Kinship determination method, device, electronic device and storage medium

Abstract

The embodiments of the present application relate to the field of computer image processing, and disclose a method, device, electronic device, and storage medium for determining kinship. The method comprises: selecting multiple facial images from a family image set of a family to be matched as input data and inputting them into a family facial feature extraction model to obtain a first family facial feature of the family to be matched; replacing at least one of the multiple facial images with a face image to be tested, and inputting the replaced multiple face images into the family facial feature extraction model to obtain a second family facial feature; comparing the first family facial feature with the second family facial feature to determine the kinship between the face image to be tested and the family to be matched. The judgment of kinship has more familial characteristics, and the judgment result is more reliable. At the same time, the kinship is obtained only by relying on the image set, which reduces the investment of manpower and material resources compared to the judgment process such as DNA.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of computer image processing, and in particular to a method, device, electronic device, and storage medium for determining kinship. Background Art

[0002] When searching for missing children, most often childhood photos are used, with an emphasis on certain innate facial features, for assistance from caring individuals. However, it is understandable that childhood and adolescence are the primary periods of human growth, and childhood photos may not be identical to the current facial features of the person being sought, making identification of the person inaccurate. At the same time, it is not ruled out that facial information with similar childhood features but not the person being sought may be discovered. In such cases, further confirmation through technical means such as DNA testing is not only costly but also disrupts the normal life of the person being misidentified to some extent.

[0003] Therefore, there is an urgent need for a kinship determination method based on face recognition. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide a method for determining kinship, which can realize kinship determination based on face recognition, improve the accuracy of kinship determination, and reduce labor costs.

[0005] To solve the above technical problems, an embodiment of the present invention provides a method for determining kinship, comprising the following steps: selecting multiple facial images from a family image set of a family to be matched as input data and inputting them into a family facial feature extraction model to obtain a first family facial feature of the family to be matched; replacing at least one of the multiple facial images with a facial image to be tested, and inputting the replaced multiple facial images into the family facial feature extraction model to obtain a second family facial feature; comparing the first family facial feature with the second family facial feature to determine the kinship between the facial image to be tested and the family to be matched.

[0006] An embodiment of the present invention also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned kinship determination method.

[0007] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned kinship determination method when executed by a processor.

[0008] This application provides a method for determining kinship based on family facial features. By comparing the features of the face of the family to be matched with the family facial features obtained by replacing some family members with the face to be tested, the kinship relationship between the face to be tested and the family to be matched is determined. This method for determining kinship no longer simply judges two specific facial features. Instead, it presents the facial features to be matched in the form of family facial features. By comparing the image of the face to be tested before and after it is inserted into the family facial images, the degree of interference caused by the family facial features is determined. This method of determining kinship has more family characteristics and the judgment results are more reliable. At the same time, obtaining kinship only by a set of images reduces the investment of manpower and material resources compared to judgment processes such as DNA. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings. These exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are represented as similar elements. Unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0010] Figure 1 This is a flow chart of a method for determining kinship provided by one embodiment of the present application;

[0011] Figure 2 This is a schematic diagram of a family face feature extraction model provided by one embodiment of the present application;

[0012] Figure 3 It is a schematic diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0013] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, each embodiment of the present invention will be described in detail below with reference to the accompanying drawings. However, it will be understood by those skilled in the art that in each embodiment of the present invention, many technical details are provided to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in this application can be implemented. The division of the following embodiments is for convenience of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined with each other and referenced to each other under the premise that there is no contradiction.

[0014] The terms "first" and "second" in the embodiments of the present application are only used for descriptive purposes and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present application, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a system, product or device comprising a series of components or units is not limited to the listed components or units, but may optionally also include components or units that are not listed, or may optionally also include other components or units that are inherent to these products or devices. In the description of the present application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0015] One embodiment of the present invention relates to a method for determining kinship. The specific process is as follows: Figure 1 shown.

[0016] Step 101: Select multiple facial images from the family image set of the family to be matched as input data and input them into the family facial feature extraction model to obtain the first family facial feature of the family to be matched;

[0017] Step 102: replacing at least one of the multiple facial images with the facial image to be tested, and inputting the replaced facial images into a family facial feature extraction model to obtain a second family facial feature;

[0018] Step 103 : Compare the facial features of the first family with the facial features of the second family to determine the relationship between the facial image to be tested and the family to be matched.

[0019] This application provides a method for determining kinship based on family facial features. By comparing the features of the face of the family to be matched with the family facial features obtained by replacing some family members with the face to be tested, the kinship relationship between the face to be tested and the family to be matched is determined. This method for determining kinship no longer simply judges two specific facial features. Instead, it presents the facial features to be matched in the form of family facial features. By comparing the image of the face to be tested before and after it is inserted into the family facial images, the degree of interference caused by the family facial features is determined. This method of determining kinship has a more familial nature and the judgment results are more reliable. At the same time, relying solely on image sets to obtain kinship reduces the investment of manpower and material resources compared to the DNA judgment process.

[0020] The following is a detailed description of the implementation details of the kinship determination method of this embodiment. The following content is only provided for ease of understanding and is not necessary for implementing this solution.

[0021] In step 101, multiple facial images are selected from a family image set of the family to be matched as input data and fed into a family facial feature extraction model to obtain a first family facial feature of the family to be matched. The family image set is a complete image database of the family to be matched. The specific number of facial images to be selected from the family image set is the same as the number of images in the training samples used by the family facial feature extraction model during training. The multiple facial images selected can be facial images of people who may have a direct blood relationship with the missing person within three generations, which can specifically obtain a first family facial feature that is more closely related to the missing person. Furthermore, the multiple facial images selected can also be directly or indirectly related to each other. In this case, the multiple facial images selected can have stable family inheritance and form a relatively unified family facial feature.

[0022] In one example, obtaining a family image set of a family to be matched includes: obtaining facial images of family members to be matched, and performing at least one of the following augmentation processing on the facial images of each member to form a family image set of the family to be matched: blurring, occlusion, shearing, and random noise addition. For example, when the number of facial images of family members to be matched is less than the number of inputs of the family face extraction model, or the attributes are too single, the acquired images can be augmented, including blurring, occlusion, shearing, and random noise, to increase the number of samples and ensure that the above-mentioned kinship determination method can be executed normally. In step 102, at least one of the multiple facial images is replaced with the facial image to be tested, and the multiple facial images after replacement are input into the family facial feature extraction model to obtain the second family facial features. That is, the facial features of the second family are obtained by combining the facial image to be tested and the facial images of the family to be matched; the facial features of the second family are compared with the facial features of the first family obtained only by the facial images of the family to be matched, and the specific impact of the addition of the facial image to be tested on the facial features of the family is obtained, so as to judge the kinship relationship between the facial image to be tested and the family to be matched.

[0023] In one example, replacing at least one of the multiple facial images with a facial image to be tested includes: copying the facial image to be tested to obtain a number of facial images to be tested that is the same as the number of the multiple facial images, and using the copied multiple facial images to be tested as the replaced multiple facial images; or, replacing an image in the multiple facial images that has facial attributes consistent with the facial image to be tested with the facial image to be tested to obtain the replaced multiple facial images; facial attributes include: gender attributes and / or age attributes.

[0024] Among them, the step of replacing the image whose facial attributes are consistent with the facial image to be tested in the above-mentioned multiple facial images with the facial image to be tested is based on the premise that there must be an image with the attributes consistent with the facial image to be tested in these multiple images; for example, a total of 6 facial images {m1, m2, m3, m4, m5, m6} are selected from the family image set of the family to be matched, among which the attributes of the two facial images m1 and m4 are consistent with the attributes of the facial image to be tested, then the facial image to be tested replaces m1 and m4, that is, the two facial images to be tested and the remaining facial images {m2, m3, m5, m6} in the 6 facial images constitute input data, which is input into the family facial feature extraction model to obtain the second family facial feature.

[0025] In actual implementation, the face image to be tested is copied to obtain multiple face images to be tested that are the same in number as the multiple face images, which makes the operation more convenient and quick, and the acquired facial features of the second family are more highly correlated with the features of the face to be tested, and the differences and similarities are more obvious when compared with the facial features of the first family; when the face image to be tested is used to replace the image with the same facial attributes as the face image to be tested in the multiple face images, the facial features of the second family combine the facial features of the family to be matched and the features of the face image to be tested, and when compared with the facial features of the first family, more attention is paid to the feature details, that is, the sensitivity of the kinship judgment is higher; at the same time, the face image to be tested replaces the face image in the family to be matched with the same facial attributes. Compared with the process of obtaining the facial features of the first family, there is no difference in the facial attribute parameters, that is, the facial attributes do not affect the extraction of the facial features of the two families, making the kinship judgment result more reliable.

[0026] In one example, the facial images to be tested are multiple images of the face to be tested belonging to different age groups; replacing at least one of the multiple facial images with the facial image to be tested includes: replacing images in the multiple facial images that have facial attributes consistent with the facial image to be tested with the facial images to be tested belonging to different age groups, thereby obtaining multiple replaced facial images; facial attributes include gender and age. When the facial images to be tested are multiple images of the face to be tested belonging to different age groups, the process of replacing at least one of the multiple facial images with the facial image to be tested can perform replacement of attributes consistent with those of different age groups, thereby reducing the influence of the age of the face to be tested on the generation of facial features for the second family.

[0027] In one example, a similarity analysis is performed on facial images in a family image set belonging to the family to which the family image sample belongs to determine the family's genetic type, which includes at least one of male inheritance, female inheritance, and at least one of the generations-abandoning inheritance. A facial image to be tested is then used to replace one of the multiple facial images, so that the facial image to be tested and at least one of the multiple facial images that has not been replaced form the genetic type of the family, thereby obtaining multiple replaced facial images. Specifically, the replacement steps for the facial image to be tested are adjusted based on the specific genetic characteristics of the family to be matched, so that the replaced facial image to be tested can specifically affect the facial features of the family to be matched. For example, if the male facial genetic characteristics of the family to be matched are more pronounced, and the facial image to be tested is a male face, the facial image to be tested is used to replace any one of the multiple facial images, so that the facial image to be tested and the facial image that has not been replaced form a male genetic relationship, thereby increasing the influence of the facial image to be tested on the characteristic changes of the obtained facial features of the second family.

[0028] In step 103, the facial features of the first family are compared with the facial features of the second family to determine the relationship between the facial image to be tested and the family to be matched. For example, the similarity between the facial features of the first family and the facial features of the second family is obtained and compared with a preset similarity threshold. If the similarity is not less than the preset similarity threshold, it is determined that the facial features of the facial image to be tested match the facial features of the family to be matched, and the facial image to be tested belongs to the family to be matched. In other words, the person corresponding to the facial image to be tested is related to the family to be matched and may be a missing person from the family to be matched.

[0029] In one example, the family face feature extraction model includes a feature extraction layer and a feature fusion layer. The training process of the family face feature extraction model includes: extracting individual face features of each individual image contained in the family image sample through the feature extraction layer; fusing the individual face features through the feature fusion layer to obtain the family face features corresponding to the family image sample. Specifically, in order to complete the training of the family face feature extraction model, it is necessary to obtain: ① Family image set: Each family corresponds to a family image set, and each family image set contains multiple face images of corresponding family members. For example: A family face image set is represented as A = (a1, a2, ..., a n ), the B family face image set is expressed as B = (b1, b2, ..., b n)…; ② Family image samples: Each family image sample contains multiple facial images belonging to the same family. The multiple facial images can be facial images belonging to the same person (different angles, different ages), or can be facial images belonging to different people in part or in whole. In this way, the same family can correspond to multiple family image samples containing non-completely identical facial images, such as: A family image sample 1, A family image sample 2, B family image sample 1, B family image sample 2…; For example: A family image sample 1 includes 4 facial images, i.e., A1 = (a1, a2, a3, a4), A family image sample 2 includes 4 facial images, i.e., A2 = (a1, a2, a6, a7); ③ Non-family image samples: Each non-family image sample contains at least two facial images belonging to different families, i.e., each non-family image sample contains at least two facial images of different people, such as: non-family image sample 1 = (a1, a2, a3, c4), non-family image sample 2 = (a1, d2, a3, c4).

[0030] When training the family facial feature extraction model, the fusion process also requires determining the fusion parameters for different facial attributes, including determining the age fusion parameter. Specifically, for the facial features of each family member, the facial features of children and the elderly typically change significantly with age, while the facial features of adults are relatively stable. Therefore, when constructing family facial features based on facial feature fusion, facial images of different age groups can be weighted and fused according to age. For example, the facial features of adults will account for a larger proportion of the fused family facial features, while the facial features of children and the elderly will account for a smaller proportion. This weighting can be specifically set using an attention mechanism. Furthermore, determining the parameters for different attributes also includes determining the gender fusion parameter. Specifically, for the facial features of each family member, there are certain regularities in the inheritance of gender between different individuals, such as male inheritance and female inheritance. Therefore, when constructing family facial features based on facial feature fusion, facial images of different genders can be weighted and fused according to gender. For example, if the family is pre-determined to be male-based based on similarity, male facial features will have a larger proportion in the fused family facial features, while female facial features will have a smaller proportion in the fused family facial features, and vice versa. In addition, the genetic type results obtained here for male or female inheritance can be used in the replacement process in step 102, that is, according to the genetic rules obtained here, the facial image to be tested replaces one of the multiple facial images, so that the facial image to be tested and at least one of the multiple facial images that has not been replaced form the genetic type to which the family belongs.

[0031] In an actual implementation, for determining the gender fusion parameters, for example, the similarity between every two male (female) face images in the family image set is calculated, and the male genetic parameters inherited by the males of the family are determined based on the similarity. The male genetic parameters are used to measure the degree of obviousness of the genetic characteristics of males in the family. Specifically, the average value of the multiple similarities calculated for males can be taken as the male genetic parameters of the family. The two face images used to calculate the similarity are face images belonging to different people. For example: from the face image set of family A (a1, a2, ..., a n ) randomly select two male face images a p 、a q (p≠q) Calculate the similarity μ between the two face images i , based on multiple similarity values ​​μ determined for every two male face images within family A i , and finally calculate the male genetic parameters of family A by calculating the average similarity. For example: the male genetic parameters of family A are S Am =(μ1+μ2+…+μ n ) / n.

[0032] In one example, the feature fusion layer is used to fuse individual facial features to obtain family facial features corresponding to the family image sample, including: setting fusion weights for the corresponding individual facial features according to the age group of each facial image in the family image sample, and performing weighted fusion of the individual facial features based on the fusion weights. Figure 2 As shown in the figure, the family face feature extraction model includes: feature extraction layer and feature fusion layer. During training, the classifier is connected to the family face feature extraction model for training, and both family image samples and non-family image samples are input into the model for training; where X j , the jth image of the i-th family; f j , facial features of the jth image of the i-th family; F 家族 , the family face features of the i-th family. n When all the sth family image samples belong to the ith family, F 家族 is the family facial feature of the sth family image sample of the i-th family.

[0033] Specifically, the feature extraction layer of the family facial feature extraction model is used to extract features from a single image and can be implemented using a convolutional neural network. The feature fusion layer can pre-determine fusion parameters for different fusion types (gender, age, etc.) by comparing similarities between family faces. It then sets weights (channel weights) for the images in the current sample based on the fusion parameters of the fusion type, fusing the facial features of the different facial images together to generate the family facial features. The fusion parameter is used to measure the importance of the fusion type during fusion; its size determines the number of channels used to fuse the corresponding face type within the fusion type to generate the family facial features.

[0034] In addition, the loss function for training the family facial feature extraction model is constructed based on the loss between the individual facial features of each single image and the family facial features, as well as the loss between each family facial feature and the corresponding family facial label. The loss between each family facial feature and the corresponding family facial label is used to measure the similarity of the family facial features of different families obtained by the feature fusion layer. The loss is controlled so that the similarity between the aforementioned family facial features is as low as possible, that is, the inter-class distance is as large as possible. For example, when the family facial labels of two families, Family A and Family B, are obtained, the model training parameters are adjusted so that the two family facial features are close to the family facial labels of their corresponding families in terms of distance, thereby reducing the similarity between the family facial features corresponding to the two families as much as possible, making the differences between the families clearer and making the differences between the families clearer when making subsequent predictions based on family facial features. The loss function also includes the loss between the individual facial features of each individual image and the family facial features, which is used to measure the feature similarity between the individual facial features obtained by the feature extraction layer and the family facial features obtained by the feature fusion layer, and by adjusting the model training parameters, the overall loss is controlled to be less than a certain range. For example, by calculating the loss between the facial features of family A and the facial features of each facial image in the face image sample of family A, the obtained family features are made closer to the individual facial features within the same family. When performing family labeling on facial images, family face labels can be used for identification and distinction; wherein, family image samples composed of facial images in the same family correspond to a family face label according to the family they belong to, and family image samples composed of facial images from different families correspond to a family face label together. That is, for family image samples, if they all belong to family A, they correspond to the family face label of family A; if they all belong to family B, they correspond to the family face label of family B; if some belong to family A and some belong to family B, that is, the family image sample is composed of face images from different families, then the family image sample corresponds to a specific family face label, for example, the K label, which indicates that the family image sample contains face images from more than one family.

[0035] In one example, the kinship determination method further includes: comparing the facial features of the second family with the single facial features extracted from the face image to be tested by the feature extraction layer to determine the kinship relationship between the face image to be tested and the family to be matched. Specifically, the facial features of the second family are obtained by fusing the face image to be tested with the facial images in the family to be matched, that is, the facial features of the second family are the result of transforming the face image to be tested under the features of the family to be matched; if the facial features of the second family are compared with the single facial features extracted from the face image to be tested by the feature extraction layer, and the similarity is greater than a preset comparison threshold, then it indicates that the facial features of the family to be matched do not have a significant impact on the single facial features of the face image to be tested, that is, the single facial features of the face image to be tested are originally similar to the facial features of the family to be matched, and further indicates that the face image to be tested belongs to the family to be matched, that is, there is a kinship relationship, and the subject of the face image to be tested may be a missing person from the family to be matched.

[0036] In this embodiment, a method for determining kinship based on family facial features is provided. By comparing the features of the family face to be matched with the family facial features obtained after extracting the features of the test face, the kinship relationship between the test face and the family to be matched is determined. This method for determining kinship no longer simply judges between two specific facial features. Instead, it presents the features of the face to be matched in the form of family facial features. By comparing the test face image before and after intervening with the family facial images, the degree of interference caused by the family facial features is determined. This method of determining kinship is more family-specific and reliable. Furthermore, by relying solely on a set of images to determine kinship, it reduces the human and material investment compared to DNA-based determination processes.

[0037] The steps of the various methods above are divided only for the purpose of clear description. During implementation, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this patent. Adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the scope of protection of this patent.

[0038] One embodiment of the present invention relates to an electronic device, such as Figure 3 As shown, it includes: at least one processor 201; and a memory 202 that is communicatively connected to the at least one processor 201; wherein the memory 202 stores instructions that can be executed by the at least one processor 201, and the instructions are executed by the at least one processor 201 to enable the at least one processor 201 to execute the above-mentioned kinship determination method.

[0039] The memory and processor are connected using a bus, which can include any number of interconnected buses and bridges. The bus connects various circuits of one or more processors and memories. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. These are all well known in the art and are therefore not described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over a wireless medium via an antenna. Furthermore, the antenna receives data and transmits it to the processor.

[0040] The processor is responsible for managing the bus and general processing, and may also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. The memory may be used to store data used by the processor when performing operations. One embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by the processor, the computer program implements the above-described method embodiment.

[0041] That is, those skilled in the art will understand that all or part of the steps in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a program, which is stored in a storage medium and includes a number of instructions for causing a device (which may be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc., various media that can store program code.

[0042] Those skilled in the art will appreciate that the above-mentioned embodiments are specific examples for implementing the present invention, and that in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for determining kinship, characterized in that: include: Selecting multiple facial images from the family image set of the family to be matched as input data and inputting them into the family facial feature extraction model to obtain the first family facial feature of the family to be matched; replacing at least one of the plurality of facial images with the facial image to be tested, and inputting the replaced plurality of facial images into the family facial feature extraction model to obtain a second family facial feature; Compare the facial features of the first family with the facial features of the second family to determine the relationship between the face image to be tested and the family to be matched; wherein, The family face feature extraction model includes a feature extraction layer and a feature fusion layer. The training process of the family face feature extraction model includes: Extracting individual facial features of each individual image contained in the family image sample through the feature extraction layer; The individual facial features are fused through a feature fusion layer to obtain family facial features corresponding to the family image samples; Among them, the loss function for training the family facial feature extraction model is constructed based on the loss between the single facial features of each single image and the family facial features, as well as the loss between each family facial feature and the corresponding family face label; the family image samples composed of facial images in the same family correspond to a family face label according to the family they belong to, and the family image samples composed of facial images from different families correspond to a family face label together.

2. The method for determining kinship according to claim 1, wherein: The replacing at least one of the plurality of facial images with the facial image to be tested comprises: Copying the facial image to be tested to obtain a plurality of facial images to be tested that is the same in number as the plurality of facial images, and using the plurality of facial images to be tested as the plurality of facial images after replacement; Alternatively, the facial image to be tested is used to replace an image in the plurality of facial images that has the same facial attributes as the facial image to be tested, to obtain the plurality of facial images after replacement; the facial attributes include: gender attribute and / or age attribute.

3. The method for determining kinship according to claim 1, wherein: The facial images to be tested are multiple images of different age groups of the faces to be tested; and replacing at least one of the multiple facial images with the facial image to be tested includes: The facial images to be tested belonging to different age groups are used to replace the images in the multiple facial images that have the same facial attributes as the facial images to be tested, to obtain the multiple facial images after replacement; the facial attributes include: gender attribute and age attribute.

4. The method for determining kinship according to claim 1, wherein: The method further comprises: Performing similarity analysis on facial images in a family image set of the family to which the family image sample belongs to determine the genetic type of the family, wherein the genetic type includes at least one of male inheritance, female inheritance, and skipped-generation inheritance; The facial image to be tested is used to replace one of the multiple facial images, so that the facial image to be tested and at least one of the multiple facial images that has not been replaced form the genetic type to which the family belongs, thereby obtaining the multiple facial images after replacement.

5. The method for determining kinship according to claim 1, wherein: The step of fusing the individual facial features through the feature fusion layer to obtain the family facial features corresponding to the family image samples includes: According to the age group to which each facial image in the family image sample belongs, a fusion weight is set for the corresponding single facial feature, and weighted fusion is performed on each single facial feature based on the fusion weight.

6. The method for determining kinship according to claim 1, wherein: The method further comprises: The second family facial features are compared with the single facial features of the face image to be tested extracted by the feature extraction layer to determine the relationship between the face image to be tested and the family to be matched.

7. The method for determining kinship according to any one of claims 1 to 6, characterized in that: Acquiring a family image set of the family to be matched, including: Acquire facial images of the family members to be matched, and perform at least one of the following augmentation processes on the facial images of each member to form a family image set of the family to be matched: Blur, occlude, clip, and randomly add noise.

8. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the kinship determination method according to any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the kinship determination method according to any one of claims 1 to 7 is implemented.

Images