KR102294741B1 - Method for psychological test based on picture using deep learning and server using the same - Google Patents

Abstract

The present invention relates to a picture-based psychological test method through deep learning and to enabling non-face-to-face picture-based psychological analysis and solutions. The picture-based psychological test method comprises the following steps: extracting component data from a picture to be analyzed; labeling the extracted component data; extracting and repeatedly labeling the component data by augmenting and repeatedly learning the analysis the picture to be analyzed; and deriving psychological diagnosis results according to the picture to be analyzed based on the extracted component data.

Description

Picture-based psychological test method and server through deep learning

The present invention relates to a picture-based psychological examination method and server through deep learning, and to enable picture-based psychological analysis and solutions non-face-to-face.

Recently, as the number of patients with mental illness increases in society, the demand for psychological counseling is increasing.

However, in the case of psychological counseling that is being conducted in general, there are prejudices against psychological counseling, fear of self-opening, burden of counseling costs, or geographical and temporal limitations.

Accordingly, there is a growing demand for a method that enables online psychological counseling and solutions without a direct visit to a counseling center for psychological counseling, and a method for solving the above-mentioned problems is urgently needed.
For reference, as a prior art document related to the present invention, Korean Patent Publication No. 10-1772987 (2017.08.31.) exists.

The present invention aims to solve the problems of the prior art described above, and to enable non-face-to-face psychological counseling and solution provision by performing picture-based psychological examination using deep learning technology.

Objects of the present invention are not limited to the objects mentioned above, and other objects not mentioned will be clearly understood from the description below.

A picture-based psychological examination method through deep learning according to an embodiment of the present invention for solving the above problem includes extracting component data from an analysis target picture; labeling the extracted component data; extracting and repeatedly labeling the component data by augmenting and repeatedly learning the data to be analyzed; and deriving a psychological diagnosis result according to the analysis target picture based on the extracted component data.

According to an embodiment of the present invention, the component data extraction step may extract characteristics of a house, a tree, and a person as the component data.

According to an embodiment of the present invention, in the component data extraction step, the characteristic of the house includes at least one or more detailed characteristics of a roof, a wall, a door, a window, a chimney, and an access road, and the characteristic of the tree is a subject, At least one detailed characteristic of a column, branch, root, leaf, and bark may be included, and the human characteristic may include at least one or more detailed characteristic of a head, face, mouth, lips, chin, eyes, and arm.

According to an embodiment of the present invention, the labeling step may include a file name, a feature class, a detailed feature class, and maximum and minimum values of x and y coordinates for labeling.

According to an embodiment of the present invention, the data augmentation, repeated learning, and repeated labeling step may augment data by inverting, rotating, cropping, or changing the size of the analysis target picture.

According to an embodiment of the present invention, the data augmentation, repeated learning, and repeated labeling step is performed by inverting, rotating, cropping, and changing the size of the ROI (Region Of Interest) in the analysis target picture based on the labeled components at least extracting one or more; machine learning the ROI and extracting component data; and labeling the component data extracted from the ROI.

According to an embodiment of the present invention, the step of deriving the psychological diagnosis result may derive the diagnosis result according to the class, size, ratio, state, and location of the extracted component.

According to the present invention, by performing a picture-based psychological examination using deep learning technology, it may be possible to provide non-face-to-face psychological counseling and solutions.

On the other hand, the effects of the present invention are not limited to the above-mentioned effects, and various effects may be included within the range apparent to those skilled in the art from the contents to be described below.

1 is a diagram showing a picture-based psychological examination server through deep learning according to an embodiment of the present invention.
2 is a diagram illustrating a picture-based psychological test result sheet through deep learning according to an embodiment of the present invention.
3 is a diagram illustrating setting values for main items and detailed features of a figure according to an embodiment of the present invention.
4 is a diagram illustrating a process of providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
5 is a diagram illustrating a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
6 is a diagram illustrating an embodiment of labeling in a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
7 is a diagram illustrating an embodiment of data augmentation in a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
8 is a diagram illustrating another embodiment of data augmentation in a method of providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
9 is a diagram illustrating an embodiment of modeling among a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
10 is a diagram illustrating an embodiment of Faster R-CNN-based modeling among a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
11 is a diagram illustrating a modeling process in a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
12 is a diagram illustrating an example of a re-learning result among a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.
13 is a diagram illustrating a picture-based psychological test result through deep learning according to an embodiment of the present invention.

Hereinafter, a 'picture-based psychological examination method and server through deep learning' according to the present invention will be described in detail with reference to the accompanying drawings. The described embodiments are provided so that those skilled in the art can easily understand the technical spirit of the present invention, and the present invention is not limited thereto. In addition, matters expressed in the accompanying drawings may be different from the forms actually implemented in the drawings schematically for easy explanation of the embodiments of the present invention.

On the other hand, each component represented below is only an example for implementing the present invention. Accordingly, other components may be used in other implementations of the present invention without departing from the spirit and scope of the present invention.

In addition, each component may be implemented purely by a configuration of hardware or software, or may be implemented by a combination of various hardware and software configurations that perform the same function. In addition, two or more components may be implemented together by one piece of hardware or software.

In addition, the expression 'including' certain components merely refers to the existence of the corresponding components as an expression of 'open type', and should not be construed as excluding additional components.

1 is a diagram showing a picture-based psychological examination server through deep learning according to an embodiment of the present invention.

First, the communication unit 110 of the psychological examination server 100 may be implemented with various communication technologies. That is, Wi-Fi (WIFI), WCDMA (Wideband CDMA), HSDPA (High Speed Downlink Packet Access), HSUPA (High Speed Uplink Packet Access), HSPA (High Speed Packet Access), Mobile WiMAX (Mobile WiMAX), WiBro (WiBro) , LTE (Long Term Evolution), 5G, Bluetooth, infrared data association (IrDA), NFC (Near Field Communication), Zigbee, wireless LAN technology, etc. can be applied. In addition, when a service is provided by being connected to the Internet, TCP/IP, which is a standard protocol for information transmission on the Internet, may be followed.

In addition, the database 130 of the psychological test server 100 has information on various psychological test results, which will be described later along with the operation of the processor 120 .

Information can be transmitted and received with the database 130, the user's terminal 200, the administrator's terminal 300, and the like through the communication unit 110, and the user's terminal 200 and the administrator's terminal 300 are desktop computers, notebook computers, and the like. A computer, workstation, PDA, web pad, mobile phone, smart remote control, various IOT main devices, etc. are equipped with a memory means while performing communication, equipped with a microprocessor to have computing power, and a camera equipped with a marker such as a QR code Any digital device capable of recognizing , may correspond to the user's terminal 200 and the administrator's terminal 300 according to the present invention.

The psychological test progress of the present invention and the report provided with the psychological test result may be checked through the user's terminal 200 , and the selected picture may be analyzed through the manager's terminal 300 .

The processor 120 of the picture-based psychological examination server 100 through deep learning according to an embodiment of the present invention collects image data, pre-processes the image data, learns an image, and extracts from the learned result. By analyzing the characteristics of the components, it is possible to interpret the psychological diagnosis results and provide a report on the interpretation contents.

The picture-based psychological examination server 100 through deep learning according to an embodiment of the present invention may include a computer or a system using a Convolutional Neural Network (CNN) using artificial intelligence, and is analyzed through image machine learning It is possible to derive psychological analysis results by extracting components from the target picture and analyzing its characteristics.

2 is a diagram illustrating a picture-based psychological test result sheet through deep learning according to an embodiment of the present invention, and FIG. 3 is a set value for main items and detailed features of a picture according to an embodiment of the present invention. It is an exemplified drawing.

First, the processor 120 of the psychological examination server 100 of the present invention may obtain an analysis target picture. In a state in which a plurality of pictures are stored in the database 130 , the processor 120 may transmit them to the user's terminal 200 to request the user to select at least one picture. The processor 120 may set at least one picture input to the user's terminal 200 as an analysis target picture, analyze the analysis target picture through AI analysis, and transmit it to the user's terminal 200 . The analyzed result may be transmitted to the manager's terminal 300 to allow the manager to review the analysis result.

In addition, the processor 120 may display the analysis result in the form of a report on the user's terminal 260 .

2 and 3 , the picture-based psychological examination method through deep learning according to an embodiment of the present invention may extract component data from an analysis target picture. In the picture-based psychological examination method through the deep learning, the processor 120 may extract the characteristics of a house, a tree, and a person as the component data. The characteristic of the house may include at least one or more detailed characteristics of a roof, a wall, a door, a window, a chimney, and a driveway, and the characteristic of the tree may include at least one or more of a subject, a column, a branch, a root, a leaf, and a bark. It may include a characteristic, and the characteristic of the person may include at least one or more detailed characteristics of a head, face, mouth, lips, chin, eyes, and arm. In the drawings, for the convenience of the private description, it may be configured so that detailed characteristics can be identified centering on the characteristics of a person.

In addition, in the picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 may derive a diagnosis result according to the class, size, ratio, state and location of the extracted components. The class, size, proportion, state, and position of the components may be absolute or relative. In addition, the processor 120 may calculate a score for the psychological state through the above components. According to an embodiment, the processor 120 divides the score out of 10 points into four zones, a first zone (ex 0 points to 3 points), a second zone (ex 4 points to 5 points), and a third zone It can be divided into (ex 6~7 points) and the 4th area (ex 8~10 points).

As a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 is interested in intellectual ability, intellectual ambition, achievement desire, expanded ego, when the size or ratio of the head is large. It may be associated with psychological states of introversion, daydreaming, hyperidealization (paranoia), aggressive tendencies, regression, inhibition, dependence, immaturity, and anxiety.

The present invention is to diagnose human psychology through pictures, and a picture with a large head is related to the activeness and unconsciousness caused by daydreaming, and may mean satisfaction with the intellectual. Conversely, small heads are associated with obsessive-compulsive or neuroticism; hiding hands may reflect aggressive and hostile emotions; clenched hands may reflect repressed aggression. Clothing buttons, pocket accents, etc. can mean compensation for dependence and deficits. As described above, when it is determined that there are some problems related to the mental state, the processor 120 may calculate a low psychological state-related score.

Specifically, based on the overall size of the picture to be analyzed, if the size of the head is greater than or equal to the preset ratio (ex 5%), the size of the head is considered large, and if it is less than the preset ratio (ex 5%), the size of the head is considered small. can see.

In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 determines that the head feature is a disproportionately large head if it is 4 or more heads, and a small head if it is 8 or less heads. In between, it can be judged as balanced, when there is no head, it can be judged that the head is omitted, and when the line is thin and indistinct (light gray line), it can be judged as a blurred hair. If it is not a head shape that anyone can recognize, it can be judged as a deformed head, and when it is not on the neck, it can be judged as a head that is not in position. In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 additionally adds the features of the head to the picture in which the head is drawn the latest, the picture in which the head of the opposite sex is drawn larger, and the adult male is female. It is possible to determine the case in which the large head of ) will be able to set the psychological state-related score lower than the preset score (ex the score corresponding to the first zone or the second zone).

In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 determines that the size or ratio of the head is small (intellectual, social, sexual) of inadequacy, incompetence, and inferiority. It may be related to the state of mind. Therefore, the processor 120 may give a score corresponding to the third zone (ex 6 to 7 points) for the psychological state related score when the head is large in the picture but the proportion is unbalanced, and the head is small in the picture If the ratio is balanced, a score corresponding to the second zone (ex 4 to 5 points) can be given for the psychological state related score.

In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 is unpleasant when the features of the head are omitted, when the head is blurred, or when it is determined that the head is a deformed head It can be associated with the psychological states of the desire to get rid of thoughts, R/O neurosis, and R/O psychosis. Accordingly, the processor 120 may give a score corresponding to the first zone (ex 0 to 3 points) for the psychological state related score. In particular, when it is determined that there is no hair of a person determined to be a woman in the picture, the processor 120 considers the possibility of a problem in the user's interpersonal relationship and determines the psychological state-related score in the first zone (ex 0 to 3 points). points can be awarded.

In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 determines that the characteristic of the house is an excessively large roof when the roof area is larger than the wall area, and the roof is the picture size If it is more than 1/2 of the figure, it is judged as a picture with the entire roof. If the roof is less than 1/2 of the picture size and smaller than the wall area, it is judged to correspond to the average roof area. It can be judged by a falling wall. In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 shows that the characteristics of the house are a one-dimensional roof, a roof line emphasis, a ground line emphasis, a wall that is not properly connected, and only one wall is drawn. cases can be distinguished.

Specifically, when it is determined that the roof area is excessively large, the processor 120 may give a score corresponding to the first zone for the psychological state-related score, and when it is determined that the roof is the whole picture, the psychological state-related score A score corresponding to the second zone can be given to the second zone for the If the area and wall are clean, a score corresponding to the fourth zone may be given for the score related to the psychological state.

In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 determines that the tree is inclined to the left when the centerline of the tree is to the left of 2/3 of the width, If the width of the center line of the tree is 2/3 to the right, it can be determined that it is inclined to the right. In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 is characterized in that the tree is an apple tree, a willow tree with drooping branches, a Christmas tree, a dead tree, a huge tree, and a hilltop. It can be distinguished as a lone tree, a tree that leans to the left, a tree that leans to the right, and a tree that looks like a keyhole.

Specifically, when it is determined that the center line of the tree is tilted to the left or right, the processor 120 may give a score corresponding to the first zone or the second zone for the psychological state related score, and the center line of the tree is tilted. If it is judged not to lose, a score corresponding to the third zone or the fourth zone may be given to the psychological state-related score.

When at least two or more of a person, a house, and a tree exist at the same time in the picture to be analyzed, the processor 120 gives the greatest weight to the psychological state score related to the person, and then assigns the weight in the order of the house and the tree. can After all, when a psychological state score related to a person is x, a mental state score related to a house is y, and a mental state score related to a tree is z, the final psychological state score may be determined by the following formula.

Formula) Ax + By + Cz = Mental state score (however, A > B > C, A, B, C are natural numbers)

4 is a diagram illustrating a process of providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.

4, in the service providing process according to the picture-based psychological examination method through deep learning according to an embodiment of the present invention, when the interviewee draws a picture for psychological diagnosis, the picture is requested as an analysis target picture, Using the picture-based psychological examination method through deep learning according to an embodiment of the invention, it is possible to confirm the psychological diagnosis result through AI-based picture analysis, and to proceed with counseling and treatment.

In addition, the processor 120 of the picture-based psychological examination method through deep learning according to an embodiment of the present invention recognizes main features and detailed features of people, houses, trees, etc., and confirms the analysis results according to the features do. The analysis target picture may be a picture including at least one of a person, a house, and a tree.

5 is a diagram illustrating a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.

Referring to FIG. 5 , as a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 is a system using artificial intelligence referring to papers, Google, etc. data can be extracted.

In addition, by preprocessing it, data such as labeling, boxing, and coordinates can be extracted and the data can be propagated.

As a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 can extract and learn the features and detailed features of the components from the analysis target picture using Faster R-CNN. And through data multiplication, iterative learning and feature extraction can increase accuracy and label.

In a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 analyzes the analysis results based on the keywords of the features and detailed features of the components, the coordinate value distance calculation, and the location of the diagnosis results. can be derived

6 is a diagram illustrating an embodiment of labeling in a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention.

Referring to FIG. 6 , in a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 may label the extracted component data. In the picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 may augment the data for the analysis target picture and repeatedly learn to extract the component data and repeatedly label it.

Specifically, as can be seen in FIG. 6 , the processor 120 determines the file name, width, length, and category (eg, person_eye, person_nose, house_wall, house_roof) for the component data extracted from the corresponding figure. , tree_root, tree_leaf, etc.) and center coordinates (ex x, y, z) can be calculated while labeling.

7 is a diagram illustrating an embodiment of data augmentation among a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention, and FIG. 8 is a view illustrating deep learning according to an embodiment of the present invention. It is a diagram showing another embodiment of data augmentation among the method of providing a picture-based psychological examination service through the

7 and 8, the data augmentation of the picture-based psychological examination method through deep learning according to an embodiment of the present invention can augment data by inverting, rotating, cropping, and changing the size of the analysis target picture. have. Data augmentation of the picture-based psychological examination method through deep learning according to an embodiment of the present invention is data by inverting, rotating, cropping, changing the size and ratio of the analysis target picture around the characteristics or detailed characteristics of the components. can be augmented. Data augmentation of the picture-based psychological examination method through deep learning according to an embodiment of the present invention generates a partial image including at least one or more characteristics or detailed characteristics of the components of the analysis target picture, and the partial image You can augment data by flipping, rotating, cropping, changing the size and scale.

On the other hand, CNN (Convolutional Neural Network) is a type of Deep Neural Network (DNN), and includes one or several convolutional layers, a pooling layer, and a fully connected layer. ) is a neural network composed of

The CNN has a structure suitable for learning two-dimensional data, and can be trained through a backpropagation algorithm. It is one of the representative models of DNN that is widely used in various application fields such as object classification in images and object detection.

As a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 may extract features of the image using the CNN. The extracted features may be classified based on a neural network.

9 is a diagram illustrating an embodiment of modeling among a method for providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention, and FIG. 10 is a view showing an embodiment of the present invention through deep learning It is a diagram showing an embodiment of Faster R-CNN-based modeling among the picture-based psychological examination service providing methods.

9 and 10 , the middle Faster R-CNN may include a Region Proposal Network (RPN), and the RPN is a Fully Convolution Network (FCN) that simultaneously predicts the presence or absence of an object's bounding box and an object. In addition, learning is carried out in the end-to-end method emphasized in Fast R-CNN. Furthermore, RPN and Detection Network (Fast R-CNN detector) shared a convolution feature and combined into a single network.

The Faster R-CNN may be composed of two modules. One is a Region Proposal Network consisting of a Fully Convolutional Network that performs Region Proposal, and the other is a Fast R-CNN detector that detects objects using the proposed region that has undergone the previous process.

In addition, the attention mechanism can be used in the process of merging the output of RPN and the result of Fast R-CNN. The attention mechanism is a technique to increase the performance of the model by making the deep learning model pay attention to a specific vector. The RPN module can tell where the Fast R-CNN detector module should look.

The RPN may receive an image of an arbitrary size as an input and output a score for whether the object exists at the position or not, and an offset of the bounding box.

The RPN may receive a feature map output by the convolution layer as an input. A sliding window method that slides an n × n spatial window on the input feature map can be used to generate a region proposal.

In this process, a region proposal is performed for each anchor, and as a result, a feature map of the same size as the original feature map is created. Feature output through this sliding window is input to two Fully Connected Layers (1 × 1 Convolution Layer), a box regression layer and a box classification layer.

The output of the box classification layer is W × H × k (number of anchor) × 2 (number of score: object / non-object), and the output of the box regression layer is W × H × k × 4 (box offset: x , y, w, h) are obtained.

Multiple region proposals are predicted at the same time in each sliding window, and the maximum number of possible proposals at each position (number of anchor boxes × aspect ratio) is called k. Therefore, the regression layer can have 4k outputs corresponding to the coordinates (x, y, w, h) of k bounding boxes, and the classification layer can have 2k outputs that predict the probability of whether each proposal is an object or not.

For k proposals, there is basically a box consisting of a predetermined size and aspect ratio, which is called an anchor. According to an embodiment of the present invention, nine anchors can be used for each sliding position using three sizes and three aspect ratios. For a feature map of size W × H, there are a total of W × H × k anchors.

11 is a diagram illustrating a modeling process among a method of providing a picture-based psychological examination service through deep learning according to an embodiment of the present invention, and FIG. 12 is a picture-based method through deep learning according to an embodiment of the present invention. It is a diagram illustrating an embodiment of a re-learning result among a method of providing a psychological examination service, and FIG. 13 is a diagram illustrating a picture-based psychological examination result through deep learning according to an embodiment of the present invention.

11 to 13 , in a picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 diagnoses according to the class, size, ratio, state and location of the extracted components. results can be derived.

For reference, component data (ex person (head), house, tree, etc.) was extracted from the picture to be analyzed, and psychological state-related scores could be extracted from this. The extracted psychological state-related score may be included in the diagnosis result and derived together.

In the picture-based psychological examination method through deep learning according to an embodiment of the present invention, the processor 120 may extract the features of the components extracted by augmenting and repeatedly learning data from the analysis target picture. In the picture-based psychological examination method through the deep learning, the processor 120 may derive a diagnosis result according to the class, size, ratio, state and location with reference to the labeled data.

According to an embodiment of the present invention, in the picture-based psychological examination method through the deep learning, the processor 120 may recognize the characteristics and detailed characteristics of a person, and derive a psychological diagnosis result according to the state. For example, having outstretched arms can be judged by the distance between the torso and arms, and having arms outstretched can be associated with outward-facing aggression or demands in the environment or in interpersonal contact while in need of help or affection. . For example, a fisted hand may be determined when a characteristic of a finger is not extracted from a hand position, and may be associated with up-pressured aggression. Therefore, according to an embodiment of the present invention, if you open your arms and clench your fists, you can be related to the tendency to like to meet people while waiting for help or affection from the people around you or the repressed things can not be expressed and can appear as aggression. have.

The diagnosis result may include an affection deficit-related index, an aggression index, and the like. If the distance between the shoulder and the hand is more than a certain percentage (15%) of the total horizontal length (a) based on the overall horizontal length (a) of the person in the figure, it can be determined that the arms are spread out, and the processor 120 ) can be calculated as the user's affection deficiency-related index is higher than the reference value.

In addition, if the finger of a person is not visible in the picture (if the length of the finger is '0'), the processor 120 considers that the person is clenching a fist in the picture and calculates that the user's aggression index is higher than the reference value. will be able

For reference, the overall process of the present invention may be summarized as follows.

First, the processor 120 may collect an image to build a data set, in which case the image may include both a live-action image or a picture image.

Next, the processor 120 may define a label for a partial area on the image (picture) as a pre-processing process. For example, a label can be defined for human eyes, a human hand, a wall of a house, a tree trunk, etc. In detail, a label can be defined with a hidden arm or a missing leg after reflecting the state or behavior. . In this case, it is also possible to calculate a psychological state-related score together with the label definition.

In addition, the processor 120 may increase the accuracy of the label process and the psychological state prediction (mental state related score) by learning the images included in the data set through an algorithm such as Faster R-CNN, and repeat the learning. By doing this, the error rate can be reduced. At this time, when the learning is repeated, the data can be augmented (ex vertical inversion, left and right inversion, etc.) to learn a larger amount of data to increase the accuracy.

Through the module on which the learning is completed, the processor 120 may extract component data, diagnose the user's psychological state from the image through label work, etc., and derive the diagnosis result as a report. The report may include a score related to a psychological state, an index related to lack of affection, an aggression index, and the like.

The embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include hard disks, magnetic media such as floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floppy disks. media), and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform processing according to the present invention, and vice versa.

In the above, the present invention has been described with specific matters such as specific components and limited embodiments and drawings, but these are provided to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments. , those of ordinary skill in the art to which the present invention pertains can devise various modifications and variations from these descriptions.

Therefore, the spirit of the present invention should not be limited to the above-described embodiments, and not only the claims described below, but also all modifications equivalently or equivalently to the claims described below belong to the scope of the spirit of the present invention. will do it

100: psychological test server
110: communication department
120: processor
130: database
200: user's terminal
300: administrator's terminal

Claims (7)

In a method for a psychological test server to perform a picture-based psychological test through deep learning,
(a) obtaining a picture to be analyzed;
(b) extracting component data from the analysis target picture;
(c) labeling the extracted component data;
(d) data augmentation and repeated learning of the analysis target picture to extract the component data and repeat labeling;
(e) deriving a psychological diagnosis result according to the analysis target picture based on the extracted component data; and
(f) calculating a psychological state score using the component data,
The component data may include detailed features of at least one of a roof, a wall, a door, a window, a chimney, and a driveway, a detailed feature of a tree of at least one of a subject, a column, a branch, a root, a leaf, and a bark, and a head; contains detailed characteristics of a person of at least one of face, mouth, lips, chin, eyes, and arms;
In the step (d), a partial image including at least one of detailed characteristics of the house, detailed characteristics of the tree, and detailed characteristics of the person among the component data is generated, and the partial image is inverted and rotated , further comprising the step of augmenting the data by cropping, changing the size, and changing the ratio,
The psychological state score is calculated by summing the scores for the detailed features of the person, the scores for the detailed features of the house, and the scores for the detailed features of the tree,
In the psychological state score, the weight of the score for the detailed characteristic of the person is greater than the weight of the score for the detailed characteristic of the house,
In the psychological state score, the weight of the score for the detailed feature of the house is greater than the weight of the score for the detailed feature of the tree, a picture-based psychological test method through deep learning.
delete delete According to claim 1,
The labeling step is
A picture-based psychological examination method through deep learning, characterized in that the labeling includes the file name, the class of features, the class of detailed features, and the maximum and minimum values of x and y coordinates.
According to claim 1,
The data augmentation, iterative learning and repeating labeling steps include:
Picture-based psychological examination method through deep learning, characterized in that the data is augmented by inverting, rotating, cropping, and changing the size of the picture to be analyzed.
6. The method of claim 5,
The data augmentation, iterative learning and repeating labeling steps include:
extracting at least one by inverting, rotating, cropping, or changing the size of an ROI (Region Of Interest) in the analysis target picture based on the labeled component;
machine learning the ROI and extracting component data; and
A picture-based psychological examination method through deep learning comprising; labeling the component data extracted from the ROI.
According to claim 1,
The psychological diagnosis result derivation step is,
A picture-based psychological examination method through deep learning, characterized in that the diagnosis results are derived according to the class, size, ratio, state and location of the extracted components.

Images