Image generation method and device, electronic equipment, storage medium and program product

Abstract

The invention provides an image generation method and device, electronic equipment and a storage medium, relates to the technical field of artificial intelligence, in particular to the technical fields of computer vision, deep learning, large models and the like, and can be applied to scenes such as AIGC and the like. The image generation method comprises the steps that M target processing paths are determined in a plurality of candidate processing paths included in an image generation model according to style configuration information, the candidate processing paths are in one-to-one correspondence with candidate styles, the style configuration information is used for indicating at least one target style, and the image generation model further comprises a pre-training module; the M target processing paths are utilized to process the image description to obtain adjustment information, and the adjustment information is used for guiding the image generation model to perform fine adjustment on the image generation process based on the at least one target style; and based on the adjustment information, adjusting the output obtained by processing the image description by using the pre-training module to obtain a target image with at least one target style.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, especially to technical fields such as computer vision, deep learning, and large models, and can be applied to scenarios such as AIGC content generation based on artificial intelligence. More specifically, it relates to an image generation method and device, electronic device, storage medium, and program product. Background Art

[0002] Generative AI (Artificial Intelligence Generated Content, AIGC) refers to the technology based on artificial intelligence technical methods such as generative adversarial networks and large pre-trained models, which generates relevant content with appropriate generalization capabilities through learning and recognition of existing data.

[0003] In one example, generative AI technology can be used to generate emoticons. Emoticons are a way of expressing specific emotions using popular celebrities, quotes, animations, and movie screenshots, along with a series of matching texts. Summary of the invention

[0004] In view of this, the present disclosure provides an image generating method and device, an electronic device, a storage medium and a program product.

[0005] According to one aspect of the present disclosure, there is provided an image generation method, comprising: determining M target processing paths among a plurality of candidate processing paths included in an image generation model according to style configuration information, wherein the candidate processing paths correspond one-to-one to the candidate styles, the style configuration information is used to indicate at least one target style, the image generation model further comprises a pre-training module, and M is a positive integer; processing an image description using the M target processing paths to obtain adjustment information, wherein the adjustment information is used to guide the image generation model to fine-tune the image generation process based on the at least one target style; and, based on the adjustment information, adjusting the output obtained by processing the image description using the pre-training module to obtain a target image having the at least one target style.

[0006] According to another aspect of the present disclosure, an image generation device is provided, including: a determination module, for determining M target processing paths from a plurality of candidate processing paths included in an image generation model according to style configuration information, wherein the candidate processing paths correspond one-to-one to the candidate styles, the style configuration information is used to indicate at least one target style, and the image generation model further includes a pre-training module, where M is a positive integer; a processing module, for processing an image description using the M target processing paths to obtain adjustment information, wherein the adjustment information is used to guide the image generation model to fine-tune the image generation process based on the at least one target style; and a generation module, for adjusting the output obtained by processing the image description using the pre-training module based on the adjustment information to obtain a target image having the at least one target style.

[0007] According to another aspect of the present disclosure, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0008] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, the steps of the above method are implemented.

[0009] According to another aspect of the present disclosure, a computer program product is provided, including a computer program or instructions, and the steps of the above method are implemented when the computer program or instructions are executed by a processor.

[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0012] Figure 1 The system architecture to which the image generation method according to the embodiment of the present disclosure can be applied is schematically shown;

[0013] Figure 2 The flowchart of the image generation method according to the embodiment of the present disclosure is schematically shown;

[0014] Figure 3 An example schematic diagram schematically illustrates a training process of an image generation model according to an embodiment of the present disclosure;

[0015] Figure 4 An example schematic diagram schematically illustrates a process of determining a target processing path according to an embodiment of the present disclosure;

[0016] Figure 5 An example schematic diagram schematically shows a target image generation process with a target style according to an embodiment of the present disclosure;

[0017] Figure 6 An example schematic diagram schematically shows a target image generation process with multiple target styles according to an embodiment of the present disclosure;

[0018] Figure 7 A block diagram schematically shows an image generating device according to an embodiment of the present disclosure; and

[0019] Figure 8 A block diagram of an electronic device suitable for implementing an image generating method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0020] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0021] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0022] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0023] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0024] In the related art, the generation methods of emoticon packages include a method based on manual creation and a method based on template editing.

[0025] The manual creation method refers to the method in which users use image processing software to manually design and modify expression images to generate expression packages. However, since this process requires users to adjust each detail frame by frame (for example, image composition, color matching, expression design, etc.), and each time a new expression package needs to be generated, it needs to start from scratch, resulting in a long production cycle and low generation efficiency for expression packages, which cannot meet the rapid needs of users in instant communication. In addition, the high threshold for using image processing software also limits the popularity of expression package creation.

[0026] Template-based editing refers to the way users generate emoticons by making partial adjustments based on pre-set emoticon templates. However, since this process relies on pre-designed emoticon templates, the user's creative freedom is limited, making it difficult to express their personalized needs based on specific conversation situations or emotions. In addition, conventional text-based graph models can only generate emoticon images with a single style, making it impossible for users to generate multi-style emoticons that fully match their own semantics or emotions.

[0027] To this end, the disclosed embodiment proposes an image generation scheme. For example, according to the style configuration information, among the multiple candidate processing paths included in the image generation model, M target processing paths are determined, the candidate processing paths correspond to the candidate styles one by one, the style configuration information is used to indicate at least one target style, and the image generation model also includes a pre-training module; using the M target processing paths, the image description is processed to obtain adjustment information, and the adjustment information is used to guide the image generation model to fine-tune the image generation process based on at least one target style; based on the adjustment information, the output obtained by processing the image description using the pre-training module is adjusted to obtain a target image with at least one target style. .

[0028] According to the embodiments of the present disclosure, by introducing style configuration information, M target processing paths that match at least one target style are accurately selected from multiple candidate processing paths included in the image generation model, so that the style diversity of the generated image can be guaranteed. By processing the image description using the M target processing paths, the obtained adjustment information can provide a basis for fine-tuning the image generation process. On this basis, by adjusting the output obtained by processing the image description using the pre-training module based on the adjustment information, the generated target image can better reflect at least one target style, thereby improving the flexibility of style selection and thereby improving the popularity and diversity of image generation.

[0029] In the technical solution of the present invention, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals. In the technical solution of the present invention, the user's authorization or consent is obtained before obtaining or collecting user personal information.

[0030] Figure 1 The system architecture to which the image generation method according to the embodiment of the present disclosure can be applied is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0031] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104 and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0032] The user may use at least one of the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0033] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0034] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0035] It should be noted that the image generation method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the image generation device provided in the embodiment of the present disclosure can generally be set in the server 105. The image generation method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the image generation device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0036] Alternatively, the image generation method provided in the embodiment of the present disclosure may also be executed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may also be executed by other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103. Accordingly, the image generation device provided in the embodiment of the present disclosure may also be provided in the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may be provided in other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0037] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0038] It should be noted that the sequence numbers of the operations in the following method are only used as representations of the operations for the purpose of description, and should not be regarded as representing the execution order of the operations. Unless explicitly stated, the method does not need to be executed completely in the order shown.

[0039] The above describes the system architecture to which the image generation method provided by the present disclosure can be applied. Figure 2 For example, the image generation process of the present disclosure is further described.

[0040] Figure 2 The flowchart of the image generating method according to the embodiment of the present disclosure is schematically shown.

[0041] like Figure 2 As shown, the image generating method includes operations S210 to S230.

[0042] In operation S210, based on the style configuration information, M target processing paths are determined from a plurality of candidate processing paths included in the image generation model, wherein the candidate processing paths correspond one-to-one to the candidate styles, the style configuration information is used to indicate at least one target style, the image generation model also includes a pre-training module, and M is a positive integer.

[0043] In operation S220, the image description is processed using the M target processing paths to obtain adjustment information, wherein the adjustment information is used to guide the image generation model to fine-tune the image generation process based on at least one target style.

[0044] In operation S230, based on the adjustment information, an output obtained by processing the image description using the pre-training module is adjusted to obtain a target image having at least one target style.

[0045] Before executing the image generation method provided by the present disclosure, an image generation model can be pre-trained. The image generation model may include a pre-training module and multiple candidate processing paths, each candidate processing path having a candidate style, and the candidate style is used to indicate the style of the generated image. The pre-training module can be used to perform a text image task, and the candidate processing path can be used to perform a style adjustment task. Through the cooperation of the above-mentioned pre-training module and multiple candidate processing paths, target images with different styles can be generated. The specific structure of the pre-training module and the candidate processing path can be configured according to actual business needs and is not limited here.

[0046] In one example, the pre-training module refers to a large-scale language model that can convert information between different sensory modalities (such as vision, language, audio, etc.). For example, the pre-training module may include at least one of the following: a diffusion model, a DALL-E model, an Imagen model, etc.

[0047] In another example, the candidate processing path refers to a processing branch constructed based on the large model fine-tuning technology, which can be used to guide the pre-trained module to perform fine-tuning in the image generation process based on the candidate style. For example, the construction method of the candidate processing path may include at least one of the following: LoRA (Low-Rank Adaptation), DoRA (Understanding Weight-Decomposed Low-Rank Adaptation) and VeRA (Vector-based Random Matrix Adaptation), etc.

[0048] The style configuration information can be used to indicate at least one target style among multiple candidate styles. The specific form of the style configuration information can be configured according to actual business needs and is not limited here. For example, the style configuration information may include at least one target style. Alternatively, the style configuration information may include respective configuration parameters for each candidate style, and the configuration parameters may belong to [0,1]. The configuration parameters are used to indicate whether the target image to be generated has a certain candidate style and the degree of contribution of the candidate style to the generated target image. After obtaining the style configuration information, the target processing path corresponding to each target style can be determined among multiple candidate processing paths according to at least one target style indicated by the style configuration information.

[0049] After obtaining M target processing paths, the image description can be input into each target processing path respectively, and the output of each target processing path is used as adjustment information for guiding the image generation model to fine-tune the image generation process based on at least one target style. Image description refers to natural language text used to describe the attributes of the target image to be generated. For example, the target style includes style A and style B, and the image description can be input into target processing path 1 corresponding to style A and target processing path 2 corresponding to style B respectively, and the outputs of target processing path 1 and target processing path 2 are used together as adjustment information.

[0050] After obtaining M target processing paths, the image description can also be input into the pre-training module to obtain the output of the pre-training module. On this basis, the output of the pre-training module can be adjusted using the adjustment information to obtain a target image having at least one target style indicated by the style configuration information. Continuing with the above example, the adjustment information can guide the pre-training module to fine-tune the image generation process based on style A and style B, thereby obtaining a target image having style A and style B.

[0051] In one example, the application scenarios of the image generation method provided by the present disclosure include at least one of the following: instant messaging software, social media platforms, digital marketing and brand promotion, and content creation platforms, etc. For example, in the scenario of instant messaging software, a target image that matches the content can be instantly generated according to the image description input by the user during the chat process to enhance the fun of the interaction. Alternatively, in the scenario of a social media platform, a target image that matches the emotional expression can be generated through a personalized image description input by the user. Alternatively, in the scenario of digital marketing and brand promotion, an enterprise can automatically generate customized target images based on hot topics or brand tonality to enhance the fun of brand communication. Alternatively, in the scenario of a content creation platform, a convenient creation tool can be provided for creators to support the generation of target images based on image descriptions to reduce creation time.

[0052] According to an embodiment of the present disclosure, by introducing style configuration information, M target processing paths that match at least one target style are accurately selected from multiple candidate processing paths included in the image generation model, so that the style diversity of the generated image can be guaranteed. By processing the image description using the M target processing paths, the obtained adjustment information can provide a basis for fine-tuning the image generation process. On this basis, by adjusting the output obtained by processing the image description using the pre-training module based on the adjustment information, the generated target image can better reflect at least one target style, conform to the user's emotional and semantic personalization, improve the flexibility of style selection, and thus improve the efficiency, universality and diversity of image generation.

[0053] In the embodiment of the present disclosure, before executing the image generation method, the candidate processing paths for different candidate styles may be trained separately to obtain an image generation model. Figure 3 Take as an example to illustrate the training process of the image generation model.

[0054] Figure 3 An example schematic diagram of the training process of the image generation model according to an embodiment of the present disclosure is schematically shown.

[0055] like Figure 3 As shown, in 300, training samples of each of the plurality of candidate styles may be generated. In one example, the training samples may include a plurality of image-text pairs, the image-text pairs including sample images and sample image descriptions for the candidate styles. The candidate styles may be configured according to actual business needs, which are not limited here. For example, the candidate styles may include at least one of the following: violent comic style, realistic animal style, stick figure style, two-dimensional style, big-headed doll style, and graffiti style, etc.

[0056] The method of obtaining the image-text pair of each candidate style can be configured according to actual business needs and is not limited here. For example, for each candidate style, P sample images belonging to the candidate style can be obtained; considering that the texture of numbers can easily lead to poor image generation, the images including numbers in the P sample images can be filtered out to obtain Q sample images, P ≥ Q; on this basis, the Q sample images are respectively input into the large model to obtain the sample image description of each sample image, which can include emotions, emotional tendencies, and scene descriptions; on this basis, the image-text pair for the candidate style can be generated based on the Q sample images and the sample image description of each sample image.

[0057] After obtaining the training samples, the pre-training module 320_0 and the N to-be-trained processing paths for different candidate styles can be trained respectively using the image-text pairs for different candidate styles. For example, the training samples may include an image-text pair 310_1 with candidate style 1, an image-text pair 310_2 with candidate style 2, ..., an image-text pair 310_N with candidate style N, where N is a positive integer and N is greater than or equal to M.

[0058] In one example, the training process of the candidate processing paths for candidate style 1 and the training process of the candidate processing paths for candidate style 2 are taken as examples to illustrate the training process of the candidate processing paths for different candidate styles.

[0059] For candidate style 1, the image-text pair 310_1 with candidate style 1 can be respectively input into the pre-training module 320_0 and the processing path 320_1 to be trained for candidate style 1 to obtain a generated image 330_1. The parameters of the processing path 320_1 to be trained are adjusted using the generated image 330_1 to obtain a candidate processing path 320_1 for candidate style 1, and the weight matrix obtained by low-rank adaptation of the candidate processing path 320_1 can be used as the weight parameter of the candidate processing path 320_1. For example, the weight parameter can be the product of two low-rank matrices.

[0060] For candidate style 2, the image-text pair 310_2 with candidate style 2 can be input into the pre-training module 320_0 and the to-be-trained processing path 320_2 for candidate style 1, respectively, to obtain a generated image 330_2. The parameters of the to-be-trained processing path 320_2 are adjusted using the generated image 330_2 to obtain a candidate processing path 320_2 for candidate style 2, and a weight matrix obtained by low-rank adaptation of the candidate processing path 320_2 can be used as a weight parameter of the candidate processing path 320_2, for example, the weight parameter can be the product of two low-rank matrices.

[0061] By analogy, the image pair 310_N with the candidate style N can be used to train a candidate processing path 320_N for the candidate style N, and the weight matrix obtained by low-rank adaptation of the candidate processing path 320_N can be used as the weight parameter of the candidate processing path 320_N. For example, the weight parameter can be the product of two low-rank matrices.

[0062] After obtaining N candidate processing paths for different candidate styles, the pre-training module 320_0 and the trained N candidate processing paths for different candidate styles (ie, candidate processing path 320_1 , candidate processing path 320_2 , . . . , candidate processing path 320_N) may be used as the image generation model 320 .

[0063] According to the embodiments of the present disclosure, since the candidate processing paths for the candidate styles are obtained by training the processing paths to be trained using the training samples corresponding to each candidate style, by adopting this method of training each candidate processing path separately, each candidate processing path can specifically capture the visual features of a specific candidate style, thereby improving the diversity and flexibility of the generated images. In addition, compared with deploying multiple models separately, more computing resources can be saved, making the process more convenient, thereby enabling the image generation model to be more accurate and diversified when generating target images of different target styles, thereby improving the quality of image generation.

[0064] The training method of the image generation model provided by the present disclosure is described above. In the embodiment of the present disclosure, after receiving the style configuration information, M target processing paths can be determined from multiple candidate processing paths included in the image generation model according to the style configuration information. Figure 4 Take as an example to illustrate the process of determining the target processing path.

[0065] Figure 4 An example schematic diagram schematically illustrates a process of determining a target processing path according to an embodiment of the present disclosure.

[0066] like Figure 4 As shown, in 400, style configuration information 410 can be used to indicate at least one target style among multiple candidate styles. The determination method of style configuration information 410 can be configured according to actual business needs and is not limited here. For example, the acquisition method of style configuration information 410 can include an active method and a passive method. The active method can refer to a method in which the style configuration information 410 is actively determined by the server after obtaining the image description. The passive method can refer to a method in which the style configuration information 410 needs to be determined based on a user request.

[0067] In one example, the active method may refer to a method of performing semantic analysis on the image description to obtain the style configuration information 410. For example, the image description may be input into a semantic analysis network to obtain the style configuration information 410. The specific structure of the semantic analysis network may be configured according to actual business needs and is not limited here. For example, the semantic analysis network may include at least one of the following: a recurrent neural network (RNN) and a long short-term memory network (LSTM). In another example, the passive method may refer to a method of determining the style configuration information 410 in response to an image generation request according to user indication information in the image generation request.

[0068] According to the embodiments of the present disclosure, by responding to image generation requests and analyzing the user-provided information therein, the user's explicit style preferences can be directly captured to ensure that the generated target image meets the user's expectations. By performing in-depth semantic analysis and processing on the image description, implicit style features can be automatically identified and extracted, so that the image generation model can intelligently identify and adapt to different style requirements. By combining the above two methods of obtaining style configuration information, the adaptability and user satisfaction of the image generation model are improved, while the complexity of user operations is reduced, making the image generation process more intuitive and convenient.

[0069] After obtaining the style configuration information 410, M target processing paths may be determined from the multiple candidate processing paths included in the image generation model according to the style configuration information 410. The style configuration information 410 may include configuration parameters for each candidate style, and the style configuration information 410 may be used to dynamically select and apply different candidate processing paths by adjusting the configuration parameters. The configuration parameters may belong to [0, 1].

[0070] In one example, when the configuration parameter is 0, it indicates that the candidate style corresponding to the configuration parameter is not the target style. When the configuration parameter is 1, it indicates that the candidate style corresponding to the configuration parameter is the target style. When the configuration parameter is between (0, 1), it indicates that the candidate style corresponding to the configuration parameter is the target style, and the contribution of the candidate style to the generated target image is equal to the configuration parameter.

[0071] For each configuration parameter, it can be determined whether the configuration parameter is greater than a preset threshold. The preset threshold can be 0. In the case where the configuration parameter is greater than the preset threshold, the candidate style corresponding to the configuration parameter can be determined as the target style. On this basis, according to the correspondence between the candidate processing path and the candidate style, the candidate processing path corresponding to the target style can be determined as the target processing path.

[0072] For example, the style configuration information 410 may include a configuration parameter 410_1 corresponding to the candidate style 1, a configuration parameter 410_2 corresponding to the candidate style 2, ..., a configuration parameter 410_N corresponding to the candidate style N. The image generation model 420 may include a pre-training module 420_0, a candidate processing path 420_1 for the candidate style 1, a candidate processing path 420_2 for the candidate style 2, ..., a candidate processing path 420_N for the candidate style N.

[0073] Taking the style configuration information as (1, 0, 0, ..., 0, 1) as an example, since the configuration parameter 410_1 is 1, the candidate style 1 corresponding to the configuration parameter 410_1 can be determined as the target style, and the candidate processing path 420_1 corresponding to the candidate style 1 can be determined as the target processing path. Similarly, since the configuration parameter 410_N is 1, the candidate style N corresponding to the configuration parameter 410_N can be determined as the target style, and the candidate processing path 420_N corresponding to the candidate style N can be determined as the target processing path.

[0074] According to the embodiments of the present disclosure, the target style is intelligently determined based on the comparison result between the configuration parameters and the preset threshold, thereby providing an accurate style guide for the image generation process. By determining the candidate style with a configuration parameter greater than the preset threshold as the target style, and determining the candidate processing path corresponding to the target style as the target processing path, this method of automatically determining the target processing path based on the configuration parameters and the corresponding relationship not only improves the accuracy and efficiency of image generation, but also provides users with a more personalized and diversified image generation experience.

[0075] In one example, the process of obtaining a target image having at least one target style is as shown in the following formula (1).

[0076] (1)

[0077] in, Characterize the target image, , , …, , …, Characterize N candidate processing paths, , , …, , …, Characterize the configuration parameters of each candidate processing path, Represents intermediate information.

[0078] The above describes how the present disclosure determines the target processing path. Figure 5 As an example, the process of generating the target image is further explained when there is only one target processing path.

[0079] Figure 5 An example schematic diagram of a target image generation process with a target style according to an embodiment of the present disclosure is schematically shown.

[0080] like Figure 5As shown, in 500, an example of a process of generating a target image with a target style is schematically shown in the case where M = 1. In this case, style configuration information 501 can be used to switch the target style using configuration parameters.

[0081] After obtaining the style configuration information 501, a target processing path may be determined from among a plurality of candidate processing paths included in the image generation model according to the style configuration information 501. For example, the target processing path 520_2.

[0082] In one example, the image description 502 can be input into the target processing path 520_2, so that the target processing path 520_2 processes the image description 502 according to its own weight parameters to obtain adjustment information 503. It should be noted that the weight parameters of each candidate processing path are obtained in the process of training the candidate processing path using image-text pairs with candidate styles, that is, the low-rank adaptive weight matrix of the candidate processing path. For example, the weight parameter can be the product of two low-rank matrices.

[0083] According to the embodiments of the present disclosure, since the weight parameters of each candidate processing path are obtained by training with image-text pairs with corresponding candidate styles, the image description is finely processed by utilizing the weight parameters of the target processing path, so that adjustment information with the target style can be obtained more accurately during the image generation process, thereby improving the quality of image generation and style consistency.

[0084] In addition, the image description 502 can also be input into the pre-training module 520_0, so that the pre-training module 520_0 processes the image description 502 to obtain intermediate information 504. On this basis, the intermediate information 504 can be adjusted based on the adjustment information 503 to obtain the target image 505. The specific adjustment method can be configured according to actual business needs and is not limited here. In one example, the adjustment information 503 and the intermediate information 504 can be processed based on the attention strategy to obtain the target image 505.

[0085] According to an embodiment of the present disclosure, the image description is processed by utilizing a pre-trained module, and based on an attention strategy, the intermediate information obtained is processed and the adjustment information obtained via a target processing path, so as to generate a high-quality target image with a target style, thereby improving the quality and practicality of image generation.

[0086] For example, the attention strategy can include a cross-attention strategy. The cross-attention strategy can be used for information interaction between different inputs, enabling the model to effectively align and focus on contexts from different sources, thereby helping the model better capture the correlation between the two inputs.

[0087] For example, the query sequence and key value sequence of the output of the target processing path 520_2 can be determined, and the adjustment information obtained includes a first key matrix (K1), a first value matrix (V1), and a first query matrix (Q1). The query sequence and key value sequence of the output of the pre-training module 520_0 are determined, and the intermediate information obtained includes a second key matrix (K2), a second value matrix (V2), and a second query matrix (Q2).

[0088] After obtaining the adjustment information 503 and the intermediate information 504, the first weight matrix can be determined according to the second query matrix (Q2) and the second key matrix (K2). The first feature map can be determined according to the first weight matrix and the first value matrix (V1). In addition, the second weight matrix can be determined according to the first query matrix (Q1) and the first key matrix (K1). The second feature map can be determined according to the second weight matrix and the second value matrix (V2). On this basis, the first feature map and the second feature map can be feature fused to obtain the target image 505.

[0089] In an example, when M=1, the process of obtaining a target image with a target style is as shown in the following formula (2).

[0090] (2)

[0091] in, Characterizing the target image 505, Characterizing the candidate processing path 520_2, Configuration parameters characterizing the candidate processing path 520_2, Characterization adjustment information 503, Characterize the intermediate information 504 .

[0092] According to an embodiment of the present disclosure, since the first feature map is determined based on the first value matrix, the first weight matrix obtained based on the second query matrix and the second key matrix, and the second feature map is determined based on the second value matrix, the second weight matrix obtained based on the first query matrix and the first key matrix, by fusing the first feature map and the second feature map, the obtained target image can have at least one target style, thereby improving the accuracy of target image generation.

[0093] The above describes the generation of the target image when the target processing path is one in the present disclosure. Figure 6 As an example, the process of generating the target image when there are multiple target processing paths is further explained.

[0094] Figure 6 An example schematic diagram of a target image generation process with multiple target styles according to an embodiment of the present disclosure is schematically shown.

[0095] like Figure 6 As shown, in 600, an example of a process of generating a target image with multiple target styles is schematically shown in the case where M>1. In this case, the style configuration information can be used to fuse multiple target styles.

[0096] After the style configuration information is obtained, multiple target processing paths may be determined from multiple candidate processing paths included in the image generation model according to the style configuration information, for example, target processing path 620_1 and target processing path 620_2.

[0097] In one example, the image description 602 can be input into the target processing path 620_1 and the target processing path 620_2 respectively, so that the target processing path 620_1 processes the image description 602 according to its own weight parameters to obtain the sub-adjustment information 603_1, and the target processing path 620_2 processes the image description 602 according to its own weight parameters to obtain the sub-adjustment information 603_2. It should be noted that the weight parameters of each candidate processing path are obtained in the process of training the candidate processing path using the image-text pairs with the candidate style, that is, the low-rank adaptive weight matrix of the candidate processing path. For example, the weight parameters can be the product of two low-rank matrices.

[0098] After obtaining the respective sub-adjustment information of each target processing path, the respective sub-adjustment information of each target processing path may be weighted based on the respective configuration parameters of each target style in the style configuration information to obtain adjustment information 604. For example, sub-adjustment information 603_1 and sub-adjustment information 603_2 may be weighted based on configuration parameters 601_1 for target processing path 620_1 and configuration parameters 601_2 for target processing path 620_2 to obtain adjustment information 604.

[0099] According to an embodiment of the present disclosure, by inputting the image description into each target processing path respectively, sub-adjustment information of each target processing path is obtained, and then the sub-adjustment information is weighted by combining the configuration parameters of each target style in the style configuration information to obtain adjustment information for guiding the image generation model to fine-tune the image generation process based on multiple target styles, which is beneficial to improving the efficiency and quality of subsequent image generation.

[0100] In addition, the image description 602 can also be input into the pre-training module 620_0, so that the pre-training module 620_0 processes the image description 602 to obtain the intermediate information 605. On this basis, the intermediate information 605 can be adjusted based on the adjustment information 604 to obtain the target image 606. The specific adjustment method can be configured according to actual business needs and is not limited here. In one example, the adjustment information 604 and the intermediate information 605 can be processed based on the attention strategy to obtain the target image 606. The specific method of processing the adjustment information 604 and the intermediate information 605 based on the attention strategy to obtain the target image 606 is as above and will not be repeated here.

[0101] In one example, when M>1, the process of obtaining a target image with multiple target styles is as shown in the following formula (3).

[0102] (3)

[0103] in, characterizing the target image 606, Characterize the target processing path 620_1, Characterize the configuration parameters of the target processing path 620_1, Characterization sub-adjustment information 603_1, Characterize the target processing path 620_2, Characterize the configuration parameters of the target processing path 620_2, Characterization sub-adjustment information 603_2, Characterize the intermediate information 605.

[0104] According to an embodiment of the present disclosure, by utilizing a pre-trained module to process the image description, and based on an attention strategy, processing the obtained intermediate information and the adjustment information obtained via multiple target processing paths, high-quality target images with multiple target styles are generated, thereby improving the quality and practicality of image generation.

[0105] It should be noted that in the embodiments of the present disclosure, certain software, components, models and other existing solutions in the industry may be mentioned, which should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present disclosure, but it does not mean that the present disclosure has or will necessarily use the solution.

[0106] The above are only exemplary embodiments, but are not limited thereto, and may also include other image generation methods known in the art, as long as they can improve the flexibility of style selection, as well as the efficiency, popularity and diversity of image generation.

[0107] Based on the above image generation method, the present invention also provides an image generation device. Figure 7 The device is described in detail.

[0108] Figure 7 A block diagram of an image generating device according to an embodiment of the present disclosure is schematically shown.

[0109] like Figure 7 As shown, the image generating device 700 may include a determining module 710 , a processing module 720 and a generating module 730 .

[0110] The determination module 710 is used to determine M target processing paths from multiple candidate processing paths included in the image generation model according to the style configuration information, wherein the candidate processing paths correspond one-to-one to the candidate styles, the style configuration information is used to indicate at least one target style, the image generation model also includes a pre-training module, and M is a positive integer.

[0111] The processing module 720 is used to process the image description using the M target processing paths to obtain adjustment information, wherein the adjustment information is used to guide the image generation model to fine-tune the image generation process based on at least one target style.

[0112] The generation module 730 is used to adjust the output obtained by processing the image description using the pre-training module based on the adjustment information to obtain a target image with at least one target style.

[0113] According to an embodiment of the present disclosure, when M=1, the style configuration information is used to switch the target style. The processing module 720 may include a first processing submodule.

[0114] The first processing submodule is used to process the image description according to the weight parameters of the target processing path to obtain adjustment information, wherein the weight parameters are obtained by training the image-text pairs with the candidate styles.

[0115] According to an embodiment of the present disclosure, in the case where M>1, the style configuration information is used to fuse the target style. The processing module 720 may include an input submodule and a second processing submodule.

[0116] The input submodule is used for inputting the image description into the target processing path for each target processing path to obtain sub-adjustment information.

[0117] The second processing submodule is used to perform weighted processing on the sub-adjustment information of each of the M target processing paths based on the respective configuration parameters of each target style in the style configuration information to obtain adjustment information.

[0118] According to an embodiment of the present disclosure, the generation module 730 may include a third processing submodule and a fourth processing submodule.

[0119] The third processing submodule is used to process the image description using the pre-training module to obtain intermediate information.

[0120] The fourth processing submodule is used to process the adjustment information and the intermediate information based on the attention strategy to obtain the target image.

[0121] According to an embodiment of the present disclosure, the attention strategy includes a cross-attention strategy, the adjustment information includes a first key matrix, a first value matrix and a first query matrix, and the intermediate information includes a second key matrix, a second value matrix and a second query matrix.

[0122] According to an embodiment of the present disclosure, the fourth processing submodule may include a first determining unit, a second determining unit, and a third determining unit.

[0123] The first determining unit is used to determine a first feature map according to the first value matrix and a first weight matrix determined based on the second query matrix and the second key matrix.

[0124] The second determining unit is used to determine a second feature map according to a second value matrix and a second weight matrix determined based on the first query matrix and the first key matrix.

[0125] The third determining unit is used to determine the target image according to the first feature map and the second feature map.

[0126] According to an embodiment of the present disclosure, the style configuration information includes configuration parameters of each of a plurality of candidate styles. The determination module 710 may include a first determination submodule and a second determination submodule.

[0127] The first determination submodule is configured to determine, in response to a configuration parameter being greater than a preset threshold, a candidate style corresponding to the configuration parameter as a target style.

[0128] The second determination submodule is used to determine the candidate processing path corresponding to the target style as the target processing path according to the correspondence between the candidate processing paths and the candidate styles.

[0129] According to an embodiment of the present disclosure, the style configuration information is obtained by one of the following methods: in response to an image generation request, the style configuration information is determined according to user indication information in the image generation request; or, the image description is semantically analyzed to obtain the style configuration information.

[0130] According to an embodiment of the present disclosure, an image generation model is trained in the following manner: generating training samples for each of a plurality of candidate styles, wherein the training samples include a plurality of image-text pairs, and the image-text pairs include sample images and sample image descriptions for the candidate styles; for each candidate style, using the training samples for the candidate style, training a pre-trained module and a processing path to be trained for the candidate style to obtain a candidate processing path for the candidate style; and using the pre-trained module and the candidate processing path for each candidate style as an image generation model.

[0131] Figure 8 A block diagram of an electronic device suitable for implementing an image generation method according to an embodiment of the present disclosure is schematically shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0132] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 to a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0133] A number of components in the device 800 are connected to the input / output I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0134] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as the image generation method. For example, in some embodiments, the image generation method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the image generation method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the image generation method in any other appropriate manner (e.g., by means of firmware).

[0135] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0136] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0137] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0138] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0139] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0140] A computer system may include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises through computer programs running on respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server combined with a blockchain.

[0141] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0142] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for generating an image, comprising: Determine, according to the style configuration information, M target processing paths from among a plurality of candidate processing paths included in the image generation model, wherein the candidate processing paths correspond one-to-one to the candidate styles, the style configuration information is used to indicate at least one target style, the image generation model further includes a pre-training module, and M is a positive integer; Processing the image description using the M target processing paths to obtain adjustment information, wherein the adjustment information is used to guide the image generation model to fine-tune the image generation process based on the at least one target style; and Based on the adjustment information, an output obtained by processing the image description using the pre-training module is adjusted to obtain a target image having the at least one target style.

2. The method according to claim 1, wherein: In the case of M=1, the style configuration information is used to switch the target style; The processing of the image description by using the M target processing paths to obtain adjustment information includes: The image description is processed according to the weight parameters of the target processing path to obtain the adjustment information, wherein the weight parameters are obtained by training with the image-text pairs having the candidate style.

3. The method according to claim 1, wherein: In the case of M>1, the style configuration information is used to fuse the target style; The processing of the image description by using the M target processing paths to obtain adjustment information includes: For each of the target processing paths, Inputting the image description into the target processing path to obtain sub-adjustment information; and Based on the respective configuration parameters of each of the target styles in the style configuration information, weighted processing is performed on the respective sub-adjustment information of the M target processing paths to obtain the adjustment information.

4. The method according to any one of claims 1 to 3, wherein: The adjusting, based on the adjustment information, the output obtained by processing the image description by using the pre-training module to obtain a target image having the at least one target style comprises: Using the pre-training module, processing the image description to obtain intermediate information; and Based on the attention strategy, the adjustment information and the intermediate information are processed to obtain the target image.

5. The method according to claim 4, wherein: The attention strategy includes a cross attention strategy, the adjustment information includes a first key matrix, a first value matrix and a first query matrix, and the intermediate information includes a second key matrix, a second value matrix and a second query matrix; The processing of the adjustment information and the intermediate information based on the attention strategy to obtain the target image includes: Determine a first feature map according to the first value matrix and a first weight matrix determined based on the second query matrix and the second key matrix; determining a second feature map according to the second value matrix and a second weight matrix determined based on the first query matrix and the first key matrix; and The target image is determined according to the first feature map and the second feature map.

6. The method according to any one of claims 1 to 3, wherein: The style configuration information includes configuration parameters of each of the plurality of candidate styles; Determining M target processing paths from among a plurality of candidate processing paths included in the image generation model according to the style configuration information includes: In response to the configuration parameter being greater than a preset threshold, determining the candidate style corresponding to the configuration parameter as the target style; as well as According to the correspondence between the candidate processing paths and the candidate styles, the candidate processing path corresponding to the target style is determined as the target processing path.

7. The method according to claim 6, wherein: The method for obtaining the style configuration information includes one of the following: In response to an image generation request, determining the style configuration information according to user instruction information in the image generation request; or Perform semantic analysis on the image description to obtain the style configuration information.

8. The method according to any one of claims 1 to 3, wherein: The image generation model is trained using the following method: Generate training samples for each of a plurality of candidate styles, wherein the training samples include a plurality of image-text pairs, and the image-text pairs include sample images and sample image descriptions for the candidate styles; For each of the candidate styles, using the training samples for the candidate style, training the pre-training module and the to-be-trained processing path for the candidate style to obtain a candidate processing path for the candidate style; and The pre-trained module and the candidate processing path for each of the candidate styles are used as the image generation model.

9. An image generating device, comprising: A determination module, configured to determine, according to the style configuration information, M target processing paths from a plurality of candidate processing paths included in the image generation model, wherein the candidate processing paths correspond one-to-one to the candidate styles, the style configuration information is used to indicate at least one target style, the image generation model further includes a pre-training module, and M is a positive integer; a processing module, configured to process the image description using the M target processing paths to obtain adjustment information, wherein the adjustment information is used to guide the image generation model to fine-tune the image generation process based on the at least one target style; and A generation module is used to adjust the output obtained by processing the image description using the pre-training module based on the adjustment information to obtain a target image with the at least one target style.

10. The device according to claim 9, wherein: In the case of M=1, the style configuration information is used to switch the target style; The processing module comprises: The first processing submodule is used to process the image description according to the weight parameters of the target processing path to obtain the adjustment information, wherein the weight parameters are obtained by training with the image-text pairs having the candidate style.

11. The device according to claim 9, wherein: In the case of M>1, the style configuration information is used to fuse the target style; The processing module comprises: an input submodule, for inputting the image description into each target processing path to obtain sub-adjustment information; and The second processing submodule is used to perform weighted processing on the sub-adjustment information of each of the M target processing paths based on the respective configuration parameters of each of the target styles in the style configuration information to obtain the adjustment information.

12. The device according to any one of claims 9 to 11, wherein: The generation module comprises: A third processing submodule is used to process the image description using the pre-training module to obtain intermediate information; and The fourth processing submodule is used to process the adjustment information and the intermediate information based on the attention strategy to obtain the target image.

13. The device according to claim 12, wherein: The attention strategy includes a cross attention strategy, the adjustment information includes a first key matrix, a first value matrix and a first query matrix, and the intermediate information includes a second key matrix, a second value matrix and a second query matrix; The fourth processing submodule includes: A first determining unit, configured to determine a first feature map according to the first value matrix and a first weight matrix determined based on the second query matrix and the second key matrix; a second determining unit, configured to determine a second feature map according to the second value matrix and a second weight matrix determined based on the first query matrix and the first key matrix; and A third determining unit is used to determine the target image according to the first feature map and the second feature map.

14. The device according to any one of claims 9 to 11, wherein: The style configuration information includes configuration parameters of each of the plurality of candidate styles; The determination module comprises: A first determination submodule, configured to determine, in response to the configuration parameter being greater than a preset threshold, a candidate style corresponding to the configuration parameter as a target style; and The second determination submodule is used to determine the candidate processing path corresponding to the target style as the target processing path according to the correspondence between the candidate processing paths and the candidate styles.

15. The device according to claim 14, wherein: The method for obtaining the style configuration information includes one of the following: In response to an image generation request, determining the style configuration information according to user instruction information in the image generation request; or Perform semantic analysis on the image description to obtain the style configuration information.

16. The device according to any one of claims 9 to 11, wherein: The image generation model is trained using the following method: Generate training samples for each of a plurality of candidate styles, wherein the training samples include a plurality of image-text pairs, and the image-text pairs include sample images and sample image descriptions for the candidate styles; For each of the candidate styles, using the training samples for the candidate style, training the pre-training module and the to-be-trained processing path for the candidate style to obtain a candidate processing path for the candidate style; and The pre-trained module and the candidate processing path for each of the candidate styles are used as the image generation model.

17. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 8.

18. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

19. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images