CN112101072A - Face matching method, device, equipment and medium - Google Patents

Abstract

The embodiment of the invention provides a face matching method, a face matching device, face matching equipment and a face matching medium, wherein the face matching method comprises the following steps: acquiring a face image of a user; extracting the face features of the user from the face image; calculating the face similarity of the face of the user and the face in each video according to the face features of the user and video face features corresponding to each video in a pre-stored video library; identifying gender characteristics of a user and hair style characteristics of the user by using the face image; calculating the hair style similarity between the hair style characteristics of the user and the hair style characteristics of the person in each video; and screening a target video corresponding to the user in a video library according to the face similarity and the hair style similarity, wherein the gender characteristics of the person in the target video are consistent with those of the user. According to the embodiment of the invention, the video corresponding to the face similar to the user can be screened out from the video library, so that the video character of the user after face changing is closer to the face of the user, and the experience degree of the user is improved.

Description

Face matching method, device, equipment and medium

technical field

The present invention relates to the technical field of face swapping, and in particular, to a face matching method, device, equipment and medium.

Background technique

Face conversion is a popular application in the field of computer vision. Face conversion can generally be used for video synthesis, providing privacy services, portrait replacement, or other innovative applications.

At present, the user's face is too different from the user's face after changing his face in the video, and it is impossible to recommend a video corresponding to a face that is closer to the user's face to the user based on the user's face characteristics, resulting in a poor user experience.

SUMMARY OF THE INVENTION

Embodiments of the present invention provide a face matching method, device, equipment and medium, which can filter out videos corresponding to faces similar to users in a video library, so that the video characters after the user changes faces are more closely related to the user's face. Proximity to improve user experience.

In a first aspect, an embodiment of the present invention provides a face matching method, including:

Get the user's face image;

Extract the user's face features from the face image;

Calculate the face similarity between the user's face and the face in each video according to the user's face features and the video face features corresponding to each video in the pre-stored video library;

Use face images to identify the user's gender characteristics and the user's hairstyle characteristics;

Calculate the hairstyle similarity between the user's hairstyle features and the hairstyle features of the characters in each video;

According to the similarity of faces and hairstyles, the target videos corresponding to the user are screened in the video library, wherein the gender characteristics of the characters in the target videos are consistent with the gender characteristics of the user.

In a second aspect, an embodiment of the present invention provides a face matching device, which includes:

The acquisition module is used to acquire the face image of the user;

The extraction module is used to extract the user's face features from the face image;

The first calculation module is used to calculate the face similarity between the user's face and the face in each video according to the user's face feature and the video face feature corresponding to each video in the pre-stored video library;

The recognition module is used to recognize the gender characteristics of the user and the hairstyle characteristics of the user by using the face image;

The second computing module is used to calculate the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video;

The selection module is used for screening the target video corresponding to the user in the video library according to the similarity of the face and the similarity of the hairstyle, wherein the gender characteristics of the characters in the target video are consistent with the gender characteristics of the user.

In a third aspect, an embodiment of the present invention provides a computer device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory, when the computer program instructions are executed by the processor, the first aspect is implemented method.

In a fourth aspect, embodiments of the present invention provide a computer-readable storage medium on which computer program instructions are stored, characterized in that, when the computer program instructions are executed by a processor, the method of the first aspect is implemented.

The face matching method, device, device and medium provided by the embodiments of the present invention obtain the user's face image; extract the user's face feature from the face image; The video face features corresponding to the video, calculate the face similarity between the user's face and the face in each video; use the face image to identify the user's gender feature and the user's hairstyle feature; calculate the user's hairstyle feature and each video. The hairstyle similarity between the hairstyle features of the characters in the video library; according to the face similarity and hairstyle similarity, the target video corresponding to the user is screened in the video library, wherein the gender features of the characters in the target video are consistent with the gender features of the user. The embodiment of the present invention can filter out videos corresponding to faces similar to the user in the video library, so that the video characters of the user after changing faces are closer to the user's face, and the user experience is improved.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the accompanying drawings required in the embodiments of the present invention will be briefly introduced below. For those of ordinary skill in the art, without creative work, the Additional drawings can be obtained from these drawings.

1 shows a flowchart of a method for face matching provided according to some embodiments of the present invention;

FIG. 2 shows a schematic structural diagram of a face matching apparatus provided according to some embodiments of the present invention;

FIG. 3 shows a schematic structural diagram of a computing device according to some embodiments of the present invention.

Detailed ways

The features and exemplary embodiments of various aspects of the present invention will be described in detail below. In order to make the objects, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only configured to explain the present invention, and are not configured to limit the present invention. It will be apparent to those skilled in the art that the present invention may be practiced without some of these specific details. The following description of the embodiments is only intended to provide a better understanding of the present invention by illustrating examples of the invention.

It should be noted that, in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any relationship between these entities or operations. any such actual relationship or sequence exists. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass a non-exclusive inclusion such that a process, method, article or device that includes a list of elements includes not only those elements, but also includes not explicitly listed or other elements inherent to such a process, method, article or apparatus. Without further limitation, an element defined by the phrase "comprises" does not preclude the presence of additional identical elements in a process, method, article, or device that includes the element.

Referring to FIG. 1, an embodiment of the present invention provides a face matching method, and the method includes: S101-S106.

S101: Obtain a face image of a user.

During specific implementation, the face image of the user may be a picture taken by the user using the terminal, or may be a picture uploaded by the user before recording the face-changing video.

S102: Extract the user's face features from the face image.

During specific implementation, the user's facial features include the shape and position of the user's facial features. For example, a three-dimensional morphable model (3D morphable model, 3DMM) can be used to extract feature point coordinates from a user's face image. The extracted face feature may be a 1×P-dimensional vector, where P is an integer greater than 1.

S103: Calculate the face similarity between the user's face and the face in each video according to the user's face feature and the video face feature corresponding to each video in the pre-stored video library.

In the specific implementation, the video face features corresponding to each video in the video library are extracted in advance, and the extracted video face features are generated into video feature vectors for storage. For example, the total number of videos in the video library is M, the video feature vector is M×P. Calculate the similarity between the face feature and the face in each video. For example, if the face feature is a 1×P-dimensional vector, and the face features of M videos are an M×P-dimensional vector, then calculate 1×P× (M×P) ^T obtains a 1×M-dimensional feature vector, which represents the similarity of M faces, and the similarity of M faces corresponds to M videos one-to-one, which means that the user’s face is respectively associated with M Face similarity of video faces.

In some embodiments, the videos whose face similarity between the M video faces and the user's face is greater than the first preset value may be recommended to the user for face swapping.

S104: Identify the gender feature of the user and the hairstyle feature of the user by using the face image.

During specific implementation, the user's gender feature and the user's hairstyle feature are identified according to the face image. For example, the user's hairstyle style, hair length, hair curl, hair color, and bangs shape all belong to the user's hairstyle feature.

S105: Calculate the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video.

During specific implementation, the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video is calculated.

In some embodiments, the hairstyle features include N types of hairstyle features, the hairstyle features of characters in each video in the video library include N types of hairstyle features, and for each video, the user's hairstyle features are calculated through the following steps The hairstyle similarity with the hairstyle features of the characters in the first video, where the first video is any video in the video library: traverse each of the N types, and use each type as the type to be processed; calculate the user The hairstyle similarity between the hairstyle features of the to-be-processed type and the hairstyle features of the to-be-processed type of characters in the first video.

For example, if the N hairstyle features include hairstyle style, hair length, hair curl, hair color, and bangs shape, the hairstyles of the user's to-be-processed hairstyle feature and the to-be-processed hairstyle feature of the character in the first video are similar in hairstyles The degree is calculated as follows:

The similarity between the hairstyle style of the user and the hairstyle style of the character in the first video is calculated as the similarity degree of the hairstyle style of the first video.

The similarity between the user's hair length and the hair length of the character in the first video is calculated as the hair length similarity of the first video.

The similarity between the hair curl of the user and the hair curl of the character in the first video is calculated as the similarity of the hair curl of the first video.

The similarity between the user's hair color and the hair color of the character in the first video is calculated as the hair color similarity of the first video.

The similarity between the user's bangs shape and the bangs shape of the characters in the first video library is calculated as the hair color similarity of the first video.

S106: Screen target videos corresponding to the user in the video library according to the similarity of faces and hairstyles, wherein the gender characteristics of the characters in the target videos are consistent with the gender characteristics of the user.

After calculating the hairstyle similarity between the hairstyle features of the type to be processed, the following two methods are used to filter the target video corresponding to the user in the video library through the hairstyle similarity and the face similarity.

One:

In the video library, select a video with a face similarity greater than the first preset value as the second video set, traverse each video in the video library, and use each video as the first video, if the hairstyle style similarity of the first video, The similarity of the hair length of the first video, the similarity of the hair curl of the first video, the similarity of the hair color of the first video, and the similarity of the bangs shape of the first video are all greater than the second preset value, then the The first video is added to the first video set, after traversing each video, the intersection of the second video set and the first video set is taken, the target video is selected from the intersected videos, and the target video is recommended to the user.

Second:

For any video in the video library, use the similarity calculation model to calculate the similarity of the face of the first video, the similarity of the hairstyle style of the first video, the similarity of the hair length of the first video, and the hair volume of the first video. The degree of similarity, the similarity of the hair color of the first video, and the similarity of the bangs shape of the first video are weighted and summed, and the comprehensive similarity between the user and the characters in the first video is obtained, and each video is traversed to obtain the user and the first video. Comprehensive similarity of characters in each video; screen target videos whose comprehensive similarity is greater than the third preset value in the video library, and recommend the target video to the user.

Here, the gender characteristics of people in all target videos are consistent with the gender characteristics of users. The gender feature can be identified through the face image, the user's gender can be identified according to the user's face feature and hairstyle feature, or the gender feature can be marked by collecting the user data and the account corresponding to the user.

In some embodiments, the face matching method provided by the embodiments of the present invention further includes training a similarity calculation model, which specifically includes:

Obtain a sample face image, and use multiple models to be trained to calculate the similarity and face similarity between the sample face image and the hairstyle features of the characters corresponding to each video. The model to be trained includes a plurality of parameters, and each parameter corresponds to the similarity of the face and the similarity of the hairstyle, that is, the weight corresponding to the similarity of the face and the hairstyle. Here, multiple sets of parameters are preset, that is, multiple models to be trained are preset. For each model to be trained, select a video with a face similarity greater than a first preset value and a hairstyle feature similarity greater than a second preset value; use the face swap model to swap the sample face image to the selected video Corresponding to the face of the character, a plurality of target face images are obtained; wherein, the face swapping model may be a generative adversarial network (Generative Adversarial Networks, GAN) or a CycleGAN. Input the sample face image and multiple target face images into the face recognition model, calculate the similarity between the sample face image and multiple target face images; calculate the multiple target face images corresponding to each model to be trained The average of the similarity; select the model to be trained corresponding to the highest average as the similarity calculation model.

The face matching method provided by the embodiment of the present invention obtains the face image of the user; extracts the face feature of the user from the face image; Features, calculate the face similarity between the user's face and the face in each video; use the face image to identify the user's gender characteristics and the user's hairstyle characteristics; calculate the user's hairstyle characteristics and the hairstyle characteristics of the characters in each video According to the similarity of face and hairstyle, the target video corresponding to the user is screened in the video library, wherein the gender characteristics of the characters in the target video are consistent with the gender characteristics of the user. The embodiment of the present invention can filter out videos corresponding to faces similar to the user in the video library, so that the video characters of the user after changing faces are closer to the user's face, and the user experience is improved.

Referring to FIG. 2, an embodiment of the present invention provides a face matching device, which includes:

an acquisition module 201, used for acquiring a face image of a user;

The extraction module 202 is used for extracting the facial features of the user from the facial image;

The first calculation module 203 is used to calculate the face similarity between the user's face and the face in each video according to the user's face feature and the video face feature corresponding to each video in the pre-stored video library;

The identification module 204 is used for identifying the gender feature of the user and the hairstyle feature of the user by using the face image;

The second calculation module 205 is used to calculate the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video;

The selection module 206 is configured to filter the target video corresponding to the user in the video library according to the similarity of the face and the similarity of the hairstyle, wherein the gender characteristics of the characters in the target video are consistent with the gender characteristics of the user.

In some embodiments, the hairstyle features of the user include N types of hairstyle features, the hairstyle features of characters in each video include N types of hairstyle features, the total number of videos in the video library is M, and M and N are respectively an integer greater than 1;

The second calculation module 205 is used to calculate the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video, including:

The hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in the first video is calculated by the following steps:

Traverse each of the N types and treat each type as a pending type;

Calculate the hairstyle similarity between the hairstyle features of the type to be processed of the user and the hairstyle features of the type to be processed of the characters in the first video; wherein the first video is any video in the video library.

In some embodiments, the N types of hairstyle features include: hairstyle style, hair length, hair curl, hair color, bang shape;

The second calculation module 205 is specifically configured to calculate the hairstyle similarity between the hairstyle features of the type to be processed of the user and the hairstyle features of the type to be processed of the characters in the first video, including:

Calculate the similarity between the user's hairstyle style and the hairstyle style of the character in the first video as the similarity of the hairstyle style of the first video;

Calculate the similarity between the user's hair length and the hair length of the character in the first video as the hair length similarity of the first video;

Calculate the similarity between the hair curl of the user and the hair curl of the character in the first video as the similarity of the hair curl of the first video;

Calculate the similarity between the user's hair color and the hair color of the character in the first video as the hair color similarity of the first video;

The similarity between the user's bangs shape and the bangs shape of the character in the first video is calculated as the hair color similarity of the first video.

The selection module 206 is used to filter the target video corresponding to the user in the video library according to the similarity of the face and the similarity of the hairstyle, including:

Selecting a video whose face similarity is greater than the first preset value in the video library to form a second video set;

Traverse each video and use each video as the first video. If the similarity of the hairstyle style of the first video, the similarity of the length of the hair of the first video, the similarity of the curl of the hair of the first video, the similarity of the hair of the first video The similarity of the color and the similarity of the bangs shape of the first video are both greater than the second preset value, then the first video is added to the first video set;

After traversing each video, the intersection of the second video set and the first video set is taken, and the target video is selected from the intersected videos.

In some embodiments, the face similarity includes the face similarity of the first video, and the face similarity of the first video is the face similarity between the user's face and the face in the first video;

The selection module 206 is used to filter the target video corresponding to the user in the video library according to the similarity of the face and the similarity of the hairstyle, including:

Using the similarity calculation model to calculate the similarity of the face of the first video, the similarity of the hairstyle style of the first video, the similarity of the hair length of the first video, the similarity of the hair curl of the first video, the similarity of the first video The similarity of the hair color and the similarity of the bangs shape of the first video are weighted and summed to obtain the comprehensive similarity between the user and the characters in the first video;

Traverse each video to get the comprehensive similarity between the user and the characters in each video;

The target videos whose comprehensive similarity is greater than the third preset value are screened in the video library.

In some embodiments, it further includes: a training module 207 for training a similarity calculation model;

The training module 207 for training the similarity calculation model includes:

Get a sample face image.

The similarity and face similarity between the sample face image and the hairstyle feature of the person corresponding to each video are calculated respectively by using multiple models to be trained.

For each model to be trained, a video with a face similarity greater than a first preset value and a hairstyle feature similarity greater than a second preset value is selected.

A face swapping model is used to swap the sample face image to the face of the person corresponding to the selected video to obtain multiple target face images.

The sample face image and the multiple target face images are input into the face recognition model, and the similarity between the sample face image and the multiple target face images is calculated.

Calculate the average value of the similarity of multiple target face images corresponding to each model to be trained;

Select the model to be trained corresponding to the highest average value as the similarity calculation model.

In some embodiments, it also includes a recommendation module 208;

The recommendation module 208 is used for recommending the target video to the user.

In addition, the face matching method of the embodiment of the present invention described in conjunction with FIG. 1 may be implemented by a computing device. FIG. 3 shows a schematic diagram of a hardware structure of a computing device provided by an embodiment of the present invention.

The computing device may include a processor 301 and a memory 302 storing computer program instructions.

Specifically, the above-mentioned processor 301 may include a central processing unit (CPU), or a specific integrated circuit (Application Specific Integrated Circuit, ASIC), or may be configured as one or more integrated circuits implementing the embodiments of the present invention.

Memory 302 may include mass storage for data or instructions. By way of example and not limitation, memory 302 may include a Hard Disk Drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive or two or more A combination of more than one of the above. Memory 302 may include removable or non-removable (or fixed) media, where appropriate. Memory 302 may be internal or external to the data processing device, where appropriate. In certain embodiments, memory 302 is non-volatile solid state memory. In particular embodiments, memory 302 includes read only memory (ROM). Where appropriate, the ROM may be a mask programmed ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), electrically rewritable ROM (EAROM) or flash memory or A combination of two or more of the above.

The processor 301 reads and executes the computer program instructions stored in the memory 302 to implement any one of the face matching methods in the foregoing embodiments.

In one example, the computing device may also include a communication interface 303 and a bus 310 . Among them, as shown in FIG. 3 , the processor 301 , the memory 302 , and the communication interface 303 are connected through the bus 310 and complete the communication with each other.

The communication interface 303 is mainly used to implement communication between modules, apparatuses, units and/or devices in the embodiments of the present invention.

The bus 310 includes hardware, software, or both, coupling the components of the computing device to each other. By way of example and not limitation, the bus may include Accelerated Graphics Port (AGP) or other graphics bus, Enhanced Industry Standard Architecture (EISA) bus, Front Side Bus (FSB), HyperTransport (HT) Interconnect, Industry Standard Architecture (ISA) Bus, Infiniband Interconnect, Low Pin Count (LPC) Bus, Memory Bus, Microchannel Architecture (MCA) Bus, Peripheral Component Interconnect (PCI) Bus, PCI-Express (PCI-X) Bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association Local (VLB) bus or other suitable bus or a combination of two or more of the above. Bus 310 may include one or more buses, where appropriate. Although embodiments of the present invention describe and illustrate a particular bus, the present invention contemplates any suitable bus or interconnect.

In addition, in combination with the face matching method in the foregoing embodiments, the embodiments of the present invention may be implemented by providing a computer-readable storage medium. Computer program instructions are stored on the computer-readable storage medium; when the computer program instructions are executed by the processor, any one of the face matching methods in the foregoing embodiments is implemented.

It is to be understood that the present invention is not limited to the specific arrangements and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above-described embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the sequence of steps after comprehending the spirit of the present invention.

The functional blocks shown in the above-described structural block diagrams may be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it may be, for example, an electronic circuit, an application specific integrated circuit (ASIC), suitable firmware, a plug-in, a function card, or the like. When implemented in software, elements of the invention are programs or code segments used to perform the required tasks. The program or code segments may be stored in a machine-readable medium or transmitted over a transmission medium or communication link by a data signal carried in a carrier wave. A "machine-readable medium" may include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, and the like. The code segments may be downloaded via a computer network such as the Internet, an intranet, or the like.

It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps may be performed in the order mentioned in the embodiment, or may be different from the order in the embodiment, or several steps may be performed simultaneously.

The above are only specific implementations of the present invention. Those skilled in the art can clearly understand that, for the convenience and brevity of the description, for the specific working process of the above-described systems, modules and units, reference may be made to the foregoing method embodiments. The corresponding process in , will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should all cover within the protection scope of the present invention.

Claims (10)

1. a face matching method, is characterized in that, described method comprises: Get the user's face image; extracting the face features of the user from the face image; Calculate the face similarity between the user's face and the face in each video according to the user's face feature and the video face feature corresponding to each video in the pre-stored video library; Using the face image to identify the gender characteristics of the user and the hairstyle characteristics of the user; Calculate the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video; According to the similarity of the face and the similarity of the hairstyle, a target video corresponding to the user is screened in the video library, wherein the gender characteristics of the characters in the target video are consistent with the gender characteristics of the user. 2. The method according to claim 1, wherein the hairstyle features of the user include N types of hairstyle features, the hairstyle features of the characters in each video include the N types of hairstyle features, and the The total number of videos in the video library is M, and M and N are integers greater than 1 respectively; The calculating the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video includes: The hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in the first video is calculated by the following steps: Traverse each of the N types, and use the each type as a type to be processed; Calculate the hairstyle similarity between the hairstyle feature of the type to be processed of the user and the hairstyle feature of the to-be-processed type of the character in the first video; wherein the first video is any video in the video library . 3. The method of claim 2, wherein The hair characteristics of the N types include: hair style, hair length, hair curl, hair color, and bangs shape; The calculating the hairstyle similarity between the hairstyle features of the type to be processed of the user and the hairstyle features of the type to be processed of the characters in the first video includes: Calculate the similarity between the hairstyle style of the user and the hairstyle style of the character in the first video as the similarity of the hairstyle style of the first video; Calculate the similarity between the hair length of the user and the hair length of the character in the first video as the hair length similarity of the first video; Calculate the similarity between the hair curl of the user and the hair curl of the character in the first video as the similarity of the hair curl of the first video; Calculate the similarity between the hair color of the user and the hair color of the character in the first video as the hair color similarity of the first video; The similarity between the user's bangs shape and the bangs shape of the character in the first video is calculated as the hair color similarity of the first video. 4. The method according to claim 3, characterized in that, according to the similarity of the face and the similarity of the hairstyle, screening the target video corresponding to the user in the video library, comprising: Selecting a video whose face similarity is greater than the first preset value in the video library to form a second video set; Traverse each video and use each video as the first video, if the hair style similarity of the first video, the similarity of the hair length of the first video, the similarity of the first video The similarity of hair curl, the similarity of the hair color of the first video, and the similarity of the bangs shape of the first video are all greater than the second preset value, then the first video is added to the first video. in the collection; After traversing each video, the intersection of the second video set and the first video set is taken, and the target video is selected from the videos in the intersection. 5. The method according to claim 3, wherein the similarity of the human face comprises the similarity of the human face of the first video, and the similarity of the human face of the first video is the human face of the user face similarity with the face in the first video; Screening the target video corresponding to the user in the video library according to the similarity of the face and the similarity of the hairstyle, including: Using the similarity calculation model to calculate the similarity of the face of the first video, the similarity of the hairstyle style of the first video, the similarity of the length of the hair of the first video, and the curl of the hair of the first video The similarity of the first video, the similarity of the hair color of the first video and the similarity of the bangs shape of the first video are weighted and summed to obtain the comprehensive similarity between the user and the characters in the first video; Traverse each video to obtain the comprehensive similarity between the user and the characters in each video; The target videos whose comprehensive similarity is greater than a third preset value are screened in the video library. 6. The method according to claim 5, wherein the method further comprises: training the similarity calculation model; The training of the similarity calculation model includes: Get a sample face image; Use a plurality of models to be trained to calculate the similarity and the face similarity between the sample face image and the hairstyle features of the corresponding characters in each video; For each model to be trained, select a video whose face similarity is greater than a first preset value and the hairstyle feature similarity is greater than a second preset value; Using a face swapping model to swap the sample face image to the face of the person corresponding to the selected video to obtain a plurality of target face images; Inputting the sample face image and the multiple target face images into a face recognition model, and calculating the similarity between the sample face image and the multiple target face images; Calculate the average value of the similarity of the multiple target face images corresponding to each model to be trained; The model to be trained corresponding to the highest average value is selected as the similarity calculation model. 7. The method of claim 1, wherein the method further comprises: The target video is recommended to the user. 8. A face matching device, wherein the device comprises: The acquisition module is used to acquire the face image of the user; an extraction module, used for extracting the facial features of the user from the facial image; The first calculation module is used to calculate the similarity between the user's face and the face in each video according to the user's face feature and the video face feature corresponding to each video in the pre-stored video library Spend; an identification module, used for identifying the gender feature of the user and the hairstyle feature of the user by using the face image; a second computing module, configured to calculate the hairstyle similarity between the hairstyle features of the user and the hairstyle features of the characters in each video; A selection module, configured to screen a target video corresponding to the user in the video library according to the face similarity and the hairstyle similarity, wherein the gender characteristics of the characters in the target video are related to the user gender characteristics are the same. 9. A computing device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory, when the computer program instructions are executed by the processor, the implementation as claimed in the claims The method of any one of 1-7. 10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, when the computer program instructions are executed by a processor, the method according to any one of claims 1-7 is implemented.

Images