Method and apparatus for age estimation
By segmenting the clothes images from human images and inputting them into neural network models, the problem of traditional age estimation methods relying on human faces or human bodies is solved, and efficient age estimation is achieved in the absence of these features.
Patent Information
- Application Number
- CN202280097872.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-14
- Publication Date
- 2025-02-18
AI Technical Summary
Traditional age estimation methods rely on detecting faces or uncovered human bodies, and cannot perform age estimation without these characteristics.
By taking images of humans, cropping out body parts, segmenting out clothes images, removing background pixels, and inputting clothes images into neural network models for age estimation.
Age estimation is achieved without a human face or an uncovered human body, improving the flexibility and accuracy of age estimation.
Smart Images
Figure CN119487558A_ABST
Abstract
Description
Technical Field
[0001] The present invention generally relates to computer technology. For example, the present invention relates to image processing. In another example, the present invention relates to a method and apparatus for performing age estimation. Background Art
[0002] Human and / or face recognition is widely used in computer vision. For example, with the help of machine learning techniques, human bodies and / or faces can be detected from images or videos. Based on the detected human bodies and / or faces, the age of the person can be estimated. Various machine learning-based technical solutions (e.g., neural networks) have been used to perform age estimation based on detected human bodies and / or faces. The estimated age can be used in various application scenarios, such as providing censored search results, filtering child abuse content, and restricting the provision of services to minors (underage). Summary of the invention
[0003] Traditional age estimation is based on faces and / or uncovered human bodies. When faces or uncovered human bodies cannot be detected due to occlusion or blockage, age estimation cannot be performed because there is no valid input.
[0004] In light of the above, there is a need to improve age estimation methods.
[0005] The present invention aims to facilitate age estimation even when no face or uncovered human body is detected in an image of a human being.The present invention aims to perform age estimation based on clothing.
[0006] These and other objects are achieved by the subject matter of the independent claims. Further implementations are apparent from the dependent claims, the description and the drawings.
[0007] A first aspect of the present invention provides a method for age estimation. The method is performed by a device and comprises the following steps:
[0008] - Acquire images of humans;
[0009] -cropping one or more body parts from the image;
[0010] - segmenting clothes from the cropped one or more body parts to obtain a clothes image;
[0011] - removing background pixels from the clothing image;
[0012] - providing the clothing image as input to a neural network model;
[0013] -obtaining an output of the neural network model, wherein the output of the neural network model indicates an age category of the human being.
[0014] Optionally, the image of the human being can be a photograph or a frame of a video. The image (or video) can be acquired by a camera of the above-mentioned device. Alternatively, the image can be acquired by the above-mentioned device from an external source (e.g., via a communication link).
[0015] Optionally, the one or more body parts may include one or more of a human torso and limbs. The limbs may include one or more of a human arm and leg. The one or more body parts cut out may be completely covered by clothing or partially covered by clothing.
[0016] Optionally, the segmented clothes may be understood as clothing, and may also include ornaments worn by humans.
[0017] Optionally, the age category of humans may include two or more age groups. For example, the two or more age groups may include minors (under 18 years old) and adults (18 years old or older). Alternatively, the two or more age groups may include children, adolescence / teen, adults, and the elderly.
[0018] The method provided in the first aspect can efficiently perform age estimation based on clothing. Thus, even when no face or uncovered human body is detected in the human image, the method can conveniently perform age estimation.
[0019] In an implementation manner of the first aspect, the neural network model may be based on an 18-layer residual neural network (ResNet-18) model.
[0020] Optionally, the neural network model can share a similar architecture of a ResNet-18 model. The architecture of ResNet-18 can provide satisfactory performance and very low error rate.
[0021] In an implementation of the first aspect, the neural network model may include three convolutional layers and three feature refinement blocks (FRBs). The step of obtaining the output of the neural network model may include:
[0022] - extracting three feature representations of the clothing image from the three outputs of the three convolutional layers;
[0023] - providing the three feature representations as inputs to the three feature refinement blocks to obtain three refined features from the three outputs of the three feature refinement blocks.
[0024] Optionally, each FRB can be used to compress and recalibrate the extracted features. Each FRB is used to learn a weight vector from different block features during the training phase. The weight vector is used as an attention vector to recalibrate the features of the output of the three convolutional layers during the inference phase. In this way, features useful for small object classification can be selected. FRB can finally be connected with global max pooling to capture global context information and obtain a compressed feature vector.
[0025] In an implementation manner of the first aspect, the method may further include: cascading the three refined features with an output of a first convolutional layer among the three convolutional layers to obtain a final feature representation of the clothing image.
[0026] Optionally, the three refined features can be first element-wise multiplied and then concatenated.
[0027] In an implementation of the first aspect, the neural network model may further include a fully-connected (FC) layer. The method may further include:
[0028] - providing the final feature representation of the clothing as input to the fully connected layer;
[0029] -Obtaining the output of the fully connected layer as the output of the neural network model.
[0030] In an implementation manner of the first aspect, the image of the human being may not include the face of the human being.
[0031] Alternatively, the human face may not be fully visible. Alternatively, the image quality of the human face may be poor, so age estimation cannot be performed based on the face.
[0032] A second aspect of the present invention provides a device for age estimation. The device is used for:
[0033] - obtaining an image of a human being; cropping one or more body parts from the image;
[0034] - segmenting clothes from the cropped one or more body parts to obtain a clothes image;
[0035] - removing background pixels from the clothing image;
[0036] - providing the clothing image as input to a neural network model implemented in the device;
[0037] -obtaining an output of the neural network model, wherein the output of the neural network model indicates an age category of the human being.
[0038] In an implementation manner of the second aspect, the neural network model may be based on a ResNet-18 model.
[0039] In an implementation manner of the second aspect, the neural network model may include three convolutional layers and three feature refinement blocks, and in order to obtain the output of the neural network model, the device is used to:
[0040] - extracting three feature representations of the clothing image from the three outputs of the three convolutional layers;
[0041] - providing the three feature representations as inputs to the three feature refinement blocks to obtain three refined features from the three outputs of the three feature refinement blocks.
[0042] In an implementation manner of the second aspect, the device may also be configured to cascade the three refined features with an output of a first convolutional layer among the three convolutional layers to obtain a final feature representation of the clothing image.
[0043] In an implementation manner of the second aspect, the neural network model may include a fully connected layer, and the device may further be used for:
[0044] - providing the final feature representation of the clothing as input to the fully connected layer;
[0045] -Obtaining the output of the fully connected layer as the output of the neural network model.
[0046] In an implementation manner of the second aspect, the image of the human being may not include the face of the human being.
[0047] The device provided by the second aspect can efficiently perform age estimation based on clothing. Thus, even when no face or uncovered human body is detected in an image of a person, the device can still perform age estimation.
[0048] A third aspect of the present invention provides a computer program including program code, wherein the program code is used to execute the method according to the first aspect or any implementation thereof.
[0049] A fourth aspect of the present invention provides a non-transitory storage medium storing executable program code. When the executable program code is executed by a processor, the method according to the first aspect or any implementation thereof is performed.
[0050] A tenth aspect of the present invention provides a chipset including a memory and a processor, wherein the memory and the processor are used to store and execute program codes to execute the method according to the first aspect or any implementation manner thereof.
[0051] It should be noted that all devices, elements, units and components described in this application can be implemented in software or hardware elements or any type of combination thereof. All steps performed by various entities described in this application and the functions described to be performed by various entities are intended to indicate that the corresponding entities are suitable for or used to perform the corresponding steps and functions. Although in the description of the following specific embodiments, the specific functions or steps performed by the external entity are not reflected in the description of the specific detailed elements of the entity performing the specific steps or functions, it should be clear to the technician that these methods and functions can be implemented by corresponding software or hardware elements or any combination thereof. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In conjunction with the accompanying drawings, the following description of specific embodiments will illustrate the above aspects and various implementations. In the drawings:
[0053] Figure 1 An example of the device provided by the present invention is shown;
[0054] Figure 2 An example of a process for age estimation provided by the present invention is shown;
[0055] Figure 3 A schematic diagram of a neural network model provided by the present invention is shown;
[0056] Figure 4 A schematic diagram of the process provided by the present invention is shown;
[0057] Figure 5 An example of an application scenario provided by the present invention is shown;
[0058] Figure 6 An example of filtering search results provided by the present invention is shown;
[0059] Figure 7 An example of other application scenarios provided by the present invention is shown;
[0060] Figure 8 An example of age verification by a vending machine provided by the present invention is shown;
[0061] Fig. 9 A schematic diagram of a method for age estimation provided by the present invention is shown;
[0062] Fig.10 An example of a neural network model provided by the present invention is shown. DETAILED DESCRIPTION
[0063] The present invention generally relates to clothing-based human age estimation, which can be performed without a face or body. Given an image or video that includes clothing, it is possible to roughly estimate the age (or at least the age group) of the person appearing in the image or video. In order to better estimate a person's age based on clothing, it is important to analyze the characteristics of the clothes and the context of the clothing.
[0064] Figure 1 An example of the device 100 provided by the present invention is shown. The device 100 is used to obtain an image 110 of a human (or person). In the present invention, the image 110 of a human may be referred to as a human image 110. Optionally, the human image 110 may be a frame of a video. Optionally, the human image 110 may be understood as a picture including at least one human. Optionally, the device 100 may include a camera ( Figure 1 ). Alternatively, the device 100 may receive the human image 110 from another device.
[0065] The device 100 is used to crop one or more body parts from the human image 110. The device 100 is used to segment clothes from the cropped one or more body parts to obtain a clothes image 130. Optionally, the one or more body parts may include one or more of a human torso and limbs (e.g., arms and legs). The segmented clothes may include one or more of clothes worn by a person, headwear and jewelry worn by the person. The device 100 is used to remove background pixels from the acquired clothes image 130. Thereby reducing noise and improving the accuracy of age estimation. Optionally, the device 100 may include an image processing module 120 for acquiring the clothes image 130 based on the human image 110. Any method well known in the field of image processing that can be used to segment clothes from the human image 110 can be used to acquire the clothes image 130.
[0066] The device 100 is used to provide the clothing image 130 as an input to the neural network model 140, and obtain the output 150 of the neural network model 140. The neural network model 140 can be simply referred to as the neural network 140. The output 150 of the neural network model 140 indicates the age category of the human being. The neural network model 140 can be included in the device 100 as a machine learning model, and the machine learning model is used to predict (or estimate) the age category of the human being based on the acquired clothing image 130. The age category can be understood as an age group, and does not necessarily indicate the exact age of the human being.
[0067] Optionally, the age categories may include at least two groups. For example, the age categories may include adults (e.g., adults) and minors (e.g., minors). In most cases, the age of adulthood may be 18 years old, which can be adjusted according to different regions and nationalities. Optionally, the age categories may be further refined to include more detailed groups. For another example, the age categories may include a children's group (e.g., 0-12 years old), teenagers (e.g., 13-17 years old), adults (e.g., 18-59 years old), and the elderly (e.g., over 60 years old). It should be noted that the age ranges given in the above examples are for reference only and can be adjusted according to different purposes, application scenarios and / or needs.
[0068] Optionally, the neural network model 140 can be trained through a training phase so that the trained neural network model 140 is suitable for age estimation based on clothing images. In order to train the neural network 140, a certain number (e.g., thousands) of clothing images can be randomly collected and labeled as a training data set according to the desired age category. Different people may participate in labeling the collected clothing images to eliminate bias as much as possible. Any training technique known in the field of machine learning can be used to train the neural network 140.
[0069] Optionally, the neural network model 140 may include a feature extraction model 142 and a classification model 144. The feature extraction model 142 may be used to extract features based on the clothing image 130. Optionally, the feature extraction model 142 may include three convolutional layers and three FRBs. The three convolutional layers may be used to extract three feature representations of the clothing image. The three feature representations may be provided as input to the three FRBs to obtain three refined features. The three refined features may be processed and fed as input to the classification model 144. For example, the classification model may be used to classify the processed features into corresponding age categories via the FC layer through max pooling.
[0070] Optionally, the neural network model 140 can be based on a ResNet-18 model. A residual network (ResNet) is a convolutional neural network (CNN) with the lowest error rate, almost equal to the human error rate. ResNet can learn through network depth, ranging from a small model with 18 layers to a complex model with 152 layers (for example, the number of layers is 18, 34, 50, 101, 152). The number of layers represents the total number of weight layers in ResNet. In the present invention, a ResNet with 18 layers can be used as the basis for constructing the neural network model provided by the present invention.
[0071] Device 100 can be used in various application scenarios. For example, device 100 can be a mobile device that serves as a content distribution terminal. The mobile device can receive distributed content, such as streaming music and videos, from a content provider. Traditionally, device 100 can require the user to manually enter their age for verification. However, there is no method to control whether the manually entered age information is true. Therefore, some explicit content or content limited to a specific age group may be improperly accessed. According to the present invention, device 100 can be used to capture a user image to verify age.
[0072] Similarly, for another example, the device 100 may be a mobile device that is a game console. Due to some regulations in certain regions, games may have different age ratings, or the gaming time may be limited to a specific age group. According to the present invention, the device 100 may be used to capture a user's image before and / or during a video (or mobile) game to verify age.
[0073] Figure 2 An example of a process for age estimation provided by the present invention is shown. The process can be applied to Figure 1 The device 100 in FIG. 100 may have similar elements and features. Figure 2 The arrows between the elements indicate the flow of information between the elements.
[0074] like Figure 2 As shown, image processing 220 is applied to the input image 210 to obtain a clothing image. The clothing image is then provided as input to a convolutional neural network model 240. The convolutional neural network 240 may include a feature extraction portion 242 and a classification portion 244 as a discard layer. The convolutional neural network 240 may be constructed according to a ResNet-18 model. For example, the convolutional neural network 240 may include a portion of the ResNet-18 model that is located before the res_conv3_1 layer and serves as a backbone network. The convolutional neural network 240 may also include (e.g., in the feature extraction portion 242) three different branches and three FRBs that are located after the res_conv3_1 layer of the ResNet-18 model. Three feature representations (or features) of clothing appearing in the clothing image , and It can be extracted from the three outputs of three different branches after the res_conv3_1 layer of the ResNet-18 model. It should be noted that the res_conv3_1 layer of the ResNet-18 model can refer to a layer of the ResNet 18 model, which is used to perform 3×1 convolution and the feature map dimension is fixed to 256. These three different branches can include three independent convolution layers, each of which is used to extract features. Then, the three FRBs can represent these three features respectively. , and Processing is performed to obtain three refined features , and .
[0075] Optionally, each FRB can be used to extract features related to the characteristics of the clothes and the background of the clothes for better prediction. This is because the characteristics of the clothes and the background are less affected by regional, weather, indoor / outdoor and cultural differences.
[0076] Optionally, each FRB can be used to compress and recalibrate the extracted feature representations. Each FRB can be used to learn a weight vector from different block features during the training phase. The weight vector can be used as an attention vector to recalibrate the features of the outputs of the three different branches during the inference phase. In this way, features useful for small object classification can be selected.
[0077] Optionally, these three refined features can be element-wise multiplied and then concatenated with the output of the first convolutional layer to obtain the final feature representation of the clothing image243.
[0078] The final feature representation 243 may be provided as an input to a classification part 244 including a FC layer to obtain an output 250 of the FC layer. The output 250 represents the age category of the human being.
[0079] Figure 3 A schematic diagram of a neural network model provided by the present invention is shown. Figure 3 The arrows between the elements indicate the flow of information between the elements. Figure 3 The neural network model in can be built based on the ResNet-18 model. The neural network model can include a part of the ResNet-18 model as the backbone network, for example, the part before the res_conv3_1 layer (or conv3_1 for short) of the ResNet-18 model. Then, after the res_conv3_1 layer, the neural network model can also include three different branches, each for extracting features. , and The extracted features can be combined with Figure 2 The three FRBs mentioned above are treated similarly. Figure 3 The neural network model shown can be applied to Figure 1 and Figure 2 .
[0080] Figure 4 A schematic diagram of a process 400 provided by the present invention is shown. The process 400 is as follows.
[0081] Step 401: Obtain an input image (or video).
[0082] Step 402: Detect human bodies from the input image.
[0083] Step 403: After the human body is detected, it is checked whether a face is detected. It should be noted that a face can be regarded as a part of the human body. Therefore, general human body detection is performed.
[0084] Step 404: If a face is detected, predict the age based on the face.
[0085] Step 405: If no face is detected, predict the age based on the body (including body parts other than the face).
[0086] Step 406: After step 402, clothing is detected based on the human body.
[0087] Step 407: Predict age based on clothing.
[0088] It should be noted that the present invention combines Figures 1 to 3 The various aspects mentioned can be applied to step 406 and step 407. It can be seen that the age estimation based on clothing provided by the present invention can be performed regardless of whether a face is detected. Therefore, the age estimation based on clothing provided by the present invention can be used as an independent method for predicting age, or as an alternative method for predicting age when a face cannot be detected, or as other (e.g., verification, anti-counterfeiting) methods other than predicting age based on face and / or body.
[0089] Figure 5 An example of an application scenario provided by the present invention is shown.
[0090] This application scenario illustrates a safe content search method 500 that can filter child abuse content from search results.
[0091] The method 500 comprises the following steps:
[0092] Step 501: A user searches for unrestricted content on a search engine.
[0093] Unrestricted Content may sometimes include inappropriate content, such as adult content, child abuse images and / or videos.
[0094] Step 502: The search engine performs age prediction on the image of the search result to determine the age group of the person in the image.
[0095] Step 503: Based on the determined age group, the search engine determines whether there is any child abuse content in the search results.
[0096] Step 504: If there is child abuse content, hide the corresponding image from the search results.
[0097] It should be noted that the present invention combines Figures 1 to 4 Various aspects provided may be applied to step 503. It can be seen that the present invention may be particularly useful for filtering content that abuses children (eg, child pornography).
[0098] Figure 6 An example of filtering search results provided by the present invention is shown. Figure 5 It can be seen that several inappropriate images have been filtered out in the final search results.
[0099] Figure 7 An example of other application scenarios provided by the present invention is shown. Figure 7 The left side of the figure shows a vending machine. The right side shows the flow of the method steps performed for the vending machine.
[0100] Vending machines are used to estimate the age of the purchaser to allow or deny the sale. For example, if the vending machine sells alcohol and / or tobacco products, the vending machine may need to verify the purchaser's age. Traditional verification methods based on identity (ID) cards can be easily bypassed. Figure 7 The vending machine in includes at least one camera. For example, Figure 7 The front camera and the rear camera are shown in FIG. The vending machine is used to obtain one or more images of the purchaser from at least one camera. Then, the vending machine is used to Figures 1 to 4 Various aspects of the invention mentioned herein estimate the age of a purchaser, either as a separate method, in addition to, or as an alternative to verifying the age of a purchaser. Figure 7 An example of a method for a vending machine to verify the age of a purchaser is shown on the right side of FIG. One advantage of the vending machine provided by the present invention is that even if the purchaser deliberately covers his / her face to avoid age verification, the vending machine can still estimate his / her age based on clothing. Optionally, the vending machine can also include liveness detection as an anti-spoofing method.
[0101] Figure 8 An example of age verification for a vending machine provided by the present invention is shown. An image of a customer is captured, and the customer's body and one or more objects are detected in the image. A cropped image is created for the object identified with high confidence. Then, according to the present invention, age prediction is performed based on the image of the body, the cropped object image, and the surrounding objects in the image. For example, if the customer is predicted to be a minor, the customer's order at the vending machine can be canceled.
[0102] Fig. 9 A schematic diagram of a method 900 for age estimation provided by the present invention is shown. The method 900 is executed by a device and includes the following steps:
[0103] - Step 901: Acquire an image of a human being;
[0104] - Step 902: cropping one or more body parts from the image;
[0105] - Step 903: Segment clothes from the cropped one or more body parts to obtain a clothes image;
[0106] - Step 904: removing background pixels from the clothing image;
[0107] - Step 905: providing the clothing image as input to the neural network model;
[0108] - Step 906: Obtain the output of the neural network model, wherein the output of the neural network model represents the age category of a human being.
[0109] It should be noted that from the above Figures 1 to 4 From the perspective of FIG. 9 , the steps of method 900 may share the same functions and details. Therefore, the corresponding method implementation will not be described in detail here.
[0110] Fig.10 An example of a neural network model provided by the present invention is shown. Fig.10 The neural network model shown is based on the PyTorch framework. It should be noted that Fig.10 The neural network model in is only an illustrative example. The neural network model that can be used in the present invention is not limited to Fig.10 . Fig.10 The neural network model in can be applied to the present invention Figures 1 to 9 .
[0111] The device in the present invention may include a processing circuit (not shown) for performing, carrying out or starting various operations of the device described herein. The processing circuit may include hardware and software. The hardware may include analog circuits or digital circuits, or both analog circuits and digital circuits. The digital circuit may include components such as an application-specific integrated circuit (ASIC), a field-programmable array (FPGA), a digital signal processor (DSP) or a multi-purpose processor. Optionally, the processing circuit includes one or more processors and a non-transitory memory connected to the one or more processors. The non-transitory memory may carry an executable program code, which, when executed by one or more processors, causes the device to perform, carry out or start the operations or methods described herein.
[0112] For example, the device provided by the present invention can be a computer, a server, a tablet computer, a mobile terminal, a graphics processing unit, a neural processing unit, or other electronic device capable of computing.
[0113] The present invention has been described in conjunction with various embodiments as examples and implementations. However, from a study of the drawings, the present invention and the independent claims, a person skilled in the art will be able to understand and implement other variations when implementing the claimed invention. In the claims and the specification, the word "comprising" does not exclude other elements or steps, and the quantifier "a" does not exclude a plurality. A single element or other unit may fulfill the functions of several entities or items described in the claims. The listing of certain methods in different dependent claims does not mean that a combination of these methods cannot be used in an advantageous implementation.
Claims
1. A method (900) for age estimation, characterized in that The method comprises: Acquire (901) an image of a person (110); cropping (902) one or more body parts from the image (110); Segmenting (903) clothing from the cropped one or more body parts to obtain a clothing image (130); removing (904) background pixels from the clothing image (130); providing (905) the clothing image (130) as input to a neural network model (140); An output (150) of the neural network model is obtained (906), wherein the output (150) of the neural network model indicates an age category of the human being.
2. The method (900) according to claim 1, characterized in that The neural network model (140) is based on an 18-layer residual neural network (ResNet-18) model.
3. The method (900) according to claim 1 or 2, characterized in that: The neural network model (140) includes three convolutional layers and three feature refinement blocks, and the obtaining (906) the output (150) of the neural network model includes: Extracting three feature representations of the clothing image (130) from the three outputs of the three convolutional layers; The three feature representations are provided as inputs to the three feature refinement blocks to obtain three refined features from the three outputs of the three feature refinement blocks.
4. The method (900) according to claim 3, characterized in that The method further comprises: concatenating the three refined features with the output of the first convolutional layer among the three convolutional layers to obtain a final feature representation of the clothing image (130).
5. The method (900) according to claim 4, characterized in that The neural network model (140) includes a fully connected layer (144), and the method further includes: providing the final feature representation of the garment as input to the fully connected layer (144); The output of the fully connected layer is obtained as the output of the neural network model (150).
6. The method (900) according to any one of claims 1 to 5, characterized in that: The image (110) of the human does not include a face of the human.
7. A device (100) for age estimation, characterized in that The device (100) is used for: Get an image of a human being (110); cropping one or more body parts from the image (110); Segmenting clothes from the cropped one or more body parts to obtain a clothes image (130); removing background pixels from the clothing image (130); providing the clothing image (130) as input to a neural network model (140) implemented in the device; An output (150) of the neural network model is obtained, wherein the output (150) of the neural network model indicates an age category of the human being.
8. The device (100) according to claim 7, characterized in that The neural network model (140) is based on an 18-layer residual neural network (ResNet-18) model.
9. The device (100) according to claim 7 or 8, characterized in that The neural network model (140) includes three convolutional layers and three feature refinement blocks. In order to obtain the output (150) of the neural network model, the device is used to: Extracting three feature representations of the clothing image (130) from the three outputs of the three convolutional layers; The three feature representations are provided as inputs to the three feature refinement blocks to obtain three refined features from the three outputs of the three feature refinement blocks.
10. The device (100) according to claim 9, characterized in that The device is also used to cascade the three refined features with the output of the first convolutional layer in the three convolutional layers to obtain a final feature representation of the clothing image (130).
11. The device (100) according to claim 10, characterized in that The neural network model (140) includes a fully connected layer, and the device is further used for: providing the final feature representation of the clothing as input to the fully connected layer; The output of the fully connected layer is obtained as the output of the neural network model (150).
12. The device (100) according to any one of claims 7 to 11, characterized in that The image (110) of the human does not include a face of the human.
13. A computer program product comprising instructions, characterized in that When the program is executed by a computer, the instructions cause the computer to perform the method according to any one of claims 1 to 6.