Virtual image generation method, apparatus and device, and computer readable storage medium

Abstract

The invention discloses a virtual image generation method and device, equipment and a computer readable storage medium, and belongs to the technical field of communication. The method comprises the steps that generation request information, input by a first user, of a target virtual image is received, the generation request information of the target virtual image comprises information and a skill name of a first virtual image, the first virtual image is generated by a first model file, and the first virtual image has first response capacity for at least one instruction; the skill name is used for indicating the skill possessed by the target virtual image; generating a target virtual image according to a target model file generated by a first model file corresponding to the information of the first virtual image and a skill model file corresponding to the skill name, the target virtual image having a second response capability for at least one instruction; and providing the target virtual image to the first user. The target virtual image generated in a customized mode according to the generation request information of the target virtual image of the first user better meets the requirements of the first user.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a method, device, equipment and computer-readable storage medium for generating a virtual image. Background Art

[0002] With the development of communication technology, the products of communication technology are becoming more and more diverse and numerous, including virtual images. Different virtual images can be generated according to different user needs. Virtual images with different appearances can be generated according to the user's face pinching, motion capture and other methods, but the other characteristics of different virtual images are still relatively simple. Summary of the invention

[0003] The present application provides a method, device, apparatus and computer-readable storage medium for generating a virtual image to generate a virtual image that better meets user needs. The technical solution is as follows:

[0004] In a first aspect, a method for generating a virtual image is provided, which is used to generate a target virtual image using a first virtual image, and the method includes: receiving a target virtual image generation request information input by a first user, the target virtual image generation request information includes information of the first virtual image and a skill name, the first virtual image is generated by a first model file, the first virtual image has a first response capability for at least one instruction, and the skill name is used to indicate the skill possessed by the target virtual image; generating a target virtual image according to the target model file, the target model file is generated by the first model file corresponding to the information of the first virtual image and the skill model file corresponding to the skill name, and the target virtual image has a second response capability for at least one instruction; and providing the target virtual image to the first user. The target virtual image generated by customization according to the target virtual image generation request information of the first user better meets the needs of the first user.

[0005] In a possible implementation, the target virtual image generation request information also includes skill level information, and the skill level information is used to indicate the target virtual image's learning degree of the skill indicated by the skill name. Before generating the target virtual image according to the target model file, the method further includes: controlling the number of iterations of generating the target model file according to the skill level information, wherein the learning degree of the skill indicated by the skill level information is positively correlated with the number of iterations of generating the target model file. Controlling the number of iterations of generating the target model file can make the learning degree of the skill indicated by the skill name of the target virtual image corresponding to the target model file more in line with the needs of the first user, thereby ensuring the accuracy of the generated target model file.

[0006] In a possible implementation, before generating the target virtual image according to the target model file, the method further includes: receiving skill data input by the second user, where the skill data is used to indicate skill information corresponding to the skill name; and generating the skill model file according to the skill data. The skill model file is generated according to the skill data input by the second user, so that the generated skill model file is more targeted and better meets the needs of the second user.

[0007] In a possible implementation, the method further includes: receiving user data generated when the first user uses the target virtual image; updating the target model file according to the user data; generating an updated target virtual image according to the updated target model file; and providing the updated target virtual image to the first user. Updating the target virtual image according to the user data can make the updated target virtual image more in line with the needs of the first user.

[0008] In a possible implementation, before receiving the generation request information of the target virtual image input by the first user, the method further includes: receiving the generation request information of the first virtual image input by the first user, the generation request information of the first virtual image is used to indicate the characteristics of the first virtual image, the generation request information of the first virtual image includes at least one feature type and a feature description corresponding to at least one feature type, the feature type includes one or more of personality, language style, target group, field, and brand; based on an artificial intelligence (AI) model file, at least one feature type and a feature description corresponding to at least one feature type, a first model file is generated. Generating the first model file according to the generation request information of the first virtual image can make the first virtual image corresponding to the first model file have unique characteristics and style, which better meets the needs of the first user.

[0009] In a second aspect, a virtual image generation device is provided, which is used to generate a target virtual image using a first virtual image, and the device includes: a receiving module, which is used to receive target virtual image generation request information input by a first user, the target virtual image generation request information includes the first virtual image information and a skill name, the first virtual image is generated by a first model file, the first virtual image has a first response capability for at least one instruction, and the skill name is used to indicate the skills possessed by the target virtual image; a generating module, which is used to generate a target virtual image according to the target model file, the target model file is generated by the first model file corresponding to the first virtual image information and the skill model file corresponding to the skill name, and the target virtual image has a second response capability for at least one instruction; and a providing module, which is used to provide the target virtual image to the first user.

[0010] In one possible implementation, the generation request information of the target virtual image also includes skill level information, and the skill level information is used to indicate the degree of learning of the skill indicated by the skill name by the target virtual image. The generation module is also used to control the number of iterations of generating the target model file according to the skill level information, wherein the degree of learning of the skill indicated by the skill level information is positively correlated with the number of iterations of generating the target model file.

[0011] In a possible implementation, the generation module is further used to receive skill data input by a second user, where the skill data is used to indicate skill information corresponding to the skill name; and to generate a skill model file based on the skill data.

[0012] In a possible implementation, the receiving module is further used to receive user data generated when the first user uses the target virtual image; the generating module is further used to update the target model file according to the user data; based on the updated target model file, an updated target virtual image is generated; and the providing module is further used to provide the updated target virtual image to the first user.

[0013] In one possible implementation, the receiving module is further used to receive generation request information of a first virtual image input by a first user, the generation request information of the first virtual image is used to indicate features of the first virtual image, the generation request information of the first virtual image includes at least one feature type and a feature description corresponding to at least one feature type, the feature type includes one or more of personality, language style, target group, field, and brand; the generating module is further used to generate a first model file based on the artificial intelligence AI model file, at least one feature type, and a feature description corresponding to at least one feature type.

[0014] In a third aspect, a computing device cluster is provided, the computing device cluster comprising at least one computing device, each computing device comprising a processor and a memory; the processor of at least one computing device is used to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster executes any one of the virtual image generation methods of the first aspect above.

[0015] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes computer program instructions. When the computer program instructions are executed by a computing device cluster, the computing device cluster executes any one of the methods for generating a virtual image in the first aspect.

[0016] In a fifth aspect, a computer program (product) comprising instructions is provided. When the instructions are executed by a computing device cluster, the computing device cluster executes any one of the methods for generating a virtual image of the first aspect.

[0017] In a sixth aspect, a communication device is provided, the device comprising: a transceiver, a memory, and a processor. The transceiver, the memory, and the processor communicate with each other through an internal connection path, the memory is used to store instructions, the processor is used to execute the instructions stored in the memory to control the transceiver to receive signals and control the transceiver to send signals, and when the processor executes the instructions stored in the memory, the processor executes the method in the first aspect or any possible implementation of the first aspect.

[0018] Optionally, the number of the processors is one or more, and the number of the memories is one or more.

[0019] Optionally, the memory may be integrated with the processor, or the memory may be provided separately from the processor.

[0020] In the specific implementation process, the memory can be a non-transitory memory, such as a read-only memory (ROM), which can be integrated with the processor on the same chip or can be set on different chips. This application does not limit the type of memory and the setting method of the memory and the processor.

[0021] In a seventh aspect, a chip is provided, comprising a processor for calling and executing instructions stored in a memory from the memory, so that a communication device equipped with the chip executes any one of the methods for generating a virtual image in the first aspect.

[0022] In an eighth aspect, another chip is provided, comprising: an input interface, an output interface, a processor and a memory, wherein the input interface, the output interface, the processor and the memory are connected via an internal connection path, and the processor is used to execute the code in the memory. When the code is executed, the processor is used to execute any one of the virtual image generation methods in the above-mentioned first aspect.

[0023] It should be understood that the beneficial effects achieved by the technical solutions of the second to eighth aspects of the present application and the corresponding possible implementation methods can be referred to the above-mentioned technical effects of the first aspect and its corresponding possible implementation methods, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of an implementation scenario provided for an embodiment of the present application;

[0025] Figure 2 A schematic diagram of a flow chart of a method for generating a virtual image provided in an embodiment of the present application;

[0026] Figure 3A schematic diagram of a process of obtaining a first virtual image provided in an embodiment of the present application;

[0027] Figure 4 A schematic diagram of a process for obtaining a skill model file provided in an embodiment of the present application;

[0028] Figure 5 A schematic diagram of a process for generating a target model file provided in an embodiment of the present application;

[0029] Figure 6 A schematic diagram of the structure of a virtual image generation device provided in an embodiment of the present application;

[0030] Figure 7 A schematic diagram of the structure of a computing device provided in an embodiment of the present application;

[0031] Figure 8 A connection diagram of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0032] The terms used in the implementation section of this application are only used to explain the specific embodiments of this application and are not intended to limit this application.

[0033] With the development of communication technology, the application scope of communication technology is becoming wider and wider, and the products of communication technology are becoming more and more diverse and increasing in number. For example, virtual images are one of them. Virtual images can be trained based on common language and behavior models. For example, virtual images can be digital people, digital animals, or digital cartoon characters. Different virtual images can be applied to different scenarios. In various scenarios, the virtual images need to be driven to realize various functions such as performance or interaction of the virtual images.

[0034] For example, the virtual image can be driven by the person in the virtual image. The person in the virtual image driving the virtual image refers to the person who plays the core control role in the virtual image system, and is the actor or actor of the virtual image. In this scenario of driving the virtual image, the performance and action of the person in the virtual image can be captured and recorded by motion capture driving technology, using sensors and cameras and other equipment, and the virtual image can be driven by anchor points or skeleton matching to make the virtual image consistent with the action of the person in the virtual image, so as to complete the performance and interactive behavior of the virtual image, thereby improving the realism of the virtual image and ensuring the interactive experience of the user with the virtual image. In this driving method, all the interactive behaviors and emotional expressions of the virtual image are given by the person in the virtual image, and the performance and actions of the person in the virtual image can directly affect the interactive behavior and emotional expression of the virtual image. However, this driving method is equivalent to changing the image and timbre of humans in the digital world, without truly giving skills, knowledge and emotions to the virtual image.

[0035] Alternatively, the virtual image can also be driven by voice or text driving technology. For example, the lip shape corresponding to the voice that the virtual image needs to output can be determined, and the virtual image can be driven by the lip shape to change the lip shape and make a sound, or the virtual image can be driven to read the text and output the voice corresponding to the text to realize the voice of the virtual image. Alternatively, the virtual image can also be driven by interactive rule driving technology. For example, the speech script and interactive rules can be preset so that when a specific event corresponding to the speech script occurs or the virtual image encounters a scene corresponding to the interactive rule, the virtual image can be triggered to perform corresponding actions based on the interactive rules and the speech script to achieve interaction. For example, when the virtual image encounters a scene corresponding to the interactive rule, the reply text set in the interactive rule can be inserted into the virtual image's playback script, and then the virtual image can be driven to interact through the text in the playback script. In a possible implementation, driving the virtual image by voice or text driving technology and by interactive rule driving technology can be considered as driving the virtual image by artificially set instructions. Voice, text and interactive rules can all be used as instructions for driving the virtual image. The virtual image driven by the instruction can be called an instructional virtual image.

[0036] However, the virtual images driven by the above-mentioned multiple technologies are relatively simple and difficult to cope with complex scenarios such as live broadcasting, intelligent customer service, and intelligent interaction. In addition, since the voice, action, expression, and reply rules of the virtual image are all preset, the virtual image cannot express rich emotions and personality characteristics. This makes it difficult for the virtual image to make a distinctive response when facing complex situations and problems, thus affecting the performance of the virtual image. In addition, since the virtual images driven by the above two technologies are not customized according to the needs of different industries, they do not have specific knowledge and skills in the industry, and different industries require different knowledge and skills. For example, in the fields of medical care, law, and finance, virtual images are required to have professional knowledge and skills to be competent for their work. Therefore, the virtual images driven by the above-mentioned multiple technologies cannot meet the specific needs of different industries, limiting the applicability of virtual images in different application scenarios. In addition, virtual images with different appearances can be generated according to the user's face pinching, motion capture, etc., but the other characteristics of different virtual images are still relatively single.

[0037] The embodiment of the present application provides a method for generating a virtual image, which can enable the generated virtual image to master professional skills in a specific field and have rich emotions and personality characteristics, so that the generated virtual image can better meet application requirements. Figure 1, showing an implementation scenario diagram of the method. The implementation scenario includes a first generation module 11, a second generation module 12, a third generation module 14 and an adjustment module 13. The third generation module 14 can be connected to the first generation module 11, the second generation module 12 and the adjustment module 13 respectively, and the adjustment module 13 can also be connected to the first generation module 11 and the second generation module 12. The first generation module 11, the second generation module 12, the third generation module 14 and the adjustment module 13 can be integrated into the same device or into different devices. The implementation scenario can be applied to scenarios such as the metaverse, intelligent customer service, live broadcast, conference chat, webinars and interactive classrooms in the field of artificial intelligence to generate virtual images and build the intelligent capabilities of virtual images.

[0038] See also Figure 2 , shows a schematic flow chart of a method for generating a virtual image provided by an embodiment of the present application, the method includes but is not limited to the following S201 to S203, the method can be applied to Figure 1 In the implementation scenario shown.

[0039] S201, receiving a target virtual image generation request information input by a first user, the target virtual image generation request information including the first virtual image information and a skill name, the first virtual image is generated by a first model file, the first virtual image has a first response capability to at least one instruction, and the skill name is used to indicate the skill possessed by the target virtual image.

[0040] The first virtual image may also be referred to as a digital human miniature. The first virtual image may be a virtual image selected by the first user from a plurality of virtual images, or may be a virtual image generated according to the needs of the first user. The information of the first virtual image may be a model file corresponding to the first virtual image, a file name of the model file corresponding to the first virtual image, or an identifier (ID) of the first virtual image.

[0041] The first virtual image is generated based on the first model file. Therefore, the first model file can be obtained before the first virtual image is generated. Exemplarily, before receiving the generation request information of the target virtual image input by the first user, the generation request information of the first virtual image input by the first user can be received first. The generation request information of the first virtual image is used to indicate the characteristics of the first virtual image. The generation request information of the first virtual image includes at least one feature type and a feature description corresponding to at least one feature type. The feature type includes one or more of personality, language style, target group, field, and brand. The feature description can also be called a tonality description or tonality description data. Afterwards, the first model file can be generated based on the AI ​​model file, at least one feature type, and a feature description corresponding to at least one feature type. The AI ​​model file can be a general model or a general large model. After the first model file is generated, the first virtual image can be generated based on the first model file. For example, see Figure 3 , showing a schematic diagram of a process of generating a first virtual image. The first generating module can receive the generating request information of the first virtual image input by the first user, and after obtaining the AI ​​model file, based on at least one feature type and at least one feature description corresponding to the feature type included in the generating request information of the first virtual image, complete the adjustment of the AI ​​model file to obtain the first model file. Afterwards, the first generating module generates the first virtual image based on the first model file.

[0042] In a possible implementation, the generation request information of the first virtual image may include a customized database or a link to a customized database provided by a user, and the customized database provided by the user stores at least one feature type and a feature description corresponding to at least one feature type. The customized database provided by the user may be a database for the first virtual image, and the customized database may store multiple feature types and feature descriptions for indicating the features of the first virtual image. Exemplarily, the multiple feature types and feature descriptions for indicating the features of the first virtual image may include tonality description data for describing the unique style of the first virtual image, and may also include brand knowledge such as brand tonality, brand history, and brand characteristics. In this implementation, some or all of the feature types and feature descriptions in the customized database may be used as at least one feature type and a feature description corresponding to at least one feature type, and the AI ​​model file may be updated to obtain a first model file, and then based on the first model file, a first virtual image having both a unique style and brand knowledge is obtained.

[0043] In another possible implementation, at least one feature type of the person in the image and the feature description corresponding to at least one feature type can be used as at least one feature type of the first virtual image and the feature description corresponding to at least one feature type. In this implementation, the request information for generating the first virtual image can include a file that can reflect the feature type of the person in the image, such as a video, picture or cartoon that can reflect the feature type of the person in the image, and then the feature description of the feature type of the person in the image can be determined based on the file that can reflect the feature type of the person in the image. Afterwards, based on at least one feature type of the person in the image, the feature description corresponding to at least one feature type of the person in the image, and the AI ​​model file, a first model file can be generated, and based on the first model file, a first virtual image can be generated. For example, a video of at least one live broadcast hosted by anchor A can be obtained, anchor A can be used as the person in the video, and the language habits, personality, and micro-expressions of anchor A reflected in the at least one live broadcast video hosted by anchor A can be used as at least one feature type of the first virtual image. Then, based on the language habits, personality, and micro-expressions of anchor A, a feature description of at least one feature type of the first virtual image is determined, the AI ​​model file is updated to obtain the first model file, and then based on the first model file, a first virtual image with the same features as anchor A is obtained.

[0044] In addition, the first virtual image has a first response capability for at least one instruction, that is, after receiving the skill instruction corresponding to the skill name, the first virtual image can output a corresponding response. The embodiment of the present application does not limit the strength of the first response capability of the first virtual image for at least one instruction.

[0045] In the embodiment of the present application, different first virtual images can be generated based on the same AI model file and different first virtual image generation request information, without the need to model multiple AI models for different user needs, thereby reducing the time and cost required to generate the first virtual image. In addition, the first virtual image generated based on the first virtual image generation request information has a unique personality and a unique performance style as well as rich brand knowledge. The first virtual image can better understand and respond to users, and can generate behaviors and performances with specific personality traits based on information such as brand product information and brand promotional materials, so as to better interact with fans of the brand.

[0046] S202, generating a target virtual image according to the target model file, wherein the target model file is generated by a first model file corresponding to information of the first virtual image and a skill model file corresponding to the skill name, and the target virtual image has a second response capability to at least one instruction.

[0047] In a possible implementation, before generating a target virtual image according to the target model file, a skill model file corresponding to the skill name may be obtained first. The present application embodiment does not limit the method for obtaining a skill model file corresponding to a skill name. Exemplarily, skill data input by a second user may be received, the skill data being used to indicate skill information corresponding to the skill name; and a skill model file may be generated according to the skill data.

[0048] The skill data may be data or domain data corresponding to at least one skill, such as video, picture, point cloud, trajectory, human body structure, text description, etc. Taking the dance field as an example, music A corresponds to dance A, and the skill data input to the AI ​​model file may be: skill data indicating to dance dance A according to the dance instructions corresponding to music A, such as a dance video with music A as the background music.

[0049] The first user and the second user may be the same user or different users. Taking the case where the first user and the second user are different users as an example, the first user may be a user who has a need to use the target virtual image, and the second user may be a user other than the first user who can create a skill model file, such as a developer or a third party. Exemplarily, a platform for training models open to users may be constructed, and the platform may be similar to a module community for games. The platform can provide users with a model training function for training skill model files. The second user can input skill data through the model training function provided by the platform, and train the multimodal AI model file according to the skill data to obtain a skill model file. The skill model corresponding to the skill model file has a reference response capability for at least one instruction.

[0050] In a possible implementation, the second user can determine the skills that the expected skill model has according to the needs of the second user or the second user's functional positioning of the skill model, thereby determining the skill data corresponding to the skill. For example, if the second user expects the skill model to have the skill of hosting live broadcasts, the historical live broadcast words, live broadcast speech, live broadcast planning scripts, common words, live broadcast strategies and live broadcast hosting skills of other anchors can be used as skill data, and the skill data can be input, so that the AI ​​model file can be updated according to the skill data input by the second user, and a skill model file with the skill of hosting live broadcasts is obtained, or a skill model file with reference response capability for at least one instruction in the field of hosting live broadcasts is obtained. The skill model has reference response capability for at least one instruction in the field of hosting live broadcasts, which may mean that the skill model can understand the basic knowledge in the field of hosting live broadcasts, and for instructions such as arranging live broadcasts, describing products, and interacting with the audience, the skill model can arrange and plan live broadcasts based on the reference response capability, describe products with common live broadcast terms according to anchor skills, and interact with the audience according to live broadcast strategies. The skill model corresponding to the skill model file generated based on the skill data has a high level of professionalism in the field corresponding to the skill data.

[0051] Alternatively, in another possible implementation, the skill data may be the skill data corresponding to the response of the person in the video to at least one instruction. Exemplarily, after obtaining the AI ​​model file, the response of the person in the video to at least one instruction may be input into the AI ​​model corresponding to the AI ​​model file to update the AI ​​model, thereby updating the AI ​​model file according to the skill data and obtaining the skill model file. For example, by inputting at least one live video to the AI ​​model, the host of the live broadcast may be used as the person in the video to be imitated, so that the AI ​​model file is updated according to the host's behavior and language to at least one instruction, and a skill model file is obtained.

[0052] Alternatively, in another possible implementation, after obtaining the AI ​​model file, learning instructions and the skill name required by the second user can be input into the AI ​​model, so that the AI ​​model can autonomously obtain the skill data corresponding to the skill name required by the user in the network based on the learning instructions, and update it based on the skill data to obtain the skill model file.

[0053] In addition, skill data may also include skill enhancement data, which may also be referred to as specific data. Skill enhancement data may be specific data for a certain skill. Inputting skill data including skill enhancement data into the AI ​​model can make the skill model file obtained based on the skill data update have a higher professional level, that is, make the skill model's reference response capability for at least one instruction stronger. Exemplarily, after inputting skill data into the AI ​​model, skill enhancement data may also be input into the AI ​​model based on the needs of the second user to enhance a certain skill of the skill model. For example, based on the user's need to enhance sentiment analysis skills, skill enhancement data such as sentiment vocabulary and sentiment analysis samples may be input into the AI ​​model to improve the level of sentiment analysis of the skill model corresponding to the skill model file obtained based on the AI ​​model file update, so as to enhance the skill model's emotional expression ability and audience interaction effect during the live broadcast.

[0054] For example, see Figure 4 , showing a schematic diagram of the process of obtaining a skill model file. The second generation module includes a skill generation module and a skill enhancement module. The skill generation module can also be called a business domain model generation module, and the skill enhancement module can also be called a skill assembly module. The skill generation module can obtain the AI ​​model file and skill data, and input the skill data into the AI ​​model file to obtain the business domain initial model file, that is, add skill data on the basis of the AI ​​model file to adjust the AI ​​model file to obtain the business domain initial model file. The business domain initial model corresponding to the business domain initial model file can understand the basic knowledge such as concepts and terms in the vertical field set by the second user, can help solve natural language processing problems in specific fields or specific tasks, and improve the performance and effect of the output response to instructions.

[0055] The skill enhancement module can input the acquired skill enhancement data into the initial model file of the business domain, so that the skills possessed by the skill model file are more professional, thereby making the skill model corresponding to the skill model file more intelligent, personalized and diversified, and improving the ability of the skill model to respond to at least one instruction, that is, improving the reference response capability of the skill model file.

[0056] The process of updating the AI ​​model file according to the skill data is also the process of training the AI ​​model file. The method of training the AI ​​model file can be divided into supervised training and unsupervised training. Unsupervised training refers to the process of not supervising the learning results of the AI ​​model file after the skill data is input into the AI ​​model file, that is, in the process of training the AI ​​model file, no requirements are made for the interim results of the training, nor are requirements made for the mastery of at least one skill by the skill model file. Supervised training refers to the process of inputting skill data into the AI ​​model file, and also inputting requirements for all or part of the training results into the AI ​​model file, that is, inputting requirements for the mastery of all or part of the skills in at least one skill by the skill model file, so as to realize the concrete definition of various skills of the skill model file. In the scenario of supervised training, the AI ​​model file can continuously test its own training results during the training according to the skill data. If the mastery of a certain skill meets the requirements, the training for the skill is stopped, and the training for other skills is continued. If the mastery of a certain skill does not meet the requirements, the training for the skill is continued.

[0057] Inputting different skill data into the AI ​​model file can also be considered as integrating or loading different skill modules and knowledge bases into the AI ​​model file. The AI ​​model file can select and combine the skill data according to the needs of the second user, and update it according to the selected and combined skill data to obtain a skill model file with at least one skill. After acquiring the skill model file, the embodiment of the present application can also inject new skill data into the skill model file, continuously optimize and upgrade the responsiveness of the skill model file, realize the continuous learning and evolution of the skill model file, and continuously improve the skill level, intelligence and expressiveness of the skill model file.

[0058] In a possible implementation, the generation request information of the target virtual image also includes skill level information, and the skill level information is used to indicate the target virtual image's learning degree of the skill indicated by the skill name. In this implementation, before generating the target virtual image according to the target model file, the number of iterations of generating the target model file can also be controlled according to the skill level information, wherein the learning degree of the skill indicated by the skill level information is positively correlated with the number of iterations of generating the target model file. Exemplarily, the reception of skill level information can be achieved by receiving the percentage or grade of the learning degree of the indicated skill input by the user. For example, the user can input the percentage value of the learning degree according to the required target virtual image's learning degree of the skill, or complete the input of the percentage of the learning degree by dragging the percentage bar of the learning degree. Or the input of the level of the learning degree can be completed by entering the level value corresponding to the learning degree or selecting different levels.

[0059] After receiving the skill level information, the number of iterations for generating the target model file can be determined based on the skill level information. The higher the degree of skill learning of the target virtual image indicated by the skill level information, the more iterations for generating the target model file; the lower the degree of skill learning of the target virtual image indicated by the skill level information, the fewer iterations for generating the target model file. After determining the number of iterations for generating the target model file, the target model file can be iteratively generated based on the first model file and the skill model file within the determined number of iterations for generating the target model file.

[0060] See also Figure 5 , showing a schematic diagram of a process of generating a target model file. The third generation module can generate a target model file based on the acquired skill model file and the first model file. The embodiment of the present application does not limit the method of generating a target model file based on the skill model file and the first model file. Below, taking two different cases A1 and A2 as examples, the process of generating a target model file is described.

[0061] In case A1, the first virtual image corresponding to the first model file has a first response capability for at least one instruction. Since the first virtual image is a virtual image that has not been updated according to the skill data, the first response capability of the first virtual image is the initial response capability for at least one instruction. For example, at least one instruction is an unknown instruction for the first model file. If the first virtual image does not respond to the unknown instruction, it can be considered that the first response capability is not outputting a response for at least one instruction. Alternatively, if the first virtual image responds partially or arbitrarily to the unknown instruction according to its own understanding, it can be considered that the first response capability is outputting a partial response or an arbitrary response to at least one instruction. For example, if at least one instruction includes dancing dance A when listening to music A, the instruction is an unknown instruction for the first virtual image. If the first virtual image does not respond to the unknown instruction, the first response capability is the response capability of not making any dance movements when the first virtual image hears music A. If the first virtual image responds partially or arbitrarily to the unknown instruction according to its own understanding, the first response capability is the response capability of the first virtual image to answer that it will not dance or make irregular dance movements when it hears music A.

[0062] In this case, generating a target model file based on the skill model file and the first model file corresponding to the first virtual image can include inputting at least one instruction into the skill model corresponding to the skill model file to obtain a first response output by the skill model based on the at least one instruction and a reference response capability; and inputting at least one instruction into the first virtual image to obtain a second response output by the first virtual image based on the at least one instruction and the first response capability.

[0063] By inputting at least one same instruction to the skill model and the first virtual image, the first response and the second response output by the skill model file and the first virtual image to the instruction can be obtained respectively. Since the skill model file is a model obtained based on the skill data update, and the first virtual image is obtained without the skill data update, or in other words, the skill model has the skill corresponding to the skill name, and the first virtual image does not have the skill corresponding to the skill name, it can be considered that the reference response capability of the skill model to at least one instruction is different from the first response capability of the first virtual image to at least one instruction, that is, the first response and the second response output by the skill model and the first virtual image to the same instruction are different. In one possible implementation, the reference response capability is stronger than the first response capability, that is, the first response output by the skill model to at least one instruction is more accurate than the second response output by the first virtual image to at least one instruction.

[0064] In an exemplary implementation, the first response output by the skill model file for at least one instruction can be used as a standard correct response, and the similarity between the first response and the second response can reflect the correctness of the second response, that is, the similarity between the first response and the second response can reflect the degree of learning of the skill by the first virtual image. The higher the similarity between the first response and the second response, the closer the second response is to the standard correct response, that is, the higher the accuracy of the second response, and the higher the degree of learning of the skill by the first virtual image. The lower the similarity between the first response and the second response, the less close the second response is to the standard response, that is, the lower the accuracy of the second response, and the lower the degree of learning of the skill by the first virtual image. For example, for the skill of dancing, at least one instruction includes dancing dance A when music A is heard, and the first response is dancing dance A when music A is heard, and the dance moves are correct, the dance moves are in place, and the dance moves are consistent with the rhythm of the music. If the second response is dancing dance A when music A is heard and the dance moves are correct, it can be considered that the similarity between the first response and the second response is 30%. If the second response is to dance A when listening to music A and the dance moves are correct and in place, the similarity between the first response and the second response can be considered to be 50%. If the second response is to dance A when listening to music A and the dance moves are correct, in place and in line with the rhythm of the music, the similarity between the first response and the second response can be considered to be 100%.

[0065] After determining the similarity between the first response and the second response, the similarity between the first response and the second response may be compared with a first similarity threshold. The first similarity threshold is determined based on the degree of learning of the skill indicated by the skill level information. For example, if the degree of learning of the skill indicated by the skill level information is 50%, the first similarity threshold may be set to 50%, and if the degree of learning of the skill indicated by the skill level information is level three, the first similarity threshold may be set to 30%.

[0066] If the similarity between the first response and the second response is less than the first similarity threshold, it means that the first response capability of the first virtual image to at least one instruction does not meet the second response capability expected by the user. In this case, the first virtual image can continue to be trained. Exemplarily, the first update instruction and the first response can be input into the first model file to obtain a first updated model file. A first updated virtual image is generated according to the first updated model file, and at least one instruction is input into the first updated virtual image to obtain a third response output by the first updated virtual image based on the at least one instruction.

[0067] If the similarity between the third response and the first response is still less than the first similarity threshold, it means that the response capability of the first updated virtual image to at least one instruction does not meet the response capability expected by the user, and the first update instruction and the first response need to be continuously input into the first updated model file to implement a secondary update of the first updated model file, thereby enhancing the response capability of the first updated virtual image to at least one instruction. Optionally, if during the training process, the similarity between the response output by the first updated virtual image and the first response is less than the first similarity threshold, it can be considered that the response output by the first updated virtual image is an erroneous response, and a penalty instruction can be input into the first updated virtual image to punish the first updated virtual image. The first updated virtual image is updated multiple times until the multiple updated response capability of the virtual image to at least one instruction meets the response capability expected by the user, and the model file obtained after the multiple updates can be used as the target model file. That is, until the similarity between the response of the virtual image corresponding to the model file obtained after multiple updates to at least one instruction output and the first response is equal to the first similarity threshold, it can be considered that the skill learning degree of the virtual image corresponding to the model file obtained after multiple updates reaches the skill learning degree of the target virtual image indicated by the skill level information, and the model file obtained after multiple updates can be used as the target model file.

[0068] The above process of obtaining a target model file that meets user needs through multiple iterative updates based on the skill model file and the first model file can be called the process of knowledge distillation or skill distillation of the skill model on the first virtual image. In the above process, the skill model is used as a teacher and the first virtual image is used as a student to transfer skills through a teaching and learning model. For example, if the user wants the target model file to load dance skills, the skill model file with dance skills can be used as a teacher and the first virtual image as a student, so that the skill model file can assemble the dance skills to the first virtual image through the skill distillation method, and then obtain the target model file.

[0069] Afterwards, a target virtual image can be generated according to the target model file. Because the first virtual image itself has a unique style and knowledge, different first virtual images perform differently in the process of learning the same skill. Optionally, the longer the first virtual image learns the skill, the better the first virtual image's mastery of the skill, the closer the first virtual image's performance in a certain skill is to the standard correct performance, and the closer the first virtual image's response ability to at least one instruction is to the reference response ability, the weaker the style of the first virtual image. Therefore, the embodiment of the present application sets a first similarity threshold, clarifies the learning degree of the first virtual image to the skill, and controls the number of iterations of the target model file generation, so that the target virtual image corresponding to the target model file generated based on the first model file and the skill model file has both skills and a unique style, that is, the target virtual image's response to at least one instruction is both characteristic and accurate, so that the target virtual image is not only a virtual image with a unique image, but also makes the behavior of the target virtual image iconic and influential, thereby protecting the intellectual property (IP) of the target virtual image.

[0070] In case A2, a comprehensive model file can be first obtained based on the skill model file and the first model file, and then a target model file can be obtained based on the comprehensive model file. Among them, the comprehensive virtual image corresponding to the comprehensive model file has a third response capability for at least one instruction, and has the same characteristics as the first virtual image. The embodiment of the present application does not limit the method for obtaining a comprehensive model file based on the skill model file and the first model file. Exemplarily, the underlying code corresponding to the skill model file and the underlying code corresponding to the first model file can be spliced, and the underlying code obtained is the underlying code corresponding to the comprehensive model file. The comprehensive model file can be obtained by compiling the underlying code corresponding to the comprehensive model file.

[0071] Alternatively, at least one feature type and at least one feature description corresponding to the feature type may be input into the skill model file, so that the skill model file is updated according to the at least one feature type and the feature description corresponding to the at least one feature type, so that the updated skill model file also has the same features as the first virtual image, and the skill model file updated according to the at least one feature type and the feature description corresponding to the at least one feature type is a comprehensive model file.

[0072] Alternatively, the skill data may be input into the first model file so that the first model file is updated according to the skill data, thereby obtaining a comprehensive model file having the same features as the first virtual image and the third response capability.

[0073] In a possible implementation, the process of obtaining the target model file based on the comprehensive model file can be divided into different situations. Below, taking situation A21 and situation A22 as examples, different situations are described.

[0074] In case A21, the third response capability is different from the target response capability expected by the user. In this case, at least one instruction can be input into the comprehensive model file to obtain a fourth response output by the comprehensive model file based on the at least one instruction and the third response capability; at least one instruction can be input into the first model file to obtain a second response output by the first model file based on the at least one instruction and the first response capability; and based on the similarity between the fourth response and the second response, the target model file is generated based on the comprehensive model file.

[0075] By inputting at least one same instruction to the comprehensive model and the first virtual image, the fourth response and the second response output by the comprehensive model and the first virtual image to the instruction can be obtained respectively. The third response capability of the comprehensive model to the at least one instruction is different from the first response capability of the first virtual image to the at least one instruction. In a possible implementation, the third response capability is stronger than the first response capability, that is, the fourth response output by the comprehensive model to the at least one instruction is more accurate than the second response output by the first virtual image to the at least one instruction.

[0076] In an exemplary implementation, the virtual image or the comprehensive model corresponding to the comprehensive model file can be used as a teacher, and the first virtual image can be used as a student. The virtual image corresponding to the comprehensive model file performs knowledge distillation on the first virtual image, and the target model file is obtained through multiple iterations. The process of performing knowledge distillation on the first virtual image by the virtual image corresponding to the comprehensive model file can refer to the process of performing knowledge distillation on the first virtual image by the skill model in situation A1, which will not be repeated here.

[0077] In case A22, the comprehensive model file can be used as the target model file. For example, if the third response capability of the virtual image corresponding to the comprehensive model file meets the response capability expected by the user, the comprehensive model file can be directly used as the target model file, and the first model file is no longer updated according to the comprehensive model file. This method of determining the first target model file can not only ensure that the target model file obtained meets the needs of the user, but also improve the efficiency of obtaining the target model file.

[0078] The embodiment of the present application obtains a target model file by updating based on the skill model file and the first model file, so that the target virtual image corresponding to the target model file can acquire skills and maintain a unique style, and different skills can make the target virtual image adapt to different application scenarios.

[0079] S203: Provide the target virtual image to the first user.

[0080] After the target virtual image is generated, the target virtual image can be provided to the first user. The first user can apply the first virtual image to the corresponding scene. For example, if the skill possessed by the target virtual image is the skill of hosting a live broadcast, the first user can apply the target virtual image to the live broadcast scene.

[0081] The first user can record the behavior and performance of the target virtual image during the application process, and organize user data based on the behavior and performance of the target virtual image during the application process. The user data can be, for example, live broadcast scripts, text data, motion capture files, real-life action videos, or the appearance of the virtual image set by the first user. In a possible implementation, see Figure 1 The adjustment module 13 can continuously adjust the target virtual image according to the user data. For example, the user data generated when the first user uses the target virtual image can be received; the target model file can be updated according to the user data; an updated target virtual image can be generated according to the updated target model file; and the updated target virtual image can be provided to the first user.

[0082] The embodiment of the present application does not limit the method for receiving the user data generated when the first user uses the target virtual image. For example, the user data collected from the first user can be received with the permission of the first user, or the user data can be received through procurement.

[0083] After receiving the user data, the user data can be input into the target model file corresponding to the target virtual image, the target model file can be updated to obtain an updated target model file, and an updated target virtual image can be generated according to the updated target model file, and the updated target virtual image can be provided to the first user again. The updated target virtual image is closer to the style expected by the first user, or can remember the latest product data of the first user. This implementation method allows users to participate in the update process of the target virtual image, so that the target virtual image can have a better grasp of the user's products and the user's expected style during the continuous update process, so as to better meet the needs of the user, that is, the target virtual image can be customized according to different user needs, so as to better adapt to various application scenarios. In addition, since the target virtual image can be continuously iterated and updated, the performance style of the target virtual image will also change over time, so as to maintain freshness and attractiveness.

[0084] The present application receives the information of the first virtual image and the skill name included in the generation request information of the target virtual image of the first user, and customizes and generates the target virtual image with skills corresponding to the skill name according to the first model file corresponding to the information of the first virtual image, so that the target virtual image better meets the needs of the first user.

[0085] In addition, a target virtual image is generated according to the characteristics of the first virtual image, making the target virtual image more distinctive, giving the target virtual image a unique style, personality and soul, as well as the ability to perceive, think and respond, realizing the virtual image's thousand people and thousand souls.

[0086] The above describes the method for generating a virtual image provided by the embodiment of the present application. Corresponding to the above method, the embodiment of the present application also provides a device for generating a virtual image. The device is used to generate a virtual image by Figure 6 The modules shown above perform the Figure 2 As shown in Figure 6 As shown, the virtual image generation device provided in the embodiment of the present application includes the following modules.

[0087] A receiving module 601 is used to receive a target virtual image generation request information input by a first user, the target virtual image generation request information includes the first virtual image information and a skill name, the first virtual image is generated by a first model file, the first virtual image has a first response capability for at least one instruction, and the skill name is used to indicate the skill possessed by the target virtual image; a generating module 602 is used to generate a target virtual image according to the target model file, the target model file is generated by a first model file corresponding to the first virtual image information and a skill model file corresponding to the skill name, and the target virtual image has a second response capability for at least one instruction; a providing module 603 is used to provide the target virtual image to the first user.

[0088] In one possible implementation, the generation request information of the target virtual image also includes skill level information, and the skill level information is used to indicate the degree of learning of the skill indicated by the skill name by the target virtual image. The generation module 602 is also used to control the number of iterations of generating the target model file according to the skill level information, wherein the degree of learning of the skill indicated by the skill level information is positively correlated with the number of iterations of generating the target model file.

[0089] In a possible implementation, the generation module 602 is further used to receive skill data input by a second user, where the skill data is used to indicate skill information corresponding to a skill name; and to generate a skill model file according to the skill data.

[0090] In a possible implementation, the receiving module 601 is further used to receive user data generated when the first user uses the target virtual image; the generating module 602 is further used to update the target model file according to the user data; based on the updated target model file, generate an updated target virtual image; and the providing module 603 is further used to provide the updated target virtual image to the first user.

[0091] In one possible implementation, the receiving module 601 is further used to receive generation request information of a first virtual image input by a first user, the generation request information of the first virtual image is used to indicate features of the first virtual image, the generation request information of the first virtual image includes at least one feature type and a feature description corresponding to at least one feature type, the feature type includes one or more of personality, language style, target group, field, and brand; the generating module 602 is further used to generate a first model file based on the artificial intelligence AI model file, at least one feature type, and a feature description corresponding to at least one feature type.

[0092] It should be understood that the above Figure 6 The device provided has the following beneficial effects when achieving its function: Figure 2The beneficial effects of the virtual image generation method provided are the same and will not be described in detail here. Figure 6 When the device provided realizes its functions, only the division of the above-mentioned functional modules is used as an example for illustration. In practical applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0093] The present application also provides a computing device 700. Figure 7 As shown, computing device 700 includes: bus 702, processor 704, memory 706 and communication interface 708. Processor 704, memory 706 and communication interface 708 communicate through bus 702. Computing device 700 can be a server or a terminal device. It should be understood that the present application does not limit the number of processors and memories in computing device 700.

[0094] The bus 702 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 The bus 702 may include a path for transmitting information between various components of the computing device 700 (eg, the memory 706, the processor 704, and the communication interface 708).

[0095] The processor 704 may include any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0096] The memory 706 may include a volatile memory, such as a random access memory (RAM). The processor 704 may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid state drive (SSD).

[0097] The memory 706 stores executable program codes, and the processor 704 executes the executable program codes to respectively implement the functions of the aforementioned receiving module, generating module, and providing module, thereby implementing the method for generating a virtual image. That is, the memory 706 stores instructions for executing the method for generating a virtual image.

[0098] The communication interface 708 uses a transceiver module such as, but not limited to, a network interface card or a transceiver to implement communication between the computing device 700 and other devices or communication networks.

[0099] The embodiment of the present application also provides a computing device cluster. The computing device cluster includes at least one computing device. The computing device can be a server, such as a central server, an edge server, or a local server in a local data center. In some embodiments, the computing device can also be a terminal device such as a desktop computer, a laptop computer, or a smart phone.

[0100] Optionally, the structure of at least one computing device included in the computing device cluster can be found in Figure 7 The computing device 700 is shown. The memory 706 in one or more computing devices 700 in the computing device cluster may store the same instructions for executing the method for generating a virtual image.

[0101] In some possible implementations, the memory 706 of one or more computing devices 700 in the computing device cluster may also store partial instructions for executing the method for generating a virtual image. In other words, the combination of one or more computing devices 700 may jointly execute the instructions for executing the method for generating a virtual image.

[0102] It should be noted that the memory 706 in different computing devices 700 in the computing device cluster may store different instructions, which are respectively used to execute part of the functions of the virtual image generation device. That is, the instructions stored in the memory 706 in different computing devices 700 may implement the functions of one or more modules among the receiving module, the generating module and the providing module.

[0103] In some possible implementations, one or more computing devices in the computing device cluster may be connected via a network, which may be a wide area network or a local area network. Figure 8 A possible implementation is shown. Figure 8 As shown, two computing devices 800A and 800B are connected via a network. Specifically, the network is connected via a communication interface in each computing device. In this type of possible implementation, computing devices 800A and 800B include a bus 802, a processor 804, a memory 806, and a communication interface 808. The memory 806 in the computing device 800A stores instructions for executing the functions of the receiving module. At the same time, the memory 806 in the computing device 800B stores instructions for executing the functions of the generating module and the providing module.

[0104] It should be understood that Figure 8 The functions of the computing device 800A shown in FIG. 8 may also be completed by multiple computing devices 800. Similarly, the functions of the computing device 800B may also be completed by multiple computing devices 800.

[0105] The embodiment of the present application also provides a communication device, which includes: a transceiver, a memory, and a processor. The transceiver, the memory, and the processor communicate with each other through an internal connection path, the memory is used to store instructions, and the processor is used to execute the instructions stored in the memory to control the transceiver to receive signals and control the transceiver to send signals, and when the processor executes the instructions stored in the memory, the processor executes the method for generating a virtual image.

[0106] It should be understood that the above processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. It is worth noting that the processor may be a processor supporting the advanced RISC machines (ARM) architecture.

[0107] Further, in an optional embodiment, the memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. The memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0108] The memory may be a volatile memory or a nonvolatile memory, or may include both volatile and nonvolatile memory. Among them, the nonvolatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM) and direct RAM bus random access memory (DR RAM).

[0109] The present application also provides a computer program (product) containing instructions. The computer program (product) may be software or a program (product) containing instructions that can be run on a computing device or stored in any available medium. When the computer program (product) is run on at least one computing device, the at least one computing device executes the method for generating a virtual image.

[0110] The embodiment of the present application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that can be stored by a computing device or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk). The computer-readable storage medium includes instructions that instruct the computing device to execute the method for generating a virtual image.

[0111] An embodiment of the present application also provides a chip, including a processor, for calling and executing instructions stored in a memory from the memory, so that a communication device equipped with the chip executes any of the above-described methods for generating a virtual image.

[0112] An embodiment of the present application also provides another chip, including: an input interface, an output interface, a processor and a memory, wherein the input interface, the output interface, the processor and the memory are connected via an internal connection path, and the processor is used to execute the code in the memory. When the code is executed, the processor is used to execute any of the virtual image generation methods described above.

[0113] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in this application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integration. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk), etc.

[0114] In order to clearly illustrate the interchangeability of hardware and software, the steps and components of each embodiment have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person of ordinary skill in the art may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0115] The computer program code for realizing the method for the embodiment of the present application can be written in one or more programming languages. These computer program codes can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable rule-finding device, so that the program code, when executed by a computer or other programmable rule-finding device, causes the function / operation specified in the flow chart and / or block diagram to be implemented. The program code can be executed completely on a computer, partially on a computer, as an independent software package, partially on a computer and partially on a remote computer or completely on a remote computer or server.

[0116] In the context of the embodiments of the present application, computer program codes or related data may be carried by any appropriate carrier to enable a device, apparatus or processor to perform the various processes and operations described above. Examples of carriers include signals, computer readable media, etc. Examples of signals may include electrical, optical, radio, acoustic or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0117] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0118] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the module is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, or it can be an electrical, mechanical or other form of connection.

[0119] The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0120] In addition, each functional module in each embodiment of the present application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.

[0121] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.) and signals involved in this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions. For example, the interactive intent information involved in this application is obtained with full authorization.

[0122] In this application, the terms "first", "second", etc. are used to distinguish between identical or similar items with substantially the same effects and functions. It should be understood that there is no logical or temporal dependency between "first", "second", and "nth", nor is the quantity and execution order limited. It should also be understood that although the following description uses the terms first, second, etc. to describe various elements, these elements should not be limited by the terms. These terms are only used to distinguish one element from another. For example, without departing from the scope of the various described examples, a first link may be referred to as a second link, and similarly, a second link may be referred to as a first link.

[0123] It should also be understood that in the various embodiments of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0124] The term "at least one" in this application means one or more, and the term "multiple" in this application means two or more, for example, multiple second messages means two or more second messages. The terms "system" and "network" are often used interchangeably herein.

[0125] It should be understood that the terms used in the description of the various examples herein are only for describing specific examples and are not intended to be limiting. As used in the description of the various examples and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

[0126] It should also be understood that the term “comprise” (also known as “includes,” “including,” “comprises” and / or “comprising”) when used in this specification specifies the presence of stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0127] It should also be understood that, depending on the context, the phrase “if it is determined that…” or “if [stated condition or event] is detected” may be interpreted to mean “upon determining that…” or “in response to determining that…” or “upon detecting [stated condition or event]” or “in response to detecting [stated condition or event]”.

[0128] It should be understood that determining B based on A does not mean determining B only based on A. B can also be determined based on A and / or other information.

[0129] It should also be understood that the references to "one embodiment", "an embodiment", or "a possible implementation" throughout the specification mean that specific features, structures, or characteristics related to the embodiment or implementation are included in at least one embodiment of the present application. Therefore, the references to "in one embodiment" or "in an embodiment", or "a possible implementation" throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for generating a virtual image, characterized in that: The method for generating a target virtual image using a first virtual image includes: receiving a request for generating the target virtual image input by a first user, wherein the request for generating the target virtual image includes information of the first virtual image and a skill name, wherein the first virtual image is generated by a first model file, the first virtual image has a first response capability to at least one instruction, and the skill name is used to indicate a skill possessed by the target virtual image; Generate the target virtual image according to the target model file, wherein the target model file is generated by the first model file corresponding to the information of the first virtual image and the skill model file corresponding to the skill name, and the target virtual image has a second response capability for at least one instruction; The target avatar is provided to the first user.

2. The method according to claim 1, characterized in that The target virtual image generation request information further includes skill level information, and the skill level information is used to indicate the target virtual image's learning level of the skill indicated by the skill name. Before generating the target virtual image according to the target model file, the process further includes: The number of iterations of generating the target model file is controlled according to the skill level information, wherein the degree of learning of the skill indicated by the skill level information is positively correlated with the number of iterations of generating the target model file.

3. The method according to claim 1 or 2, characterized in that: Before generating the target virtual image according to the target model file, the method further includes: receiving skill data input by a second user, wherein the skill data is used to indicate skill information corresponding to the skill name; The skill model file is generated according to the skill data.

4. The method according to any one of claims 1 to 3, characterized in that: The method further comprises: receiving user data generated when the first user uses the target virtual image; Updating the target model file according to the user data; generating an updated target virtual image according to the updated target model file; The updated target virtual image is provided to the first user.

5. The method according to any one of claims 1 to 4, characterized in that: Before receiving the target virtual image generation request information input by the first user, the method further includes: receiving generation request information of the first virtual image input by the first user, the generation request information of the first virtual image being used to indicate features of the first virtual image, the generation request information of the first virtual image comprising at least one feature type and a feature description corresponding to the at least one feature type, the feature type comprising one or more of personality, language style, target group, field, and brand; The first model file is generated based on the artificial intelligence AI model file, the at least one feature type and the feature description corresponding to the at least one feature type.

6. A device for generating a virtual image, characterized in that: The device is used to generate a target virtual image using a first virtual image, and comprises: a receiving module, configured to receive a request for generating the target virtual image input by a first user, wherein the request for generating the target virtual image includes information and a skill name of the first virtual image, wherein the first virtual image is generated by a first model file, the first virtual image has a first response capability to at least one instruction, and the skill name is used to indicate a skill possessed by the target virtual image; a generating module, configured to generate the target virtual image according to a target model file, wherein the target model file is generated by a first model file corresponding to the information of the first virtual image and a skill model file corresponding to the skill name, and the target virtual image has a second response capability to at least one instruction; A providing module is used to provide the target virtual image to the first user.

7. The device according to claim 6, characterized in that The generation request information of the target virtual image also includes skill level information, and the skill level information is used to indicate the degree of learning of the skill indicated by the skill name by the target virtual image. The generation module is also used to control the number of iterations of generating the target model file according to the skill level information, wherein the degree of learning of the skill indicated by the skill level information is positively correlated with the number of iterations of generating the target model file.

8. The device according to claim 6 or 7, characterized in that The generation module is further used to receive skill data input by a second user, where the skill data is used to indicate skill information corresponding to the skill name; and to generate the skill model file according to the skill data.

9. The device according to any one of claims 6 to 8, characterized in that: The receiving module is further used to receive user data generated when the first user uses the target virtual image; the generating module is further used to update the target model file according to the user data; generate an updated target virtual image according to the updated target model file; the providing module is further used to provide the updated target virtual image to the first user.

10. The device according to any one of claims 6 to 9, characterized in that: The receiving module is further used to receive generation request information of the first virtual image input by the first user, the generation request information of the first virtual image is used to indicate the characteristics of the first virtual image, the generation request information of the first virtual image includes at least one feature type and a feature description corresponding to the at least one feature type, the feature type includes one or more of personality, language style, target group, field, and brand; the generating module is further used to generate the first model file based on the artificial intelligence AI model file, the at least one feature type and the feature description corresponding to the at least one feature type.

11. A computing device cluster, characterized in that: The computing device cluster includes at least one computing device, each computing device includes a processor and a memory; the processor of the at least one computing device is used to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster executes the method for generating a virtual image described in any one of claims 1-5.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes computer program instructions. When the computer program instructions are executed by a computing device cluster, the computing device cluster executes the method for generating a virtual image as described in any one of claims 1-5.

Images