[0035] A channel enhancement module, used for respectively obtaining a third assessment painting feature map and a fourth assessment painting feature map by using a convolutional neural network model with a channel attention mechanism for the first academic year art design assessment painting image and the second academic year art design assessment painting image;
[0036] a multi-dimensional aggregation module, configured to fuse the corrected first assessment painting feature map and the third assessment painting feature map to obtain a first multi-dimensional aggregated assessment painting feature map, and to fuse the corrected second assessment painting feature map and the fourth assessment painting feature map to obtain a second multi-dimensional aggregated assessment painting feature map;
[0037] A difference module, used to calculate a difference feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map; and
[0038] The evaluation result generation module is used to pass the differential feature map through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has improved.
[0039] According to another aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the above-mentioned art design ability training information management method.
[0040] According to another aspect of the present application, a computer-readable medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the above-mentioned art design ability training information management method.
[0041] Compared with the prior art, the art design ability training information management method and system provided by the present application are based on deep learning artificial intelligence detection technology, which uses a convolutional neural network model with a dual attention mechanism to perform feature mining on the first-year art design assessment painting images and the second-year art design assessment painting images respectively, and capture the implicit difference feature distribution information of the two in high-dimensional space, so as to accurately evaluate the students' art design ability, make corresponding adjustments to the art design ability training program, and thus optimize the art design ability training system. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0043] Figure 1 The present invention is a flowchart of an art design ability training information management method according to an embodiment of the present application.
[0044] Figure 2 It is an architectural diagram of the art design ability training information management method according to an embodiment of the present application.
[0045] Figure 3 A flowchart of the art design ability training information management method according to an embodiment of the present application is provided for obtaining a first assessment painting feature map and a second assessment painting feature map by respectively using a convolutional neural network model with a spatial attention mechanism for the first academic year art design assessment painting image and the second academic year art design assessment painting image.
[0046] Figure 4 In the art design ability training information management method according to the embodiment of the present application, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has been improved.
[0047] Figure 5 It is a block diagram of an art design ability training information management system according to an embodiment of the present application.
[0048] Figure 6 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION
[0049] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0050] Application Overview
[0051] As mentioned above, currently, most of the art design ability assessment schemes for students rely on manual statistical comparison of students' art design assessment information. Not only will manual statistical assessments be incorrect, but they are also highly subjective, making it difficult to accurately assess students' art design abilities. Therefore, an optimized art design ability training information management scheme is desired.
[0052] Accordingly, in order to improve the accuracy of the art design ability assessment of students, the art design ability can be assessed by comparing the art design assessment information of the same student in different academic years, such as comparing the assessment paintings of the same student in different academic years. However, considering that there is a lot of information in the art design assessment painting images and the abstract nature of art design ability, it is difficult to evaluate the art design ability of students through traditional evaluation methods. In the actual evaluation process, the difficulty lies in how to fully and accurately dig out the implicit difference feature distribution information of the art design assessment painting images of the students in the first academic year and the art design assessment painting images of the students in the second academic year, so as to accurately evaluate the art design ability of the students and optimize the art design ability training system.
[0053] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields.
[0054] The development of deep learning and neural networks has provided new solutions and plans for mining the implicit difference feature distribution information of students' first-year art design assessment painting images and second-year art design assessment painting images.
[0055] Specifically, in the technical solution of the present application, first, the first-year art design assessment painting image and the second-year art design assessment painting image of the student to be evaluated are obtained. Then, for the first-year art design assessment painting image and the second-year art design assessment painting image, a convolutional neural network model with excellent performance in image implicit feature extraction is used as a feature extractor to extract high-dimensional local image hidden feature information in the first-year art design assessment painting image and the second-year art design assessment painting image.
[0056] In particular, considering that when extracting features from students' art design assessment painting images to evaluate their art design abilities, due to the abstractness of art design abilities, the art design abilities also have abstract expressions in the art design assessment painting images, and it is difficult to effectively capture and extract the feature information of the art design abilities. In addition, the hidden feature information about the students' art design abilities in the art design assessment painting images is small-scale feature information relative to the painting images, and is manifested in local areas of the images. Therefore, in order to fully and accurately extract the small-scale local correlation hidden feature information about the students' art design abilities in the first academic year art design assessment painting images and the second academic year art design assessment painting images, so as to accurately evaluate the students' art design abilities, in the technical solution of the present application, a convolutional neural network model with a dual attention mechanism is used to perform feature mining on the first academic year art design assessment painting images and the second academic year art design assessment painting images, respectively.
[0057] That is, it should be understood that, considering that the network will obtain partial feature information after a series of convolutions, but will not automatically distinguish the differences between the features of each category, the network's ability to selectively use features is limited. Given that the attention mechanism can select the focus position, it can produce a more distinguishable feature representation, and the features after the attention module is added will change adaptively as the network deepens. Therefore, in the technical solution of the present application, convolutional neural network models with channel attention mechanisms and spatial attention mechanisms are introduced to perform feature mining on the first academic year art design assessment painting image and the second academic year art design assessment painting image, respectively, so as to extract the high-dimensional implicit feature distribution information about the student's art design ability that focuses on the spatial position and channel content in the first academic year art design assessment painting image and the second academic year art design assessment painting image, respectively, so as to obtain the first assessment painting feature map and the second assessment painting feature map, as well as the third assessment painting feature map and the fourth assessment painting feature map. It should be understood that the image features extracted by the channel attention reflect the correlation and importance between the feature channels, and the image features extracted by the spatial attention reflect the weight of the spatial dimension feature difference, which is used to suppress or strengthen the features of different spatial positions. The channel attention and the spatial attention can respectively focus on the feature content and feature position in the image, and complement each other to a certain extent, thereby improving the feature extraction effect of the network, so that a large amount of distribution information on the characteristics of the student's artistic design ability in the assessment painting image at different spatial positions and between channel contents can be captured, which can effectively enhance the feature recognition learning ability. During the network training process, the task processing system is more focused on finding useful information in the input image data that is significantly related to the current output, thereby improving the quality of the output, and the increasing attention module will bring continuous performance improvement.
[0058] Furthermore, the first assessment painting feature map and the third assessment painting feature map are fused to fuse the high-dimensional hidden features of the art design ability of the student to be assessed that are focused on the spatial position and the channel content in the art design assessment painting image of the first academic year, thereby obtaining a first multi-dimensional aggregated assessment painting feature map; and the second assessment painting feature map and the fourth assessment painting feature map are fused to fuse the high-dimensional hidden features of the art design ability of the student to be assessed that are focused on the spatial position and the channel content in the art design assessment painting image of the second academic year, thereby obtaining a second multi-dimensional aggregated assessment painting feature map.
[0059] Then, the differential feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map is further calculated to represent the differential feature distribution information between the hidden performance features of the art design ability in the art design assessment painting image of the student to be evaluated in the first academic year and the second academic year. Then, the differential feature map is used as a classification feature map for classification processing in a classifier to obtain a classification result for indicating whether the art design ability of the student to be evaluated has been improved. That is, in the technical solution of the present application, the label of the classifier includes the art design ability of the student to be evaluated has been improved, and the art design ability of the student to be evaluated has not been improved, wherein the classifier determines which classification label the classification feature map belongs to by a soft maximum function. It should be understood that in the technical solution of the present application, the classification label of the classifier is a judgment label for whether the art design ability of the student to be evaluated has been improved. Therefore, after obtaining the classification result, it can be accurately judged based on the classification result whether the art design ability of the student to be evaluated has been improved during this period, so as to make corresponding adjustments to the art design ability training program to optimize the art design ability training system.
[0060] In particular, in the technical solution of the present application, when the first assessment painting feature map and the third assessment painting feature map are fused to obtain the first multi-dimensional aggregated assessment painting feature map, and the second assessment painting feature map and the fourth assessment painting feature map are fused to obtain the second multi-dimensional aggregated assessment painting feature map, it is considered that the first assessment painting feature map and the second assessment painting feature map are obtained by using a convolutional neural network model using a spatial attention mechanism, which may have channel dimension distribution differences relative to the third assessment painting feature map and the fourth assessment painting feature map obtained by using a convolutional neural network model using a channel attention mechanism, thereby affecting the fusion effect.
[0061] Therefore, the applicant of the present application considers that if the respective channel positions of the first assessment painting feature map and the second assessment painting feature map can be appropriately weighted to distinguish the importance of the feature values at respective positions of the first assessment painting feature map and the second assessment painting feature map, that is, applying a channel confidence weighting effect similar to the channel attention mechanism, then it is obvious that the expression effect between the feature maps can be improved.
[0062] Specifically, the first assessment painting feature graph F1 and the second assessment painting feature graph F2 are respectively modeled with a feature correlation accumulation and differentiation mechanism to obtain a first weighted feature vector V1 and a second weighted feature vector V2, which are expressed as:
[0063]
[0064]
[0065] Here, cov1(F) and cov2(F) respectively represent single-layer convolution operations based on different convolution kernels on the feature map, and GAP represents a global pooling operation on each feature matrix of the feature map.
[0066] The feature correlation accumulation and differentiation mechanism modeling first generates two new local correlation units of the feature map through convolution operations, and then uses Sigmoid function and ReLU function to perform simple embedding, resetting and updating of local correlation features similar to the neural network architecture, and then accumulates the correlation of local features with respect to the overall features through global average pooling operations, so as to use the feature differentiation mechanism to explicitly model the feature importance sequence, and the feature accumulation and differentiation mechanism based on each feature matrix of the feature map can determine the appropriate weighting factor under the channel dimension. In this way, by weighting the first assessment painting feature map F1 and the second assessment painting feature map F2 in the channel dimension respectively with the first weighted feature vector V1 and the second weighted feature vector V2, the fusion effect of the first assessment painting feature map and the second assessment painting feature map with the third assessment painting feature map and the fourth assessment painting feature map can be improved. In this way, the art design ability of students can be accurately evaluated to optimize the training system of art design ability.
[0067] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.
[0068] Exemplary Methods
[0069] Figure 1 Flow chart of the method for managing information on art design ability training according to an embodiment of the present application. Figure 1As shown, according to the art design ability training information management method of the embodiment of the present application, it includes: S110, obtaining the first academic year art design assessment painting image and the second academic year art design assessment painting image of the student to be evaluated; S120, respectively using the convolutional neural network model of the first academic year art design assessment painting image and the second academic year art design assessment painting image to obtain a first assessment painting feature map and a second assessment painting feature map; S130, performing feature distribution correction along the channel dimension on the first assessment painting feature map and the second assessment painting feature map to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map; S140, the first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively obtained by using a convolutional neural network model of the spatial attention mechanism. The art design assessment painting images of the year are respectively obtained by using a convolutional neural network model with a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map; S150, the corrected first assessment painting feature map and the third assessment painting feature map are fused to obtain a first multi-dimensional aggregated assessment painting feature map, and the corrected second assessment painting feature map and the fourth assessment painting feature map are fused to obtain a second multi-dimensional aggregated assessment painting feature map; S160, a differential feature map between the first multi-dimensional aggregated assessment painting feature map and the second multi-dimensional aggregated assessment painting feature map is calculated; and, S170, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has improved.
[0070] Figure 2 FIG. 1 is a schematic diagram of an art design capability training information management method according to an embodiment of the present application. Figure 2As shown, in this architecture, first, the first-year art design assessment painting image and the second-year art design assessment painting image of the student to be assessed are obtained; then, the first-year art design assessment painting image and the second-year art design assessment painting image are respectively passed through a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map; then, the first assessment painting feature map and the second assessment painting feature map are corrected for feature distribution along the channel dimension to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map, and at the same time, the first-year art design assessment painting image and the second-year art design assessment painting image are respectively passed through a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map. For example, a convolutional neural network model with a channel attention mechanism is used to obtain a third assessment painting feature map and a fourth assessment painting feature map; then, the corrected first assessment painting feature map and the third assessment painting feature map are fused to obtain a first multi-dimensional aggregated assessment painting feature map, and the corrected second assessment painting feature map and the fourth assessment painting feature map are fused to obtain a second multi-dimensional aggregated assessment painting feature map; then, a differential feature map between the first multi-dimensional aggregated assessment painting feature map and the second multi-dimensional aggregated assessment painting feature map is calculated; finally, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the artistic design ability of the student to be evaluated has improved.
[0071] As mentioned above, currently, most of the art design ability assessment schemes for students rely on manual statistical comparison of students' art design assessment information. Not only will manual statistical assessments be incorrect, but they are also highly subjective, making it difficult to accurately assess students' art design abilities. Therefore, an optimized art design ability training information management scheme is desired.
[0072] Accordingly, in order to improve the accuracy of the art design ability assessment of students, the art design ability can be assessed by comparing the art design assessment information of the same student in different academic years, such as comparing the assessment paintings of the same student in different academic years. However, considering that there is a lot of information in the art design assessment painting images and the abstract nature of art design ability, it is difficult to evaluate the art design ability of students through traditional evaluation methods. In the actual evaluation process, the difficulty lies in how to fully and accurately dig out the implicit difference feature distribution information of the art design assessment painting images of the students in the first academic year and the art design assessment painting images of the students in the second academic year, so as to accurately evaluate the art design ability of the students and optimize the art design ability training system.
[0073] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields.
[0074] The development of deep learning and neural networks provides new solutions and plans for mining the implicit difference feature distribution information of students' first-year art design assessment painting images and second-year art design assessment painting images.
[0075] In step S110 and step S120, the first-year art design assessment painting image and the second-year art design assessment painting image of the student to be assessed are obtained, and the first-year art design assessment painting image and the second-year art design assessment painting image are respectively subjected to a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map. For the first-year art design assessment painting image and the second-year art design assessment painting image, a convolutional neural network model with excellent performance in extracting implicit features of images is used as a feature extractor to extract high-dimensional local image hidden feature information in the first-year art design assessment painting image and the second-year art design assessment painting image.
[0076] In particular, considering that when extracting features from students' art design assessment painting images to evaluate their art design abilities, due to the abstractness of art design abilities, the art design abilities also have abstract expressions in the art design assessment painting images, and it is difficult to effectively capture and extract the feature information of the art design abilities. In addition, the hidden feature information about the students' art design abilities in the art design assessment painting images is small-scale feature information relative to the painting images, and is manifested in local areas of the images. Therefore, in order to fully and accurately extract the small-scale local correlation hidden feature information about the students' art design abilities in the first academic year art design assessment painting images and the second academic year art design assessment painting images, so as to accurately evaluate the students' art design abilities, in the technical solution of the present application, a convolutional neural network model with a dual attention mechanism is used to perform feature mining on the first academic year art design assessment painting images and the second academic year art design assessment painting images, respectively.
[0077] That is, it should be understood that, considering that the network will obtain partial feature information after a series of convolutions, but will not automatically distinguish the differences between the features of each category, the network's ability to selectively use features is limited. Given that the attention mechanism can select the focus position, it produces a more discriminative feature representation, and the features after the attention module is added will produce adaptive changes as the network deepens. Therefore, in the technical solution of the present application, a convolutional neural network model with a spatial attention mechanism is introduced to perform feature mining on the first academic year art design assessment painting image and the second academic year art design assessment painting image, so as to extract the high-dimensional implicit feature distribution information about the art design ability of the student focused on the spatial position in the first academic year art design assessment painting image and the second academic year art design assessment painting image, thereby obtaining the first assessment painting feature map and the second assessment painting feature map. It should be understood that the image features extracted by the spatial attention reflect the weight of the spatial dimension feature difference, which is used to suppress or strengthen the features of different spatial positions.
[0078] Figure 3 The present invention is a flowchart of the method for managing information on art design ability training according to an embodiment of the present application, wherein the first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively obtained by using a convolutional neural network model with a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map. Figure 3 As shown, the process of respectively passing the first academic year art design assessment painting image and the second academic year art design assessment painting image through a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map comprises: S210, using the convolutional coding part of the convolutional neural network model using the spatial attention mechanism to perform deep convolution coding on the first academic year art design assessment painting image and the second academic year art design assessment painting image to obtain a first academic year initial convolutional feature map and a second academic year initial convolutional feature map; S220, inputting the first academic year initial convolutional feature map and the second academic year initial convolutional feature map into the convolutional neural network model respectively; The spatial attention part of the convolutional neural network model using the spatial attention mechanism is used to obtain the spatial attention map of the first academic year and the spatial attention map of the second academic year; S230, the first academic year spatial attention map and the second academic year spatial attention map are respectively passed through the Softmax activation function to obtain the first academic year spatial attention feature map and the second academic year spatial attention feature map; and, S240, the point-by-point multiplication of the first academic year spatial attention feature map and the first academic year initial convolution feature map is calculated to obtain the first assessment painting feature map, and the point-by-point multiplication of the second academic year spatial attention feature map and the second academic year initial convolution feature map is calculated to obtain the second assessment painting feature map.
[0079] In step S130, the first assessment painting feature map and the second assessment painting feature map are subjected to feature distribution correction along the channel dimension to obtain the corrected first assessment painting feature map and the corrected second assessment painting feature map. In particular, in the technical solution of the present application, when the first assessment painting feature map and the third assessment painting feature map are subsequently fused to obtain the first multi-dimensional aggregated assessment painting feature map, and the second assessment painting feature map and the fourth assessment painting feature map are fused to obtain the second multi-dimensional aggregated assessment painting feature map, it is considered that the first assessment painting feature map and the second assessment painting feature map are obtained by using a convolutional neural network model using a spatial attention mechanism, which may have a channel dimension distribution difference relative to the third assessment painting feature map and the fourth assessment painting feature map obtained by using a convolutional neural network model using a channel attention mechanism, thereby affecting the fusion effect thereof.
[0080] Therefore, the applicant of the present application considers that if the respective channel positions of the first assessment painting feature map and the second assessment painting feature map can be appropriately weighted to distinguish the importance of the feature values at respective positions of the first assessment painting feature map and the second assessment painting feature map, that is, applying a channel confidence weighting effect similar to the channel attention mechanism, then it is obvious that the expression effect between the feature maps can be improved.
[0081] Specifically, the first assessment painting feature graph F1 and the second assessment painting feature graph F2 are respectively modeled with a feature correlation accumulation and differentiation mechanism to obtain a first weighted feature vector V1 and a second weighted feature vector V2, which are expressed as:
[0082]
[0083]
[0084] Wherein F1 and F2 represent the first assessment painting feature map and the second assessment painting feature map, respectively, cov1(·) and cov2(·) represent single-layer convolution operations based on different convolution kernels on the feature map, ReLU(·) represents the ReLU activation function, Sigmoid(·) represents the Sigmoid activation function, and GAP represents the global pooling operation on each feature matrix of the feature map. represents positional addition, V1 represents the first weighted feature vector, and V2 represents the first weighted feature vector.
[0085] The feature correlation accumulation and differentiation mechanism modeling first generates two new local correlation units of the feature map through convolution operations, and then uses Sigmoid function and ReLU function to perform simple embedding, resetting and updating of local correlation features similar to the neural network architecture, and then accumulates the correlation of local features with respect to the overall features through global average pooling operations, so as to use the feature differentiation mechanism to explicitly model the feature importance sequence, and the feature accumulation and differentiation mechanism based on each feature matrix of the feature map can determine the appropriate weighting factor under the channel dimension. In this way, by weighting the first assessment painting feature map F1 and the second assessment painting feature map F2 in the channel dimension respectively with the first weighted feature vector V1 and the second weighted feature vector V2, the fusion effect of the first assessment painting feature map and the second assessment painting feature map with the third assessment painting feature map and the fourth assessment painting feature map can be improved. In this way, the art design ability of students can be accurately evaluated to optimize the training system of art design ability.
[0086] In step S140, the first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively subjected to a convolutional neural network model using a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map. That is, in order to fully and accurately extract the small-scale local correlation hidden feature information about the art design ability of the students in the first academic year art design assessment painting image and the second academic year art design assessment painting image, so as to accurately evaluate the art design ability of the students, in the technical solution of the present application, a convolutional neural network model with a dual attention mechanism is used to perform feature mining on the first academic year art design assessment painting image and the second academic year art design assessment painting image, respectively. In step S140, a convolutional neural network model with a channel attention mechanism is introduced to perform feature mining on the first academic year art design assessment painting image and the second academic year art design assessment painting image, so as to extract high-dimensional implicit feature distribution information about the student's art design ability focusing on channel content in the first academic year art design assessment painting image and the second academic year art design assessment painting image, thereby obtaining a third assessment painting feature map and a fourth assessment painting feature map. It should be understood that the image features extracted by the channel attention reflect the correlation and importance between feature channels.
[0087] In the technical solution of the present application, the channel attention and the spatial attention can respectively focus on the feature content and feature position in the image, complement each other to a certain extent, and improve the feature extraction effect of the network, so that a large amount of characteristic distribution information about the student's artistic design ability in the assessment painting image at different spatial positions and between channel contents is captured, which can effectively enhance the feature recognition learning ability. During the network training process, the task processing system is more focused on finding useful information in the input image data that is significantly related to the current output, thereby improving the output quality, and the increasing attention module will bring continuous performance improvement.
[0088] Specifically, in an embodiment of the present application, the encoding process of respectively passing the first academic year art design assessment painting image and the second academic year art design assessment painting image through a convolutional neural network model using a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map includes: using each layer of the convolutional neural network model using the channel attention mechanism to perform the following steps on the input data in the forward pass of the layer: performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution feature map; performing pooling processing on the convolution feature map to generate a pooling feature map; performing activation processing on the pooling feature map to generate an activation feature map; Calculate the global mean of each feature matrix along the channel dimension of the activation feature map to obtain a channel feature vector; calculate the ratio of the eigenvalue of each position in the channel feature vector to the weighted sum of the eigenvalues of all positions of the channel feature vector to obtain a channel weighted feature vector; and, perform dot multiplication on the feature matrix along the channel dimension of the activation feature map using the eigenvalue of each position of the channel weighted feature vector as a weight to obtain a generated feature map; wherein the generated feature map output by the last layer of the convolutional neural network model using the channel attention mechanism is the third assessment painting feature map or the fourth assessment painting feature map.
[0089] In step S150, the corrected first assessment painting feature map and the third assessment painting feature map are fused to obtain a first multi-dimensional aggregated assessment painting feature map, and the corrected second assessment painting feature map and the fourth assessment painting feature map are fused to obtain a second multi-dimensional aggregated assessment painting feature map. That is, the high-dimensional hidden features of the art design ability of the student to be evaluated that focus on the spatial position and the channel content in the art design assessment painting image of the first academic year are fused to obtain the first multi-dimensional aggregated assessment painting feature map; and the high-dimensional hidden features of the art design ability of the student to be evaluated that focus on the spatial position and the channel content in the art design assessment painting image of the second academic year are fused to obtain the second multi-dimensional aggregated assessment painting feature map.
[0090] Specifically, in the embodiment of the present application, the corrected first assessment painting feature map and the third assessment painting feature map are fused by the following formula to obtain a first multi-dimensional aggregated assessment painting feature map; wherein the formula is:
[0091] X=Concat[F1 ′ ,F3]
[0092] Among them, F1 ′ represents the corrected first assessment painting feature map, F3 represents the third assessment painting feature map, Concat[·,·] represents a cascade function, and X represents the first multi-dimensional aggregated assessment painting feature map; and the corrected second assessment painting feature map and the fourth assessment painting feature map are fused with the following formula to obtain a second multi-dimensional aggregated assessment painting feature map; wherein the formula is:
[0093] Y=Concat[F2 ′ ,F4]
[0094] Among them, F2 ′ represents the corrected second assessment painting feature map, F4 represents the four assessment painting feature maps, Concat[·,·] represents the cascade function, and Y represents the second multi-dimensional aggregated assessment painting feature map.
[0095] In step S160, a differential feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map is calculated. That is, the differential feature map is used to represent the differential feature distribution information between the hidden features of the art design ability in the art design assessment painting images of the student to be assessed in the first academic year and the second academic year.
[0096] Specifically, in the embodiment of the present application, the differential feature map between the first multi-dimensional aggregated assessment painting feature map and the second multi-dimensional aggregated assessment painting feature map is calculated using the following formula; wherein the formula is: Among them, X represents the first multi-dimensional aggregate assessment painting feature map, Y represents the second multi-dimensional aggregate assessment painting feature map, and Z represents the difference feature map. Indicates difference by position.
[0097] In step S170, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has been improved. That is, in the technical solution of the present application, the labels of the classifier include whether the art design ability of the student to be evaluated has been improved, and whether the art design ability of the student to be evaluated has not been improved, wherein the classifier determines to which classification label the classification feature map belongs through a soft maximum function. It should be understood that in the technical solution of the present application, the classification label of the classifier is a judgment label of whether the art design ability of the student to be evaluated has been improved. Therefore, after obtaining the classification result, it is possible to accurately judge whether the art design ability of the student to be evaluated has been improved during this period based on the classification result, so as to make corresponding adjustments to the art design ability training program to optimize the art design ability training system.
[0098] Figure 4 The present invention provides a flowchart of the art design ability training information management method according to the embodiment of the present application, wherein the differential feature graph is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has been improved. Figure 4 As shown, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has been improved, including: S310, expanding each differential feature matrix in the differential feature map into a one-dimensional feature vector according to a row vector or a column vector and then cascading to obtain a classification feature vector; S320, inputting the classification feature vector into the Softmax classification function of the classifier to obtain the probability value of the classification feature vector belonging to each classification label; and, S330, determining the classification label corresponding to the maximum probability value as the classification result.
[0099] In summary, the art design ability training information management method based on the embodiment of the present application is explained, which is based on deep learning artificial intelligence detection technology, and uses a convolutional neural network model with a dual attention mechanism to perform feature mining on the first-year art design assessment painting images and the second-year art design assessment painting images, and capture the implicit difference feature distribution information of the two in the high-dimensional space, so as to accurately evaluate the students' art design ability, make corresponding adjustments to the art design ability training program, and thus optimize the art design ability training system.
[0100] Exemplary Systems
[0101] Figure 5 FIG. 1 is a block diagram of an art design capability training information management system according to an embodiment of the present application. Figure 5As shown, according to the art design ability training information management system 100 of the embodiment of the present application, it includes: an assessment painting acquisition module 110, which is used to obtain the first-year art design assessment painting image and the second-year art design assessment painting image of the student to be evaluated; a spatial enhancement module 120, which is used to obtain the first assessment painting feature map and the second assessment painting feature map by using a convolutional neural network model with a spatial attention mechanism respectively; a feature distribution correction module 130, which is used to perform feature distribution correction along the channel dimension on the first assessment painting feature map and the second assessment painting feature map to obtain the corrected first assessment painting feature map and the corrected second assessment painting feature map; a channel enhancement module 140, which is used to convert the first-year art design assessment painting image and the second-year art design assessment painting image into a convolutional neural network model with a spatial attention mechanism respectively. The image and the second academic year art design assessment painting image are respectively obtained by using a convolutional neural network model with a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map; a multi-dimensional aggregation module 150 is used to fuse the corrected first assessment painting feature map and the third assessment painting feature map to obtain a first multi-dimensional aggregated assessment painting feature map, and to fuse the corrected second assessment painting feature map and the fourth assessment painting feature map to obtain a second multi-dimensional aggregated assessment painting feature map; a difference module 160 is used to calculate a difference feature map between the first multi-dimensional aggregated assessment painting feature map and the second multi-dimensional aggregated assessment painting feature map; and an evaluation result generation module 170 is used to pass the difference feature map through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has improved.
[0102] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the art design ability training information management system 100 have been described in detail above. Figures 1 to 4 The art design ability training information management method has been described in detail, and therefore, its repeated description will be omitted.
[0103] As described above, the art design ability training information management system 100 according to the embodiment of the present application can be implemented in various terminal devices, such as a server for art design ability training information management, etc. In one example, the art design ability training information management system 100 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the art design ability training information management system 100 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the art design ability training information management system 100 can also be one of the many hardware modules of the terminal device.
[0104] Alternatively, in another example, the art design ability training information management system 100 and the terminal device may also be separate devices, and the art design ability training information management system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.
[0105] Exemplary Electronic Devices
[0106] Below, reference Figure 6 To describe an electronic device according to an embodiment of the present application. Figure 6 FIG. 1 is a block diagram of an electronic device according to an embodiment of the present application. Figure 6 As shown, the electronic device 10 includes one or more processors 11 and a memory 12 .
[0107] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
[0108] The memory 12 may include one or more computer program products, and the computer program product 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 (cache), etc. 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 11 may run the program instructions to implement the functions of the art design ability training information management method of each embodiment of the present application described above and / or other desired functions. Various contents such as the first-year art design assessment painting images and the second-year art design assessment painting images of the students to be evaluated may also be stored in the computer-readable storage medium.
[0109] In one example, the electronic device 10 may further include: an input device 13 and an output device 14, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).
[0110] The input device 13 may include, for example, a keyboard, a mouse, etc.
[0111] The output device 14 can output various information to the outside, including classification results, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.
[0112] Of course, to simplify, Figure 6Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 10 may also include any other appropriate components.
[0113] Exemplary computer program products and computer-readable storage media
[0114] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the functions of the art design ability training information management method according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.
[0115] The computer program product may be written in any combination of one or more programming languages to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.
[0116] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps in the functions of the art design ability training information management method according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.
[0117] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.
[0118] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.
[0119] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.
[0120] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.
[0121] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
[0122] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.
Claims
1. A method for managing information on art design ability training, characterized in that: include: Obtaining images of the first-year art design assessment paintings and second-year art design assessment paintings of the students to be assessed; The first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively passed through a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map; Performing feature distribution correction along the channel dimension on the first assessment painting feature map and the second assessment painting feature map to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map; The first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively passed through a convolutional neural network model using a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map; The corrected first assessment painting feature map and the third assessment painting feature map are merged to obtain a first multi-dimensional aggregated assessment painting feature map, and the corrected second assessment painting feature map and the fourth assessment painting feature map are merged to obtain a second multi-dimensional aggregated assessment painting feature map; Calculating a differential feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map; as well as The differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has improved.
2. The method for managing information on art design ability cultivation according to claim 1, characterized in that: The first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively subjected to a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map, including: Using the convolutional coding part of the convolutional neural network model using the spatial attention mechanism, respectively, the first academic year art design assessment painting image and the second academic year art design assessment painting image are subjected to deep convolutional coding to obtain the first academic year initial convolutional feature map and the second academic year initial convolutional feature map; Inputting the initial convolution feature map of the first academic year and the initial convolution feature map of the second academic year into the spatial attention part of the convolution neural network model using the spatial attention mechanism respectively to obtain the spatial attention map of the first academic year and the spatial attention map of the second academic year; The spatial attention map of the first academic year and the spatial attention map of the second academic year are respectively subjected to a Softmax activation function to obtain a spatial attention feature map of the first academic year and a spatial attention feature map of the second academic year; and The first assessment painting feature map is obtained by multiplying the spatial attention feature map of the first academic year and the initial convolution feature map of the first academic year by the position point, and the second assessment painting feature map is obtained by multiplying the spatial attention feature map of the second academic year and the initial convolution feature map of the second academic year by the position point.
3. The method for managing information on art design ability cultivation according to claim 2, characterized in that: Correcting the first assessment painting feature map and the second assessment painting feature map along the channel dimension to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map, including: Performing feature correlation accumulation and differentiation mechanism modeling on the first assessment painting feature graph and the second assessment painting feature graph respectively to obtain a first weighted feature vector and a second weighted feature vector; and The first assessment painting feature map and the second assessment painting feature map are weighted in channel dimension using the first weighted feature vector and the second weighted feature vector respectively to obtain the corrected first assessment painting feature map and the corrected second assessment painting feature map.
4. The method for managing information on art design ability cultivation according to claim 3, characterized in that: Modeling the feature correlation accumulation and differentiation mechanism of the first assessment painting feature graph and the second assessment painting feature graph to obtain a first weighted feature vector and a second weighted feature vector includes: Modeling the feature correlation accumulation and differentiation mechanism of the first assessment painting feature graph using the following formula to obtain the first weighted feature vector; Wherein, the formula is: Where F1 represents the feature map of the first test painting, cov1(·) and cov2(·) represent single-layer convolution operations based on different convolution kernels on the feature map, ReLU(·) represents the ReLU activation function, Sigmoid(·) represents the Sigmoid activation function, and GAP represents the global pooling operation on each feature matrix of the feature map. represents positional addition, V1 represents the first weighted feature vector; Modeling the feature correlation accumulation and differentiation mechanism of the second assessment painting feature graph using the following formula to obtain the second weighted feature vector; Wherein, the formula is: Where F2 represents the feature map of the second test painting, cov1(·) and cov2(·) represent single-layer convolution operations based on different convolution kernels on the feature map, ReLU(·) represents the ReLU activation function, Sigmoid(·) represents the Sigmoid activation function, and GAP represents the global pooling operation on each feature matrix of the feature map. represents positional addition, and V2 represents the first weighted feature vector.
5. The method for managing information on art design ability cultivation according to claim 4, characterized in that: The first academic year art design assessment painting image and the second academic year art design assessment painting image are respectively subjected to a convolutional neural network model using a channel attention mechanism to obtain a third assessment painting feature map and a fourth assessment painting feature map, including: Each layer of the convolutional neural network model using the channel attention mechanism performs the following on the input data in the forward pass of the layer: Performing convolution processing based on a two-dimensional convolution kernel on the input data to generate a convolution feature map; Performing pooling processing on the convolutional feature map to generate a pooled feature map; Performing activation processing on the pooled feature map to generate an activation feature map; Calculating the global mean of each feature matrix along the channel dimension of the activation feature map to obtain a channel feature vector; Calculating the ratio of the eigenvalue of each position in the channel eigenvector to the weighted sum of the eigenvalues of all positions in the channel eigenvector to obtain a channel weighted eigenvector; and Performing a dot multiplication on the feature matrix along the channel dimension of the activation feature map using the eigenvalues of each position of the channel weighted feature vector as weights to obtain a generated feature map; Among them, the generated feature map output by the last layer of the convolutional neural network model using the channel attention mechanism is the third assessment painting feature map or the fourth assessment painting feature map.
6. The method for managing information on art design ability cultivation according to claim 5, characterized in that: The first and third assessment painting feature maps after correction are merged to obtain a first multi-dimensional aggregated assessment painting feature map, and the second and fourth assessment painting feature maps after correction are merged to obtain a second multi-dimensional aggregated assessment painting feature map, including: The corrected first assessment painting feature map and the third assessment painting feature map are fused using the following formula to obtain a first multi-dimensional aggregated assessment painting feature map; Wherein, the formula is: X=Concat[F1 ′ ,F3] Among them, F1 ′ represents the corrected first assessment painting feature map, F3 represents the third assessment painting feature map, Concat[·,·] represents a cascade function, and X represents the first multi-dimensional aggregated assessment painting feature map; The corrected second assessment painting feature map and the fourth assessment painting feature map are fused using the following formula to obtain a second multi-dimensional aggregated assessment painting feature map; Wherein, the formula is: Y=Concat[F2 ′ ,F4] Among them, F2 ′ represents the corrected second assessment painting feature map, F4 represents the four assessment painting feature maps, Concat[·,·] represents the cascade function, and Y represents the second multi-dimensional aggregated assessment painting feature map.
7. The method for managing information on art design ability cultivation according to claim 6, characterized in that: Calculating a differential feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map includes: The differential feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map is calculated using the following formula; Wherein, the formula is: Among them, X represents the first multi-dimensional aggregate assessment painting feature map, Y represents the second multi-dimensional aggregate assessment painting feature map, and Z represents the difference feature map. Indicates difference by position.
8. The method for managing information on art design ability cultivation according to claim 7, characterized in that: The differential feature map is passed through a classifier to obtain a classification result, which is used to indicate whether the art design ability of the student to be evaluated has improved, including: Expanding each differential feature matrix in the differential feature map into a one-dimensional feature vector according to a row vector or a column vector and then concatenating them to obtain a classification feature vector; Inputting the classification feature vector into the Softmax classification function of the classifier to obtain the probability value of the classification feature vector belonging to each classification label; and The classification label corresponding to the maximum probability value is determined as the classification result.
9. An art design ability training information management system, characterized in that: include: An assessment painting acquisition module is used to acquire the first-year art design assessment painting images and the second-year art design assessment painting images of the students to be assessed; A spatial enhancement module, for respectively applying the first academic year art design assessment painting image and the second academic year art design assessment painting image to a convolutional neural network model using a spatial attention mechanism to obtain a first assessment painting feature map and a second assessment painting feature map; a feature distribution correction module, configured to perform feature distribution correction on the first assessment painting feature map and the second assessment painting feature map along a channel dimension to obtain a corrected first assessment painting feature map and a corrected second assessment painting feature map; A channel enhancement module, used for respectively obtaining a third assessment painting feature map and a fourth assessment painting feature map by using a convolutional neural network model with a channel attention mechanism for the first academic year art design assessment painting image and the second academic year art design assessment painting image; a multi-dimensional aggregation module, configured to fuse the corrected first assessment painting feature map and the third assessment painting feature map to obtain a first multi-dimensional aggregated assessment painting feature map, and to fuse the corrected second assessment painting feature map and the fourth assessment painting feature map to obtain a second multi-dimensional aggregated assessment painting feature map; A difference module, used to calculate a difference feature map between the first multi-dimensional aggregate assessment painting feature map and the second multi-dimensional aggregate assessment painting feature map; as well as The evaluation result generation module is used to pass the differential feature map through a classifier to obtain a classification result, and the classification result is used to indicate whether the art design ability of the student to be evaluated has improved.
10. The art design ability training information management system according to claim 9, characterized in that: The evaluation result generating module comprises: An expansion unit, used for expanding each differential feature matrix in the differential feature map into a one-dimensional feature vector according to a row vector or a column vector and then cascading to obtain a classification feature vector; A probabilization unit, used for inputting the classification feature vector into the Softmax classification function of the classifier to obtain a probability value of the classification feature vector belonging to each classification label; and The classification result determination unit is used to determine the classification label corresponding to the maximum probability value as the classification result.
Citation Information
Patent Citations
Universal image aesthetic evaluation method, system and equipment based on attention mechanism
CN109886317A
Transportation junction security check image recognition method based on autonomous learning
CN110287773A
Artistic design ability cultivation information management method and system and storage medium
CN111444234A
And evaluation method is used for evaluating learning efficiency of students and improving and optimizing learning efficiency
CN111738285A
SAR image change detection method based on multi-scale differential feature attention mechanism
CN114926746A
