Image assessment system and image assessment method

Abstract

To provide an image determination system and the like that determine presence / absence of a blur and / or out-of-focus of an image of an animal using a simple method.SOLUTION: An image determination system comprises: acceptance means for accepting input of an image of an animal; and determination means for determining presence / absence of a blur and / or out-of-focus of the image of the animal from the image of the animal input to the acceptance means using a learned model. The learned model is obtained by learning relation between images of animals and presence / absence of a blur and / or out-of-focus of the images of the animals.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an image assessment system and an image assessment method, and more particularly to an image assessment system and an image assessment method that provide an assessment result regarding blurring and / or out-of-focus of an animal image from an animal image. [Background technology]

[0002] Pets such as dogs, cats, and rabbits, as well as livestock such as cows and pigs, are irreplaceable to humans. With the spread of smartphones and digital cameras, opportunities to take photos of animals are also increasing.

[0003] Unlike humans, animals are covered in hair all over their bodies, and their facial and body shapes are also different from humans. This can make it difficult to focus when photographing animals. Furthermore, unless you are a highly trained photographer, animals tend to move while you are taking a photograph, which can lead to blurring. Such blurring and out-of-focus images are often not easily noticeable, and can only be discovered when you enlarge the image long after it has been taken.

[0004] Meanwhile, pet insurance, which covers medical expenses for pets, is becoming increasingly popular. When offering pet insurance, it is necessary to distinguish between insured and non-insured pets. For this reason, applicants are sometimes asked to submit an image of the pet when applying for insurance. The submitted image of the pet is used to determine whether or not the policyholder is eligible for pet insurance and is also used on the pet insurance card. For example, to provide a pet insurance card to a policyholder and enable the policyholder to receive insurance payments by presenting the card at a veterinary clinic, the card typically includes a photo of the pet, allowing the policyholder to verify that the pet listed on the card is the same as the pet that actually received medical treatment. In such cases, blurry or out-of-focus photos of the pet may interfere with the pet insurance application screening process and the determination of identity on the card.

[0005] Patent Document 1 discloses a system for improving the quality of selfies, the system including a selfie quality index (SQI) module that analyzes a normalized selfie for quality metrics selected from occlusion, blur, distance to the camera, facial expression, lighting, and combinations thereof, and the system generates an SQI score using the analyzed image quality metrics, in particular, the SQI module includes a blur detection module in the form of a CNN trained with a blur feature vector formed from at least one of blur coefficients and features extracted from the last layer of an Alexnet CNN.

[0006] However, the system for improving the quality of selfies described in Patent Document 1 is a system for improving the image quality when a person takes a photo of themselves, a so-called selfie, and is not intended for images of animals. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Patent No. 6849824 Summary of the Invention [Problem to be solved by the invention]

[0008] Therefore, an object of the present invention is to provide an image assessment system and an image assessment method that provide assessment results regarding blurring and / or out-of-focus of an animal image using a simple method. [Means for solving the problem]

[0009] As a result of intensive research into solving the above problem, the inventors discovered that the above problem could be solved by a trained model that had learned the relationship between images of animals and blur and / or out-of-focus images of those animals, and thus completed the present invention.

[0010] That is, the present invention relates to the following [1] to

[13] . [1] An image judgment system comprising: a receiving means for receiving an input of an image of an animal; and a judgment means for using a trained model to judge whether or not the image of the animal input to the receiving means is blurred and / or out of focus, An image judgment system characterized in that the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus. [2] An image judgment system comprising: a receiving means for receiving an input of an image of an animal; and a judgment means for using a trained model to judge whether or not the image of the animal is blurred and / or out of focus from the image of the animal input to the receiving means, An image judgment system characterized in that the trained model is trained using images of animals and labels regarding whether the images of the animals are blurred and / or out of focus as training data, and the input is an image of an animal and the output is a judgment regarding whether the images of the animal are blurred and / or out of focus. [3] The image judgment system according to [1] or [2], wherein the image of the animal accepted by the accepting means is an image of the animal's face taken from the front. [4] The image judgment system according to [3], wherein the face image of the animal accepted by the accepting means is an image including the eyes, nose, and ears of the animal. [5] The image judgment system according to [4], wherein the facial image of the animal accepted by the accepting means is a facial image for individual identification. [6] An animal insurance subscription system, an acceptance means for accepting input of an image of an animal submitted by an applicant for animal insurance; a determination means for determining whether or not the image of the animal input to the receiving means is blurred and / or out of focus using a trained model; output means for outputting a determination result by the determination means; An animal insurance system characterized in that the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus. [7] An animal insurance subscription system, an acceptance means for accepting input of an image of an animal submitted by an applicant for animal insurance; a determination means for determining whether or not the image of the animal input to the receiving means is blurred and / or out of focus using a trained model; output means for outputting a determination result by the determination means; An animal insurance system characterized in that the trained model is trained using images of animals and labels regarding whether the images of the animals are blurred and / or out of focus as training data, and the input is an image of an animal and the output is a determination of whether the image of the animal is blurred and / or out of focus. [8] An animal insurance system according to [6] or [7], wherein if the judgment result by the judgment means is that the image of the animal is blurred and / or out of focus, the output means outputs the judgment result together with a message urging the user to submit another image of the animal. [9] An animal insurance application system according to any one of [6] to [8], wherein the determination means determines that the image of the animal input to the insurance card image receiving means contains out-of-focus images if the entire face of the animal is not in focus.

[10] A method for generating a trained model that determines whether an image of an animal is blurred and / or out of focus from an image of the animal, the method comprising inputting an image of the animal and a label indicating whether the image of the animal is blurred and / or out of focus into a computer as training data, and having an artificial intelligence learn the trained model.

[11] preparing an image of an animal; An image judgment method comprising: inputting the image into a trained model; and outputting a judgment on whether or not the image of the animal is blurred and / or out of focus from the input image of the animal using the trained model; An image assessment method characterized in that the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus.

[12] preparing an image of an animal; An image judgment method comprising: inputting the image into a trained model; and outputting a judgment on whether or not the image of the animal is blurred and / or out of focus from the input image of the animal using the trained model; An image judgment method characterized in that the trained model is trained using images of animals and labels indicating whether the images of the animals are blurred and / or out of focus as training data, and the input is an image of an animal and the output is a judgment as to whether the images of the animal are blurred and / or out of focus. [Effects of the Invention]

[0011] The present invention makes it possible to provide an image assessment system and an image assessment method that provide, in a simple manner, assessment results on blur and / or out-of-focus of an animal image from an animal image. It also makes it possible to provide an animal insurance application system that uses an animal image by utilizing the ability to easily determine whether an animal image is blurred and / or out-of-focus. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 10 is a diagram showing an example of a suitable animal image. [Figure 2] FIG. 10 is a diagram showing an example of a suitable animal image. [Figure 3] This is an example image that does not contain any blur or out-of-focus. [Figure 4] 1 is an example image that includes blur or out of focus. [Figure 5]1 is a schematic diagram illustrating a configuration of an embodiment of an image assessment system according to the present invention. [Figure 6] FIG. 1 is a flowchart illustrating an example of a flow of image assessment by the image assessment system of the present invention. [Figure 7] FIG. 1 is a flowchart showing an example of the flow of image assessment by the animal insurance application system of the present invention. [Figure 8] FIG. 1 is a flowchart showing an example of the flow of image assessment by the animal insurance application system of the present invention. [Figure 9] 10 is an example of an image used in the embodiment. [Figure 10] 10 is an example of an image used in the embodiment. [Figure 11] This is an example of an animal image. [Figure 12] (A) An example image where the line is drawn, and (B) an example image where the line is not drawn. [Figure 13] (A) An example image where the cut surfaces are aligned, and (B) an example image where the cut surfaces are not aligned. [Figure 14] (A) An example image where the left and right sides are balanced, and (B) an example image where the left and right sides are not balanced. DETAILED DESCRIPTION OF THE INVENTION

[0013] <Image Judgment System> The image judgment system of this embodiment is an image judgment system that includes a reception means for receiving input of an image of an animal, and a judgment means for using a trained model to judge whether the image of the animal input to the reception means is blurred and / or out of focus, and is characterized in that the trained model is a trained model that has learned the relationship between the image of the animal and whether the image of the animal is blurred and / or out of focus.

[0014] [Method of reception] The receiving means is a means for receiving input of an image of an animal. Examples of animals include mammals such as dogs, cats, and rabbits, birds, and reptiles, with dogs and cats being preferred. The image may be received by any method, such as scanning, inputting image data, transmitting, or capturing an image on the spot. The image format is not particularly limited. While the parts of the animal depicted in the image are not particularly limited, an image showing the animal's face is preferred, more preferably a photograph of the animal's face taken from the front as in Figure 1, and even more preferably a photograph showing a large image of the animal's face. Furthermore, an image showing the animal's face up to the ears, as in Figure 11(A), is particularly preferred, rather than an image cropped to show only the muzzle area as in the circular frame in Figure 11(B) or only the eyes as in the square frame in Figure 11(C). Examples of such photographs include those used on human driver's licenses. Images such as those used on animal health insurance cards, as in Figure 2, are also preferred. The image may be black and white, grayscale, or color. Images that do not show the animal's entire face, images whose shape has been edited using image editing software, images that show multiple animals, images where the face is so small that the eyes and ears cannot be distinguished, or images that are unclear are not acceptable. Images that have been normalized or have a uniform resolution are preferred.

[0015] [Judgment means] The determination means is means for determining whether or not an image of an animal input to the receiving means is blurred and / or out of focus using a trained model. In this embodiment, the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus.

[0016] The trained model described above can be generated, for example, by supervised learning or unsupervised learning. In the case of supervised learning, examples of training data include images of animals and data on whether the images of the animals are blurred and / or out of focus. Images of animals may be tagged with tags indicating whether the images of the animals are blurred and / or out of focus, and used as training data.

[0017] Blurred and / or out-of-focus images refer to both or either blur caused by camera shake or subject movement, and out-of-focus images caused by being out of focus or out of focus. Blurred or out-of-focus images can make it difficult to identify individual animals in the images. Additionally, an image of an animal's face in which the nose is in focus but the depth of field is shallow, leaving other parts of the face, such as the eyes and ears, out of focus, can also be set as out of focus. Animals, particularly dogs, have protruding noses that are often separated from the rest of their face. When photographed with a shallow depth of field, certain parts of the face may be in focus but other parts may be out of focus. In such cases, the photo is not suitable for individual identification, and can be tagged as an out-of-focus image. On the other hand, if the eyes are in focus but the nose is out of focus, this poses little problem for individual identification, and the image can be set and tagged as not including out-of-focus (not out-of-focus).

[0018] The format of the animal images used as training data is not particularly limited. For example, images used for individual identification or application for animal health insurance may be images that primarily show the face or images that show the entire body. Therefore, while the parts of the animal depicted in the image are not particularly limited, images that show the animal's face are preferred, more preferably photographs of the animal's face taken from the front, and even more preferably photographs that show the animal's face enlarged, as in Figure 1. Furthermore, images that show the animal's face up to the ears, as in Figure 11(A), are particularly preferred, rather than images cropped to show only the muzzle area, as in the circular frame in Figure 11(B), or only the eyes, as in the square frame in Figure 11(C). Examples of such photographs include those used on human driver's licenses. Images such as those used on animal health insurance cards, as in Figure 2, may also be used. The images may be black and white, grayscale, or color. Images that do not show the entire animal's face, images whose shape has been edited using image editing software, images that show multiple animals, images in which the face is so small that the eyes and ears cannot be distinguished, or images that are unclear are not preferred. Images that have been normalized or have a uniform resolution are preferred. It is more preferable that the resolution of the images used in the training data and the images accepted by the accepting means be uniform.

[0019] Artificial intelligence (AI) is preferred as a trained model. Artificial intelligence (AI) refers to software or systems that mimic the intellectual tasks performed by the human brain, specifically computer programs that can understand natural language used by humans, perform logical inference, and learn from experience. AI can be either general-purpose or specialized, and can be any of deep neural networks, convolutional neural networks, etc., and publicly available software can be used.

[0020] To generate a trained model, artificial intelligence is trained. Either machine learning or deep learning can be used for the training, but deep learning is preferred. Deep learning is an advanced version of machine learning, and is characterized by its ability to automatically find features.

[0021] The training method for generating a trained model is not particularly limited, and publicly available software can be used. For example, DIGITS (the Deep Learning GPU Training System) published by NVIDIA can be used. Alternatively, training may be performed using a known support vector machine method, such as that published in "Introduction to Support Vector Machines" (Kyoritsu Shuppan).

[0022] The image assessment system and image assessment method of the present invention may include, in addition to a judging means for judging whether an image is blurred and / or out of focus, a trimming result judging means for judging the quality of trimming and scoring of the animal in the image. The trimming result judging means is a judging means for judging the quality of trimming of an animal based on an image of the animal input to the receiving means using a trained model, where the trained model is a trained model that has learned the relationship between the image of the trimmed animal and the quality of the trimming of the animal. Specifically, the trained model is preferably a trained model that uses trained images of the trimmed animal and labels indicating the quality of the trimming of the animal as training data, and inputs the image of the animal and outputs a quality of trimming of the animal. The trimming judgment is preferably performed after the image judgment (judging whether the image is blurred or out of focus), and outputs the image judgment together with the quality of the trimming and a scoring result. The trimming result judging means is configured, for example, similarly to the judging means (trained model) 11 of the embodiment described below. The quality of trimming refers to, for example, whether the trimming is good or bad. Specific examples of the criteria for this include whether the trimming is symmetrical, whether the outline is clear, whether each part is well-defined, and whether the cut surfaces are even. In all of the examples listed here, the former are good examples, and the latter are bad examples. Other criteria include whether the trimming is three-dimensional, whether the cuts are well-balanced, and whether the connections between flat surfaces are smooth and even. Trimming criteria can also be based on, for example, the criteria for trimming competitions established by the All Japan Association of Animal Professional Education. Figure 12 shows an example of (A) an image with even lines and (B) an image with inconsistent lines. Figure 13 shows an example of (A) an image with even cut surfaces and (B) an image with inconsistent cut surfaces. FIG. 14 shows (A) an example image where the left and right sides are balanced, and (B) an example image where the left and right sides are not balanced.

[0023] The animal images may be scored for each individual criterion, or may be scored based on a combination of these criteria, or may be judged as "good" or "bad." That is, each image used as training data may be tagged with a score according to the above criteria, or tagged according to whether it is good or bad, and used as training data. The judgement of whether the trimming is good or bad may be made, for example, by an expert in trimming or by someone with knowledge of scoring methods used in trimming competitions.

[0024] [output] When the determination means receives an image of an animal as input information, it determines whether the image of the animal is blurred and / or out of focus using the trained model. The output format is not particularly limited, and the judgment can be output, for example, by displaying on a computer screen the quality of the image, such as "very good" or "not good," depending on whether or not there is blur or out-of-focus, or by indicating the presence of blur or out-of-focus, such as "This image is blurry" or "This image is out of focus," or by displaying the presence or absence of blur or out-of-focus, along with the probability and confidence level. The image assessment system of the present invention may also have a separate output means for receiving the assessment result from the assessment means and outputting the assessment result.

[0025] Furthermore, if the image of the animal is blurred and / or out of focus, the system may output a message urging the user to submit another image of the animal along with the output of the determination result. Examples of such messages include "This image is blurry, please take another photo" or "This image is out of focus, please upload another image."

[0026] Furthermore, if the image of the animal is blurred and / or out of focus, the system may output a message with advice on how to take the photo along with the output of the determination result. Examples of such messages include "This image is blurred. Please take the photo in a well-lit area," "This image is blurred. Please calm your dog," "Please change the camera settings to increase the shutter speed," "Please stabilize the camera," and "This image is out of focus. Please change the camera settings to increase the depth of field."

[0027] Hereinafter, an embodiment of the image assessment system of the present invention will be described with reference to FIG.

[0028] In FIG. 5, terminal 40 is a terminal used by a user. Examples of users include a pet owner who takes a photo of their pet and uploads it to a social networking site, a hair salon operator who takes a photo of a customer's pet after a haircut and uploads it to a website, and a person who applies for animal health insurance. Examples of terminal 40 include a personal computer, a smartphone, and a tablet terminal. Terminal 40 includes a processing unit such as a CPU, a storage unit such as a hard disk, ROM, or RAM, a display unit such as an LCD panel, an input unit such as a mouse, keyboard, or touch panel, and a communication unit such as a network adapter.

[0029] A user accesses the server from terminal 40 and inputs and transmits an image (photograph) of an animal. If necessary, information such as the animal's species, breed, sex, and weight may also be input. Alternatively, when using the image assessment system, the user may take a photo of the target animal using a smartphone camera and input and transmit the photo. For example, the user follows instructions displayed on the screen of terminal 40 to take a photo of the target animal and transmit it to the server. At this time, a determination means within the server determines whether the photo is blurred or out of focus, and outputs the result. The server may also be configured to include a separate photography assistance means consisting of an image assessment program, which determines whether the photo is suitable, depending on the purpose of the photo, such as whether the entire face or body of the animal is captured, or whether the photo is taken from the front of the animal's face, and communicates the determination result, including whether the photo is blurred or out of focus, to the user via an interface or terminal. Furthermore, the user can receive the result of the determination as to whether or not there is any blur or out-of-focus by having the terminal 40 access the server.

[0030] In this embodiment, the server is configured by a computer, but may be any device as long as it has the functions according to the present invention. The server may also be a server on a cloud.

[0031] The storage unit 10 is configured with, for example, a ROM, a RAM, a hard disk, etc. The storage unit 10 stores an information processing program for operating each unit of the server, and in particular, stores a determination means (trained model) 11.

[0032] The determination means (trained model) 11 receives an image of a target animal input by a user and outputs a determination as to whether the image is blurred and / or out of focus. The determination means (trained model) 11 in this embodiment is configured to include, for example, a deep neural network or a convolutional neural network.

[0033] The processing calculation unit 20 uses the determination means (trained model) 11 stored in the storage unit to determine whether or not the image is blurred and / or out of focus.

[0034] The interface unit (communication unit) 30 has a receiving means 31 and an output means 32, and receives images of animals and other information from the user's terminal, and outputs the results of a determination as to whether the image is blurred and / or out of focus to the user's terminal.

[0035] With the image assessment system of this embodiment, a user can easily obtain a judgment as to whether or not the image of the pet is blurred and / or out of focus by uploading a photo or video of the pet to a server.

[0036] In this embodiment, the determination means and reception means are stored in a server and connected to the user's terminal via a connection means such as the Internet or a LAN, but the present invention is not limited to this, and may also be in a form in which the determination means, reception means, and interface unit are stored in a single server or device, or in which a separate terminal is not required for use by the user.

[0037] In addition, the image judgment system of the present invention may be configured as an app or software in which software encoding an acceptance means and an judgment means including a trained model is downloaded to a user's terminal, and image acceptance, judgment, and output of judgment results are performed consistently within the user's terminal.

[0038] A flowchart of image judgment based on one embodiment of the image judgment system of the present invention is shown in Figure 6. A user uploads an image of a target animal to the reception means (step S1). The processing and calculation unit of the server uses the judgment means (trained model) to judge whether the image of the animal is blurred and / or out of focus from the uploaded image (step S2). The output means outputs the derived judgment result, for example by displaying it on a screen, and presents it to the user (step S3).

[0039] <Other embodiments> Another embodiment of the present invention is an image judgment system comprising a reception means for receiving input of an image of an animal, and a judgment means for using a trained model to judge whether the image of the animal input to the reception means is blurred and / or out of focus, wherein the trained model is trained using the image of the animal and a label regarding whether the image of the animal is blurred and / or out of focus as training data, and the input is an image of the animal, and the output is a judgment regarding whether the image of the animal is blurred and / or out of focus. This embodiment is characterized in that the trained model is trained using images of animals and labels related to blurring and out-of-focus of the images of the animals as training data, and the input is an image of the animal and the output is a determination of whether the image of the animal is blurred and / or out-of-focus.Other points are the same as those described above.

[0040] <Image evaluation method> The image judgment method of the present invention is an image judgment method comprising the steps of preparing an image of an animal, inputting the image into a trained model, and having a computer use the trained model to output a judgment on whether the input image of the animal is blurred and / or out of focus, wherein the trained model is a trained model that has learned the relationship between an image of an animal and whether the image of the animal is blurred and / or out of focus.

[0041] Another aspect of the image judgment method of the present invention is an image judgment method comprising the steps of preparing an image of an animal, inputting the image into a trained model, and having a computer use the trained model to output a judgment as to whether the input image of the animal is blurred and / or out of focus, wherein the trained model is trained using an image of the animal and a label indicating whether the image of the animal is blurred and / or out of focus as training data, and the input is an image of the animal and the output is a judgment as to whether the image of the animal is blurred and / or out of focus.

[0042] The animal images and trained models are the same as those of the image judgment system of the present invention described above.

[0043] <Insurance enrollment system> The animal insurance system of the present invention comprises a reception means for receiving input of an animal image submitted by an animal insurance applicant, a judgment means for judging whether the animal image input to the reception means is blurred and / or out of focus using a trained model, and an output means for outputting the judgment result by the judgment means, wherein the trained model is a trained model that has learned the relationship between the animal image and whether the animal image is blurred and / or out of focus. Another aspect of the present invention provides an animal insurance system, which includes: a reception unit that receives an input of an image of an animal submitted by an applicant for animal insurance; and a determination unit that determines whether the image of the animal input to the reception unit is blurred and / or out of focus using a trained model. output means for outputting a determination result by the determination means; The trained model is characterized in that it is trained using images of animals and labels relating to the presence or absence of blur and / or out-of-focus of the images of the animals as training data, and the input is an image of the animal and the output is a determination of the presence or absence of blur and / or out-of-focus of the image of the animal.

[0044] The preferred animal insurance to be covered is pet health insurance.

[0045] The receiving means, determination means, and output means are the same as those in the image determination system of the present invention. The receiving means may have a function to receive, along with the image of the animal to be insured, information necessary for applying for insurance, such as the animal's name, age, medical history, type, breed, sex, type of insurance desired, name, address, and credit card information of the applicant.

[0046] In the animal insurance system of the present invention, if the judgment result by the judgment means is that the image of the animal is blurred and / or out of focus, it is preferable that the output means outputs a message urging the user to submit another image of the animal along with the judgment result.

[0047] In the animal insurance application system of the present invention, it is preferable that the determination means determines that the image of the animal input to the insurance card image receiving means contains out-of-focus images if the entire face of the animal is not in focus.

[0048] 7 and 8 show flowcharts for applying for insurance based on one embodiment of the animal insurance system of the present invention. The flowchart in Figure 7 is a flowchart for when the input image is not blurred or out of focus. A user who wishes to purchase animal insurance enters necessary information about the animal to be insured, such as the name, type, breed, weight, sex, date of birth, whether or not there are other insurance policies, the selected insurance plan, and payment method, into an input form on the website provided by the animal insurance management company, and uploads an image of the animal on the website or sends it separately by email, whereupon the image of the animal is accepted by the acceptance means (step S1). The processing and calculation unit of the server uses the judgment means (trained model) to judge whether the uploaded image of the animal is blurred and / or out of focus (step S2). If the judgment result is that the image is good and not blurred or out of focus, a message to that effect is output (step S3), and the website displays a message such as "The image has been accepted. Please proceed to the next step," urging the user to proceed with the insurance application procedure (step S4).

[0049] The flowchart in Figure 8 illustrates a case where the input image is blurred or out of focus. A user who wishes to purchase animal insurance enters necessary information about the animal to be insured, such as the name, type, breed, weight, sex, date of birth, whether or not there are other insurance policies, the selected insurance plan, and payment method, into an input form on a website provided by the animal insurance management company. The user then uploads an image of the animal to the website or sends it separately by email, and the image is accepted by the acceptance means (step S1). The processing and calculation unit of the server uses the judgment means (trained model) to judge whether the uploaded image of the animal is blurred and / or out of focus (step S2). If the judgment result in step S2 indicates that the image is blurred or out of focus, a message to that effect is output (step S3), and a message is displayed on the website urging the user to submit another image, such as, "The image you uploaded is blurred. We apologize for the inconvenience, but please upload another image." (step S4). In this case, the user uploads another image of the animal on the website or sends it separately by email, and the image of the animal is accepted by the accepting means (step S1). Steps S1 and onwards are the same as above. If an image that is not blurred or out of focus is not input, steps S1 to S4 will be repeated. [Example]

[0050] [Example 1] We prepared 632 photos of dogs that were not blurred or out of focus (photos with a shallow depth of field, in which the eyes were in focus but the nose was out of focus were considered not out of focus. On the other hand, photos in which the dog was out of focus and the background was in focus were considered to be out of focus. Examples are shown in Figures 3(A) to (D)), as well as 632 photos of dogs that were blurred (either due to camera shake or subject blur) or out of focus (examples are shown in Figures 4(A) to (D)). These photos of dogs included photos of the face or the entire body. Using these facial photographs, deep learning was performed to generate a trained model. Each photo was tagged with a "1" if it was not blurred or out of focus (hereinafter also referred to as "good"), and a "0" if it was blurred or out of focus (hereinafter also referred to as "bad"), and used for learning. The learning method was transfer learning using Resnet as the artificial intelligence (neural network), and Pytorch was used as the machine learning library (deep learning library). In addition, 50 good and 50 bad photos were used as data for evaluation during the learning process. The evaluation results during the learning process are shown in Table 1.

[0051] [Table 1]

[0052] Next, using the trained model, we conducted a test using 106 good and 94 bad photos that were different from the training and validation data. The accuracy rate was 90.0%. The results are shown in Table 2.

[0053] [Table 2]

[0054] [Example 2] The good and bad photos used as training data in Example 1 above were each trimmed to include the dog's eyes and nose, but not the ears (examples of good photos are shown in Figures 9(A) to (D), and examples of bad photos are shown in Figures 10(A) to (D)). Using these facial photographs, deep learning was performed to generate a trained model. Each image was tagged with a "1" if it was a good photo and a "0" if it was a bad photo and used for learning. The learning method was transfer learning using Resnet as the artificial intelligence (neural network), and Pytorch was used as the machine learning library (deep learning library). In addition, 50 good and 50 bad photos were used as data for evaluation during the learning process. The evaluation results during the learning process are shown in Table 3.

[0055] [Table 3]

[0056] Next, using the trained model, we conducted a test using 106 good and 94 bad photos, which were different from the training and validation data. The accuracy rate was 85.1%. The results are shown in Table 4.

[0057] [Table 4]

Claims

1. An image judgment system comprising: a receiving means for receiving input of an image for identifying an animal as a subject; and a judgment means for using a trained model to judge whether or not the image of the animal as a subject input to the receiving means is blurred and / or out of focus, An image judgment system characterized in that the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus.

2. An image judgment system as described in claim 1, wherein the facial image for identifying the subject animal is a facial image for individual identification and a facial image for determining identity with the subject animal.

3. The trained model is trained using an image of an animal and a label regarding whether the image of the animal is blurred and / or out of focus as training data, and the input is an image of the animal. The image judgment system described in claim 1 is a trained model in which the input is an image of the animal and the output is a judgment regarding whether the image of the animal is blurred and / or out of focus.

4. 3. The image judgment system according to claim 1, wherein the image for identifying the subject animal accepted by the accepting means is an image of the animal's face taken from the front.

5. 5. The image judgment system according to claim 4, wherein the face image for identifying the subject animal accepted by the accepting means is an image including the eyes, nose, and ears of the animal.

6. An animal insurance subscription system, an acceptance means for accepting input of an image of an animal submitted by an applicant for animal insurance; a determining means for determining whether the image of the animal input to the receiving means is blurred and / or out of focus using a trained model; An animal insurance system characterized in that the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus.

7. An animal insurance subscription system, an acceptance means for accepting input of an image of an animal submitted by an applicant for animal insurance; a determining means for determining whether the image of the animal input to the receiving means is blurred and / or out of focus using a trained model; output means for outputting a determination result by the determination means; An animal insurance system characterized in that the trained model is trained using images of animals and labels regarding whether the images of the animals are blurred and / or out of focus as training data, and the input is an image of an animal and the output is a determination of whether the image of the animal is blurred and / or out of focus.

8. An animal insurance system as described in claim 6 or 7, wherein if the judgment result by the judgment means is that the image of the animal is blurred and / or out of focus, the output means outputs the judgment result along with a message prompting the user to submit another image of the animal.

9. An animal insurance system according to any one of claims 6 to 8, wherein the determination means determines that an image of an animal input to the reception means contains out-of-focus images if the entire face of the animal is not in focus.

10. A method of photographing an animal, comprising: preparing an image for identifying the animal; An image judgment method comprising: inputting an image for identifying the subject animal into a trained model; and outputting a judgment on whether or not the image of the animal is blurred and / or out of focus from the input image of the animal using the trained model; An image assessment method characterized in that the trained model is a trained model that has learned the relationship between an image of an animal and whether or not the image of the animal is blurred and / or out of focus.

11. The image judgment method described in claim 10, characterized in that the trained model is trained using images of animals and labels regarding whether or not the images of the animals are blurred and / or out of focus as training data, and the input is an image of an animal and the output is a judgment regarding whether or not the images of the animal are blurred and / or out of focus.

Images