Topic generation method, apparatus, electronic device and readable storage medium

Abstract

This application provides a theme generation method, apparatus, electronic device, and readable storage medium, applied to a server. The method includes: receiving a theme generation instruction, an environmental image of the vehicle's environment, and the vehicle's driving data; performing multimodal fusion processing on the theme generation instruction, environmental image, and driving data to determine theme parameters, including a reference hue range; extracting theme keywords from the theme generation instruction and generating prompt words based on the theme keywords, including color prompt words and content prompt words; and using a theme generation model to generate theme resource data based on the color prompt words, content prompt words, reference hue range, and preset theme generation information. This application can dynamically generate theme resource data based on theme generation instructions, meeting users' personalized needs.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a topic generation method, apparatus, electronic device, and readable storage medium. Background Technology

[0002] With the rapid development of smart cockpit technology, the personalization level of in-vehicle themes, as a visual carrier for human-computer interaction, has become key to improving user experience.

[0003] In related technologies, the development and use of in-vehicle theme are separated. Users can only download pre-set theme templates from the app store or combine pre-set theme templates, which cannot achieve personalized customization and results in a poor user experience. Summary of the Invention

[0004] This application provides a topic generation method, apparatus, electronic device, and readable storage medium, which can dynamically generate topic resource data based on topic generation instructions, thereby meeting users' personalized needs.

[0005] In a first aspect, embodiments of this application provide a topic generation method applied to a server. The method includes: receiving a topic generation instruction, an environmental image of the vehicle's environment, and the vehicle's driving data; performing multimodal fusion processing on the topic generation instruction, the environmental image, and the driving data to determine topic parameters, including a reference hue range; extracting topic keywords from the topic generation instruction and generating prompt words based on the topic keywords, including color prompt words and content prompt words; and using a topic generation model to generate topic resource data based on the color prompt words, content prompt words, the reference hue range, and preset topic generation information.

[0006] In one possible implementation, the above-mentioned multimodal fusion processing of topic generation instructions, environmental images, and driving data to determine topic parameters includes: semantic recognition of topic generation instructions to determine topic keywords, and feature extraction of topic keywords to obtain semantic feature vectors; feature extraction of environmental images to determine visual feature vectors, which are used to characterize the environmental information of the vehicle's environment; alignment and fusion processing of semantic feature vectors, visual feature vectors, and data feature vectors corresponding to driving data to obtain multimodal fusion features; and determination of topic parameters based on the parameter generation model and the multimodal fusion features.

[0007] In one possible implementation, the aforementioned preset theme generation information includes preset wallpaper information and preset icon information, and the aforementioned theme resource data includes wallpaper data and icon data. The aforementioned generation of theme resource data using a theme generation model based on color hints, content hints, a reference color range, and the preset theme generation information includes: determining a target color range based on the reference color range and color hints, wherein the target color range is contained within the reference color range; generating wallpaper hints based on the color hints, content hints, and preset wallpaper information; generating wallpaper data using the theme generation model based on the wallpaper hints and the target color range; generating icon hints based on the color hints, content hints, and preset icon information; and generating icon data using the theme generation model based on the icon hints and the target color range.

[0008] In one possible implementation, the above-mentioned use of the theme generation model to generate wallpaper data based on wallpaper tips and target color range includes: using the theme generation model to generate wallpaper data based on wallpaper tips, target color range, and style feature vector, wherein the style feature vector is determined based on the theme reference image; the above-mentioned use of the theme generation model to generate icon data based on icon tips and target color range includes: using the theme generation model to generate icon data based on icon tips, target color range, and style feature vector.

[0009] In one possible implementation, the above-mentioned use of the theme generation model to generate wallpaper data based on wallpaper hints, target color range, and style feature vector includes: using the theme generation model to generate wallpaper data based on wallpaper hints, target color range, style feature vector, and preset quality hints; the above-mentioned use of the theme generation model to generate icon data based on icon hints, target color range, and style feature vector includes: using the theme generation model to generate icon data based on icon hints, target color range, style feature vector, and preset quality hints.

[0010] In one possible implementation, the method further includes: receiving a subject reference image sent by the vehicle; and extracting a style feature vector from the reference subject image.

[0011] In one possible implementation, the aforementioned theme resource data also includes a theme preview image, and the method further includes generating a theme preview image based on wallpaper data, icon data, and display parameters of the vehicle's infotainment system.

[0012] In one possible implementation, the aforementioned theme resource data further includes color value data, and the method further includes: after generating wallpaper data, performing cluster analysis on the wallpaper data to determine the theme color of the wallpaper data; comparing the theme color with candidate colors to determine the highlight color; determining the text color based on the brightness of the theme color; and generating color value data based on the theme color, highlight color, and text color. The above-mentioned generation of a theme preview image based on wallpaper data and icon data includes: generating a theme preview image based on wallpaper data, icon data, display parameters of the vehicle system, and color value data.

[0013] In one possible implementation, generating prompt words based on topic keywords includes: performing color association on topic keywords to obtain color prompt words; and expanding topic keywords to obtain content prompt words.

[0014] In one possible implementation, the method further includes: receiving a theme download instruction sent by a vehicle; in response to the theme download instruction, determining theme resource data matching the theme download instruction; and sending the theme resource data to the vehicle so that the vehicle updates the theme of the vehicle system according to the theme resource data.

[0015] Secondly, embodiments of this application provide a topic generation apparatus configured on a server, comprising:

[0016] The receiving module is used to receive the theme generation command, environmental images of the vehicle's environment, and the vehicle's driving data;

[0017] The parameter determination module is used to perform multimodal fusion processing on the theme generation command, environmental image and driving data to determine the theme parameters, including the reference tone range.

[0018] The prompt word generation module is used to extract theme keywords from theme generation instructions and generate prompt words based on the theme keywords. The prompt words include color prompt words and content prompt words.

[0019] The theme generation module is used to generate theme resource data based on the theme generation model, color prompts, content prompts, reference color ranges, and preset theme generation information.

[0020] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0021] The memory stores the instructions that the computer executes;

[0022] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0023] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0024] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0025] In this embodiment, upon receiving a user-inputted theme generation command, theme resource data can be dynamically generated based on the command, meeting the user's personalized needs. Furthermore, theme parameters are generated by integrating the user-inputted theme generation command, environmental images of the vehicle's surroundings, and the vehicle's current driving data. Then, prompt words are generated based on these parameters. Using a theme generation model, theme resource data is generated according to color prompt words, content prompt words, reference color ranges, and preset theme generation information. This generates theme resources that better match the user's intent and harmonize with the vehicle's environment, improving visual comfort and providing a more immersive interactive experience. Additionally, it avoids excessively high or low saturation and brightness of theme colors, preventing the theme wallpaper from interfering with driving and improving driving safety.

[0026] Furthermore, the embodiments of this application can realize the automated generation of theme resources for the vehicle system, and achieve dynamic adaptation of theme resources and vehicle system through cloud collaboration, thereby improving wallpaper generation efficiency. Attached Figure Description

[0027] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0028] Figure 1 This is a schematic diagram illustrating an application scenario of a topic generation method according to an embodiment of this application;

[0029] Figure 2 One of the flowcharts illustrating the subject generation method provided in this application embodiment;

[0030] Figure 3 A second schematic flowchart illustrating the subject generation method provided in this application embodiment;

[0031] Figure 4 The third flowchart illustrating the subject generation method provided in this application embodiment;

[0032] Figure 5The fourth flowchart illustrating the subject generation method provided in this application embodiment;

[0033] Figure 6 Fifth flowchart illustrating the subject generation method provided in this application embodiment;

[0034] Figure 7 A flowchart illustrating the subject generation method provided in this application embodiment is shown in Figure 6.

[0035] Figure 8 A schematic diagram of a theme preview image provided for an embodiment of this application;

[0036] Figure 9 A schematic diagram of another subject preview image provided for an embodiment of this application;

[0037] Figure 10 Seventh schematic flowchart of the subject generation method provided in the embodiments of this application;

[0038] Figure 11 This is a schematic diagram of the structure of a topic generation device provided in an embodiment of this application;

[0039] Figure 12 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0040] The embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described below do not represent all embodiments consistent with this application. They are merely examples of systems and methods consistent with some aspects of this application as detailed in the claims.

[0041] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.

[0042] The terms "first," "second," "third," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar or related objects or entities, and do not necessarily imply a specific order or sequence, unless otherwise specified. It should be understood that such terms are interchangeable where appropriate.

[0043] The terms “comprising” and “having”, and any variations thereof, are intended to cover but not exclude inclusion, for example, a product or device that includes a range of components is not necessarily limited to all of the components that are clearly listed, but may include other components that are not clearly listed or that are inherent to such product or device.

[0044] The term "module" refers to any known or subsequently developed hardware, software, firmware, artificial intelligence, fuzzy logic, or combination of hardware and / or software code that is capable of performing the functions associated with that element.

[0045] With the rapid development of smart cockpit technology, the personalization level of in-vehicle themes, as a visual carrier for human-computer interaction, has become key to improving user experience.

[0046] In related technologies, the development and use of in-vehicle theme are separated. Users can only download pre-set theme templates from the app store or combine pre-set theme templates, which cannot achieve personalized customization and results in a poor user experience.

[0047] In addition, the development process of in-vehicle infotainment themes involves multiple stages such as requirements analysis, visual design, multi-resolution adaptation, development and launch. The process is cumbersome and relies on manual labor, resulting in a long development cycle and low efficiency.

[0048] In some embodiments, wallpapers are dynamically generated using existing text-based image models. However, this approach does not consider the functional and security requirements of the vehicle infotainment system, resulting in numerous adaptation issues with the generated content, such as complex compositions, obstruction of interactive controls, and color mismatches. These issues make it difficult to meet the functional zoning and layout specifications of the vehicle infotainment interface, thus affecting the user experience.

[0049] In view of this, embodiments of this application provide a theme generation method. Upon receiving a theme generation instruction from a user, the method dynamically generates theme resource data based on the instruction, meeting the user's personalized needs. Furthermore, by integrating the user-input theme generation instruction, an environmental image of the vehicle's surroundings, and the vehicle's current driving data, theme parameters are generated. Then, prompt words are generated based on these theme parameters. Using a theme generation model, theme resource data is generated according to color prompt words, content prompt words, a reference color range, and preset theme generation information. This generates theme resources that better match the user's intent and are harmonious with the vehicle's environment, improving visual comfort and providing a more immersive interactive experience. Additionally, it avoids excessively high or low saturation and brightness of theme colors, preventing the theme wallpaper from interfering with driving and improving driving safety.

[0050] Before introducing the topic generation method provided in the embodiments of this application, the application scenarios of the topic generation method will be explained first.

[0051] Figure 1 This is a schematic diagram illustrating an application scenario of a topic generation method according to an embodiment of this application.

[0052] In the embodiments of this application, such as Figure 1 As shown, the vehicle infotainment system (also known as the in-vehicle system) on vehicle 10 is connected to server 20 (also known as the cloud). In response to a user-inputted theme generation command, the vehicle infotainment system acquires environmental images of the environment in which vehicle 10 is located and the current driving data of vehicle 10. After acquiring the environmental images and driving data, the vehicle infotainment system sends a theme generation request to server 20, carrying the theme generation command, environmental images, and driving data. In response to the theme generation request, server 20 determines theme parameters based on the theme generation command, environmental images, and driving data; server 20 performs semantic parsing of the theme generation command to determine theme keywords, and generates color prompts and content prompts based on the theme keywords. Then, using a theme generation model, based on the color prompts, content prompts, theme parameters, and preset theme generation specifications, it generates theme resource data. After generating the theme resource data, server 20 stores the theme resource data and sends theme generation completion information to the vehicle infotainment system of vehicle 10. Upon receiving the theme generation completion information, in response to a user-inputted theme download command, the vehicle infotainment system retrieves the corresponding theme resource data from server 20 and updates its theme.

[0053] Of course, in some examples, after generating the theme resource data, server 20 can also send the generated theme resource data to the vehicle infotainment system. After receiving the theme resource data sent by server 20, the vehicle infotainment system updates its theme according to the theme resource data.

[0054] The subject generation method provided in this application embodiment will be described in detail below with reference to the accompanying drawings and application scenarios.

[0055] Figure 2 This is a flowchart illustrating a topic generation method provided in an embodiment of this application. Figure 2 As shown, this topic generation method may include the following steps:

[0056] S201 receives a topic generation command, an environmental image of the vehicle's surroundings, and the vehicle's driving data.

[0057] S202, perform multimodal fusion processing on the theme generation instruction, environmental image and driving data to determine theme parameters, including reference tone range.

[0058] S203: Extract theme keywords from theme generation instructions, and generate prompt words based on theme keywords. Prompt words include color prompt words and content prompt words.

[0059] S204 utilizes a theme generation model to generate theme resource data based on color cue words, content cue words, reference color range, and preset theme generation information.

[0060] In this embodiment, upon receiving a user-inputted theme generation command, theme resource data can be dynamically generated based on the command, meeting the user's personalized needs. Furthermore, theme parameters are generated by integrating the user-inputted theme generation command, environmental images of the vehicle's surroundings, and the vehicle's current driving data. Then, prompt words are generated based on these parameters. Using a theme generation model, theme resource data is generated according to color prompt words, content prompt words, reference color ranges, and preset theme generation information. This generates theme resources that better match the user's intent and harmonize with the vehicle's environment, improving visual comfort and providing a more immersive interactive experience. Additionally, it avoids excessively high or low saturation and brightness of theme colors, preventing the theme wallpaper from interfering with driving and improving driving safety.

[0061] The subject generation method provided in the embodiments of this application will be described in detail below.

[0062] Figure 3 Another topic generation method provided in this application embodiment. This topic generation method is applied to vehicles, such as... Figure 3 As shown, the topic generation method includes the following steps:

[0063] S301 receives the topic generation command input by the user.

[0064] Specifically, the vehicle's infotainment system receives a topic generation command input by the user. For example, the topic generation command can be a voice command or a text command. A voice command can be a spoken message input by the user, while a text command can be a text message input by the user.

[0065] In some examples, in response to a user's action, the vehicle's infotainment system displays a theme generation entry point, and in response to selecting the theme generation entry point, the user enters a theme generation command. In other examples, the user can also trigger the theme generation function through the vehicle's infotainment system's voice assistant and enter the corresponding voice command.

[0066] Taking the theme generation command as a voice command as an example, the theme generation command is "Generate a starry sky theme with a cool color tone for the car system". The theme generation command can also be "Generate a Kobe basketball-style car system theme".

[0067] S302, in response to the topic generation command, acquires environmental images of the vehicle's surroundings and the vehicle's driving data.

[0068] Specifically, after receiving the user's input instruction to generate a theme, the vehicle uses image sensors (i.e., onboard cameras) to collect environmental images of the vehicle's surroundings and obtain the vehicle's current driving data.

[0069] The environmental image represents the vehicle's current environment. Through the environmental image, environmental information about the vehicle's current environment can be identified. For example, environmental information may include light intensity and scene element information within the vehicle's environment. Scene element information may include, for example, road and building information.

[0070] Driving data can characterize the vehicle's current driving status. For example, driving data may include vehicle speed data, timestamp data, and geographic location data. Vehicle speed data characterizes the vehicle's speed at the time the user-inputted topic generation command is received. Timestamp data represents the current time at which the user-inputted topic generation command is received. Geographic location data represents the vehicle's current map location at the time the user-inputted topic generation command is received.

[0071] S303, a theme generation request is sent to the server. The theme generation request carries theme generation instructions, environmental images, and driving data. The theme generation request is used to request the server to generate theme resource data based on the theme generation instructions, environmental images, and driving data.

[0072] It should be noted that the process of generating the topic resource data can be found below, and will not be repeated here.

[0073] In some embodiments, after the server generates and stores the theme resource data, such as Figure 4 As shown, the method may further include:

[0074] S401 receives the user's input command to download the theme.

[0075] After generating the theme resource data, the server can send a notification message to the vehicle's infotainment system to indicate that the theme resource generation is complete. Users can then download the required theme resource data.

[0076] S402, in response to a theme download command, retrieves theme resource data matching the theme download command.

[0077] S403 updates the theme of the vehicle's infotainment system based on theme resource data.

[0078] In this embodiment, after generating theme resource data, in response to a user-inputted theme download command, the corresponding theme resource data is obtained, and the theme of the vehicle's infotainment system is updated. This ensures that the generated resource data is perfectly compatible with the vehicle's infotainment system, allowing users to instantly access and apply a vast array of personalized themes, significantly improving theme update efficiency and user experience. Furthermore, this embodiment automates the entire process from understanding user intent and intelligent generation to application download, completely changing the traditional model that relies on manual design, packaging, and deployment. Additionally, cloud collaboration enables dynamic adaptation of theme resources to the vehicle's infotainment system, improving wallpaper generation efficiency.

[0079] The process of generating thematic resource data will be explained in detail below with reference to the accompanying diagram.

[0080] like Figure 2 As shown, this topic generation method may include the following steps:

[0081] S201 receives a topic generation command, an environmental image of the vehicle's surroundings, and the vehicle's driving data.

[0082] Specifically, the server receives a topic generation request from the vehicle's infotainment system. The topic generation request carries a topic generation instruction, an environmental image of the vehicle's surroundings, and the vehicle's current driving data.

[0083] The topic generation instruction can characterize the topic content and / or tone corresponding to the topic resource data to be generated. For example, the topic generation instruction can be a voice instruction or a text instruction. A voice instruction can be a piece of speech input by the user, and a text instruction can be a piece of text input by the user.

[0084] Taking the theme generation command as a voice command as an example, the theme generation command is "Generate a starry sky theme with a cool color tone for the car system". The theme generation command can also be "Generate a Kobe basketball-style car system theme".

[0085] Environmental images can characterize the visual environment in which a vehicle is currently located. Through environmental images, environmental information about the vehicle's current surroundings can be identified. For example, environmental information may include light intensity and scene element information within the vehicle's environment. Scene element information may include, for example, road and building information.

[0086] Driving data can characterize the vehicle's current driving status. For example, driving data may include vehicle speed data, timestamp data, and geographic location data. Vehicle speed data characterizes the vehicle's speed at the time the user-inputted topic generation command is received. Timestamp data represents the current time at which the user-inputted topic generation command is received. Geographic location data represents the vehicle's current map location at the time the user-inputted topic generation command is received.

[0087] S202, performs multimodal fusion processing on the theme generation command, environmental images and driving data to determine the theme parameters.

[0088] In some embodiments, performing multimodal fusion processing on topic generation instructions, environmental images, and driving data to determine topic parameters may include: extracting semantic features from the topic generation instructions to obtain a semantic feature vector; extracting visual features from the environmental images to obtain a visual feature vector; aligning and fusing the semantic feature vector, the visual feature vector, and the data feature vector corresponding to the driving data to obtain multimodal fusion features; and determining topic parameters based on the multimodal fusion features, wherein the topic parameters include a reference tone range.

[0089] The theme parameters may include a reference hue range and / or a theme display mode. The reference hue range characterizes the color tone of the theme to be generated. For example, the reference hue range may include the value ranges of the R, G, and B channels. The theme display mode may include a dark mode and a light mode.

[0090] In some examples, such as Figure 5 As shown, S202 may include the following steps:

[0091] S501 performs semantic recognition on the topic generation instruction, determines the topic keywords, and extracts features from the topic keywords to obtain a semantic feature vector.

[0092] Specifically, in response to a topic generation request, the server performs semantic recognition and parsing on the topic generation instruction to determine the topic keywords.

[0093] Thematic keywords can be keywords that characterize the theme content and / or thematic tone. It is understood that there can be one or more thematic keywords.

[0094] For example, if the theme generation command is "generate a starry sky theme with a cool color tone for the vehicle system", the theme keywords can include "starry sky" and "cool color tone".

[0095] For example, if the theme generation command is "generate a Kobe basketball-style car infotainment theme", the theme keywords can include "Kobe" and "basketball".

[0096] For example, a pre-trained large language model can be used to perform semantic recognition on topic generation instructions to determine topic keywords. For example, the pre-trained large language model can be an automatic speech recognition (ASR) model, a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory (LSTM) network, etc.

[0097] After determining the topic keywords, features are extracted from them to obtain semantic feature vectors. For example, a word embedding model can be used to process the topic keywords and obtain semantic feature vectors. For example, the word embedding model can be a bidirectional encoder representation from transformers (BERT).

[0098] S502, extract features from the environmental image to determine the visual feature vector, which can characterize the environmental information of the vehicle's surroundings.

[0099] The environmental information may include light intensity and scene element information in the vehicle's environment. Scene element information may include, for example, road information (such as urban roads, highways, and tunnels) and building information.

[0100] For example, the light intensity of the environment in which the vehicle is located can be classified as high light intensity, medium light intensity, and low light intensity. In other words, the obtained light intensity can be used as light intensity classification information.

[0101] In some examples, a pre-trained visual processing model is used to extract features from the environmental image, resulting in a visual feature vector. This visual processing model can be a deep convolutional neural network, such as ResNet-50 or a visual Transformer.

[0102] In practice, the visual processing model is trained on an initial visual processing model based on multiple sets of first sample data. Each set of first sample data can include historical environmental images collected by any vehicle, along with the corresponding actual illumination intensity and scene element information. The historical environmental images are input into the initial visual processing model, which outputs a set of predicted illumination intensity and predicted scene element information. Using the actual illumination intensity and scene element information corresponding to the historical environmental images as supervision information, the initial visual processing model is iteratively trained to obtain the final visual processing model.

[0103] S503 aligns and fuses the semantic feature vector, visual feature vector, and data feature vector corresponding to the driving data to obtain multimodal fusion features.

[0104] Specifically, the pre-trained cross-modal alignment model uses an attention mechanism to learn and establish semantic relationships between features of different modalities. Within a unified semantic space, it aligns and fuses semantic feature vectors, visual feature vectors, and data feature vectors corresponding to driving data, and outputs a multimodal fusion feature matrix, i.e., multimodal fusion features.

[0105] For example, a cross-modal alignment model can be constructed using the Transformer architecture.

[0106] S504, determine the topic parameters based on the parameter generation model and multimodal fusion features.

[0107] In this embodiment, different topic keywords (i.e., semantic feature vectors) correspond to different topic colors, and thus different color tones. For example, taking the topic keyword "starry sky" as an example, the corresponding topic colors could be blue, dark blue, purple, or dark purple. Taking the topic keyword "sunflower" as an example, the corresponding topic colors could be green and yellow. Additionally, topic keywords can also include hue keywords, such as cool tones or warm tones. For the same topic color, different hue keywords will result in different color tones for the corresponding topics.

[0108] Different environmental conditions (such as light intensity) correspond to different saturation and brightness of the theme color. For example, if the light intensity of the vehicle's environment is high, the saturation and brightness of the theme color will be high. If the light intensity of the vehicle's environment is low, the saturation and brightness of the theme color will be low.

[0109] Based on different driving data, the vehicle's timestamp can be used to determine whether the current driving is at night or during the day. If the vehicle is currently driving at night, the saturation and brightness of the theme color will be higher. If the vehicle is currently driving during the day, the saturation and brightness of the theme color will be lower.

[0110] Furthermore, based on the vehicle's geographical location information, the dominant color of the environment in which the vehicle is located can be determined. Based on the dominant color of the environment in which the vehicle is located, the color tone of the theme can be adjusted to enhance the immersive experience of the theme.

[0111] Based on this, the parameter generation model uses topic keywords (i.e. semantic feature vectors) to determine the initial color tone range, and then adjusts the initial color tone according to the visual feature vectors representing environmental information and the feature vectors of driving data to obtain the reference color tone range.

[0112] For example, if the theme generation command is "generate a starry sky theme with a cool color tone for the vehicle", and the light intensity is low and the vehicle is currently driving at night, the reference color tone range is RGB: 30-50, 60-80, 120-150.

[0113] Additionally, theme parameters can include the theme display mode. Based on the vehicle's timestamp, it can be determined whether the vehicle is currently driving at night or during the day, thus determining the theme display mode. If the vehicle is currently driving at night, the theme display mode is dark mode. If the vehicle is currently driving during the day, the theme display mode is light mode. The color tone of the theme also differs depending on the theme display mode. Therefore, reference color gradations can be obtained for both dark and light modes.

[0114] In some examples, a pre-trained parameter generation model is used to process multimodal fusion features to obtain topic parameters. This parameter generation model can be a context-aware model, such as a Long Short-Term Memory network, a Bayesian network, a memory-based neural network, an attention mechanism, or a Transformer encoder.

[0115] In practice, the parameter generation model is trained on the initial parameter generation model based on multiple sets of second sample data. Each set of second sample data may include multimodal fusion features determined using a set of sample data (i.e., topic generation instructions, environmental images, and driving data), and may also include actual topic parameters corresponding to the multimodal fusion features (such as actual tone range and / or actual topic display mode). During training, the multimodal fusion features used for training are input into the initial parameter generation model, which outputs a set of predicted topic parameters (such as predicted tone range and / or predicted topic display mode). Using the actual topic parameters as supervision information, the initial parameter generation model is iteratively trained to obtain the parameter generation model.

[0116] In this embodiment, upon receiving a user-inputted theme generation command, theme resource data can be dynamically generated based on the command, thus meeting the user's personalized needs. Furthermore, theme parameters are generated by integrating the user-inputted theme generation command, environmental images of the vehicle's surroundings, and the vehicle's current driving data. Then, prompt words are generated based on these parameters. Using a theme generation model, theme resource data is generated according to color prompt words, content prompt words, reference color ranges, and preset theme generation information. This approach generates theme resources that better match the user's intent and are harmonious with the vehicle's environment, improving visual comfort and providing a more immersive interactive experience.

[0117] S203: Extract theme keywords from theme generation instructions, and generate prompt words based on theme keywords. Prompt words include color prompt words and content prompt words.

[0118] Color cues indicate the color tone of the generated theme resource. Content cues indicate the theme elements that can be included in the generated theme resource.

[0119] Specifically, after obtaining the theme keywords, color associations are used to generate color-coded prompts. The theme keywords are then expanded to generate content-coded prompts.

[0120] For example, taking the theme generation command as "generate a starry sky theme with a cool color tone for the in-vehicle infotainment system", the theme keyword includes "starry sky". Color associations with "starry sky" yield color prompts such as "dark blue" and "dark purple". Expanding on "starry sky" yields content prompts such as "starry sky", "stars", "Milky Way", "fireflies", and "autumn".

[0121] For example, taking the theme generation command "Generate a Kobe Bryant basketball-themed car infotainment theme" as an example, the theme keywords include "Kobe" and "basketball". Color associations for "Kobe" yield color hints such as "dark purple" and "matte gold". Expanding on "Kobe" and "basketball" yields content hints such as "basketball court", "number 24", "championship trophy", and "jersey texture".

[0122] After generating color and content hints, the hints are checked for violations to ensure that the hints used to generate wallpapers and icons meet the requirements.

[0123] S204 utilizes a theme generation model to generate theme resource data based on color cue words, content cue words, reference color range, and preset theme generation information.

[0124] The preset theme generation information may include preset wallpaper information and preset icon information. The preset wallpaper information may include at least one of the following: wallpaper layout information, wallpaper resolution, wallpaper lighting and shadow information, wallpaper visual focus information, and wallpaper element density. The preset icon information may include at least one of the following: icon rounded corner information, icon line width information, icon size information, icon resolution, and icon symbol semantic style. It is understood that the preset wallpaper information and preset icon information can be set according to the design requirements of the vehicle infotainment system, and this application embodiment does not specifically limit them.

[0125] In some examples, such as Figure 6 As shown, S204 may include the following steps:

[0126] S601, determine the target color range based on the reference color range and color cues, the target color range being contained within the reference color range.

[0127] For example, taking the color cues "dark blue" and "dark purple" as examples, the sub-color ranges corresponding to "dark blue" and "dark purple" within the reference color range are determined as the target color range.

[0128] For example, taking the color cues "dark purple" and "matte gold" as examples, the sub-color ranges within the reference color range that correspond to "dark purple" and "matte gold" are determined as the target color range.

[0129] S602 generates wallpaper prompts based on color prompts, content prompts, and preset wallpaper information.

[0130] For example, wallpaper tips are positive tips used by the theme generation model to generate wallpaper data.

[0131] For example, using color cues including "deep blue" and "deep purple," and content cues including "starry sky," "stars," "Milky Way," "fireflies," and "aurora," the wallpaper cues would be: "A minimalist deep-space universe wallpaper, primarily in deep blue and deep purple, with matte silver stars. The image depicts a rotating Milky Way and hazy nebulae, with shooting stars streaking across the sky. The background features large areas of dark white space and a depth-of-field blurring effect, with visual elements concentrated on the right side of the image. 4K resolution, cinematic lighting and shadows, rich detail, low saturation, and low brightness."

[0132] For example, if the color cues include "dark purple" and "matte gold", and the content cues include "Kobe", "basketball", "basketball court", "number 24", "championship trophy" and "jersey texture", the wallpaper cues would be "a minimalist wallpaper using dark purple and matte gold, showcasing an abstract Kobe theme, with a large area of ​​white space and depth of field blur in the background, 4K resolution, and cinematic lighting".

[0133] S603 uses a theme generation model to generate wallpaper data based on wallpaper tips and target color range.

[0134] In some examples, wallpaper tips are input into the CLIP text encoder, which outputs the corresponding text semantic vectors. These text semantic vectors and the target tonal range are then used as conditions to input into a topic generation model. Through configured sampler parameters, iterative denoising sampling is performed in the latent space to generate latent space features. A VAE decoder is then used to convert these latent space features into wallpaper images, thus obtaining the wallpaper data.

[0135] For example, the topic generation model can be a latent space diffusion model, such as a visual model (StableDiffusion).

[0136] Optionally, when generating wallpaper data, style feature vectors can be incorporated. These style feature vectors are determined based on a user-inputted theme reference image.

[0137] For example, a theme generation model can be used to generate wallpaper data based on wallpaper tips, target color range, and style feature vectors.

[0138] Specifically, the wallpaper prompts are input into the CLIP text encoder, which outputs the corresponding text semantic vectors. These text semantic vectors, style feature vectors, and target tone ranges are then input into a theme generation model. Through configured sampler parameters, iterative denoising sampling is performed in the latent space to generate latent space features. A VAE decoder is then used to convert these latent space features into wallpaper images, thus obtaining the wallpaper data.

[0139] In this embodiment, when a user inputs a theme reference image, the theme reference image can be used to determine a style feature vector, and then combined with the style feature vector to generate wallpaper data. This can generate theme wallpapers that better match the user's intentions and improve wallpaper generation efficiency.

[0140] Optionally, when generating wallpaper data, quality tips can be incorporated. These quality tips can be positive or negative. Examples of quality tips include "high resolution," "high-quality vector graphics," "avoid highly saturated reds," "prohibit complex textures," and "prohibit disturbing elements such as damaged animal faces and multiple eyes."

[0141] Specifically, wallpaper hints and quality hints are input into the CLIP text encoder, which outputs text semantic vectors corresponding to the wallpaper hints and quality hints. These text semantic vectors, style feature vectors, and target tone ranges are then input into a theme generation model. Through configured sampler parameters, iterative denoising sampling is performed in the latent space to generate latent space features. A VAE decoder is then used to convert these latent space features into wallpaper images, thus obtaining the wallpaper data.

[0142] In this embodiment, wallpaper data is generated by combining quality prompts, which can avoid the generated wallpapers being incompatible with the vehicle system and improve the compatibility of the theme wallpapers.

[0143] S604 generates icon prompts based on color prompts, content prompts, and preset icon information.

[0144] For example, the icon hints are positive hints, used by the theme generation model to generate icon data.

[0145] S605 uses a theme generation model to generate icon data based on icon hints and target color range.

[0146] In some examples, icon prompts are input into the CLIP text encoder, which outputs text semantic vectors corresponding to the icon prompts. These text semantic vectors and the target tonal range are then input into a topic generation model. Through configured sampler parameters, iterative denoising sampling is performed in the latent space to generate latent space features. A VAE decoder is then used to convert these latent space features into icon images, thus obtaining the icon data.

[0147] Optionally, when generating icon data, style feature vectors can be incorporated. These style feature vectors are determined based on a user-inputted theme reference image. For example, a theme reference image sent by a vehicle is received; a style feature vector is extracted from the reference theme image.

[0148] For example, a theme generation model can be used to generate icon data based on icon hints, target color range, and style feature vectors.

[0149] Specifically, the icon prompts are input into the CLIP text encoder, which outputs the corresponding text semantic vectors. These text semantic vectors, style feature vectors, and target tone ranges are then input into the topic generation model. Using configured sampler parameters, iterative denoising sampling is performed in the latent space to generate latent space features. A VAE decoder is then used to convert these latent space features into icon images, thus obtaining the icon data.

[0150] In this embodiment, when a user inputs a theme reference image, the style feature vector can be determined using the theme reference image, and then icon data can be generated by combining the style feature vector. This can generate theme icons that better match the user's intentions and improve icon generation efficiency.

[0151] Optionally, quality cues can be incorporated into the icon data generation process. These quality cues can be positive or negative. Examples of quality cues include "high resolution," "high-quality vector graphics," "avoid high-saturation reds," "avoid complex textures," and "flat design."

[0152] Specifically, icon hints and quality hints are input into the CLIP text encoder, which outputs text semantic vectors corresponding to the icon hints and quality hints. These text semantic vectors, style feature vectors, and target tone ranges are then input into the topic generation model. Through configured sampler parameters, iterative denoising sampling is performed in the latent space to generate latent space features. A VAE decoder is then used to convert these latent space features into icon images, thus obtaining the icon data.

[0153] In this embodiment, combining quality prompts with icon data generation can avoid generating icons that are incompatible with the vehicle system and improve the compatibility of theme icons.

[0154] After generating wallpaper and icon data, the wallpaper data undergoes post-processing based on the vehicle's infotainment system's display parameters. Using the icon draft, edge control technology is applied to the icon images for edge control. Furthermore, the icon images are scaled and cropped according to the vehicle's infotainment system's display parameters to ensure the wallpaper data and icon images are compatible with the system. The vehicle's infotainment system's display parameters may include the screen resolution and screen size.

[0155] In some embodiments, the theme resource data also includes color value data. For example... Figure 7 As shown, the process of generating color value data in this theme generation method may include the following steps:

[0156] S701: After generating wallpaper data, perform cluster analysis on the wallpaper data to determine the theme color of the wallpaper data.

[0157] For example, based on wallpaper data, a clustering algorithm (e.g., K-means, setting K=10) is performed in the HSL color space to extract the main color features of the image, and the extracted color features are sorted according to their brightness. Through principal component analysis, color features with a weight ratio greater than a preset ratio (e.g., 35%) in the cluster are identified as the theme color.

[0158] S702 compares the theme color with the candidate colors to determine the highlight color.

[0159] Among multiple candidate colors, the candidate color with a contrast greater than the preset contrast to the theme color is selected as the highlight color.

[0160] Specifically, from the preset candidate color set, a color difference analysis algorithm is applied to determine the color difference (i.e., ΔE value) between each candidate color and the theme color. Candidate colors with a color difference greater than a preset threshold (such as 12) are identified as high-quantity colors.

[0161] S703, determine the text color based on the brightness of the theme color.

[0162] For example, if the brightness of the theme color is greater than the preset brightness, black will be used as the text color; if the brightness of the theme color is less than or equal to the preset brightness, white will be used as the text color. The preset brightness is, for example, 50.

[0163] S704 generates color value data based on theme color, highlight color, and text color.

[0164] In some embodiments, after generating color value data, the generated theme colors, highlight colors, and text colors are converted into key-value pair data structures that the vehicle system can directly parse, i.e., color value files, according to a predefined JSON Schema mapping specification. When packaging the color value files together with other theme resources (wallpapers, icons), a hash verification module is integrated. This hash verification module generates a unique SHA-256 checksum for the entire theme resource package (i.e., the theme resource data mentioned above) or its color value files, and stores or transmits it synchronously with the theme resource package to verify whether the theme resource package is complete, unaltered, or undamaged during transmission.

[0165] The serialized color value file (such as theme.json), along with wallpaper data, icon data, etc., is organized according to the directory structure required by the vehicle system and assembled into the final theme resource package (i.e., the theme resource data mentioned above). After receiving the theme resource package, the vehicle system can directly read and parse the color value file and automatically complete the color switching of the system interface.

[0166] In this embodiment of the application, after generating wallpaper data, the theme color (i.e., the main color tone) is extracted from the generated wallpaper image, and the highlight color and text color are determined according to the theme color to output color value data. In this way, by using color value data, wallpaper data and icon data to generate theme resource data, the user's visual comfort can be improved and the problem of poor readability in night driving scenarios can be avoided.

[0167] In some embodiments, the theme resource data also includes a theme preview image. After generating wallpaper data and icon data, a theme preview image is generated based on the wallpaper data, icon data, and display parameters of the vehicle's infotainment system.

[0168] Optionally, a theme preview image can be generated based on wallpaper data, icon data, display parameters of the vehicle's infotainment system, and color value data. For example, the theme preview image can be as follows: Figure 8 and Figure 9 As shown. The theme preview image allows you to demonstrate the theme's effect to the user.

[0169] Specifically, a pre-developed Node.js service node is invoked. This Node.js service node integrates the target vehicle infotainment system's UI simulation framework. Based on the system's display parameters, this framework automatically renders a high-fidelity H5 preview page of the vehicle's interface. Wallpaper, icon, and color value data generated in the cloud are loaded into the H5 preview page, ensuring that the vehicle's UI elements (such as the status bar, cards, and buttons) apply the color values ​​and correctly display icons against the wallpaper background. A screenshot of the rendered preview page is taken using a headless browser (such as Puppeteer), generating a high-fidelity theme preview image. The generated theme preview image, along with wallpaper, icons, color value files, and other resources, are packaged into the final theme resource package.

[0170] In other words, the theme resource data in this application embodiment may include wallpaper data, icon data, color value files, and theme preview images.

[0171] In some embodiments, after generating topic resource data, the topic resource data is validated to output validated topic resource data.

[0172] For example, color space verification: verifies whether the color space (e.g., sRGB) of the wallpaper and icon data conforms to the display specifications of the in-vehicle infotainment system. Resolution and size verification: verifies whether the size of the wallpaper data is within the allowable tolerance range based on the screen resolution of the in-vehicle infotainment system. Icon quality verification: detects the edge sharpness of the icon data to ensure that it remains clearly distinguishable at different resolutions. Content security verification: performs theme element analysis on the wallpaper and icon data to verify whether there is any inappropriate or non-compliant content. This ensures that the generated theme resource data meets user needs and is compatible with the in-vehicle infotainment system.

[0173] After verifying the theme resource data, the data is packaged, packaged, and stored. This allows the vehicle's infotainment system to retrieve the theme resource data from the server in response to a user's input request to download the theme.

[0174] The following is a specific example illustrating the subject generation method provided in the embodiments of this application.

[0175] Figure 10 This is a flowchart illustrating a topic generation method provided in an embodiment of this application. Figure 10 As shown, the topic generation method includes the following steps:

[0176] S1001 receives the topic generation command input by the user.

[0177] S1002, in response to the topic generation command, acquires the environmental image of the vehicle's surroundings and the vehicle's current driving data.

[0178] S1003 performs multimodal fusion processing on the theme generation command, environmental images, and driving data to determine the theme parameters.

[0179] S1004: Extract theme keywords from theme generation instructions, and generate color prompts and content prompts based on the theme keywords.

[0180] S1005 generates wallpaper prompts and icon prompts based on color prompts, content prompts, preset wallpaper information, and preset icon information.

[0181] S1006, determine whether a subject reference image has been received. If yes, execute S1007; otherwise, execute S1009.

[0182] S1007, Extract style feature vectors from reference subject images.

[0183] S1008 optimizes wallpaper and icon prompts using style feature vectors.

[0184] S1009 uses a theme generation model to generate wallpaper data and icon data based on wallpaper tips, icon tips, and theme parameters.

[0185] S1010: After generating wallpaper data, analyze the wallpaper data to determine color value data.

[0186] S1011 generates a theme preview image based on wallpaper data, icon data, vehicle infotainment system display parameters, and color value data.

[0187] S1012, the wallpaper data, icon data, color value data and theme preview image are packaged and processed to obtain theme resource data.

[0188] This application also provides a topic generation apparatus. For example... Figure 11 As shown, the theme generation device 1100 includes a receiving module 1101, a parameter determination module 1102, a prompt word generation module 1103, and a theme generation module 1104. The receiving module 1101 receives a theme generation command, an environmental image of the vehicle's environment, and the vehicle's driving data. The parameter determination module 1102 performs multimodal fusion processing on the theme generation command, the environmental image, and the driving data to determine theme parameters, including a reference hue range. The prompt word generation module 1103 extracts theme keywords from the theme generation command and generates prompt words based on these keywords, including color prompt words and content prompt words. The theme generation module 1104 uses a theme generation model to generate theme resource data based on color prompt words, content prompt words, the reference hue range, and preset theme generation information.

[0189] In some embodiments, the parameter determination module 1102 is specifically used to perform semantic recognition on the topic generation instruction, determine topic keywords, and extract features from the topic keywords to obtain a semantic feature vector; extract features from the environmental image to determine a visual feature vector, which is used to characterize the environmental information of the vehicle's environment; align and fuse the semantic feature vector, visual feature vector, and data feature vector corresponding to the driving data to obtain a multimodal fusion feature; and determine the topic parameters based on the parameter generation model and the multimodal fusion feature.

[0190] In some embodiments, the preset theme generation information includes preset wallpaper information and preset icon information, and the theme resource data includes wallpaper data and icon data. The theme generation module 1104 is specifically used for: determining a target color range based on a reference color range and color cues, wherein the target color range is contained within the reference color range; generating wallpaper cues based on color cues, content cues, and preset wallpaper information; generating wallpaper data using a theme generation model based on the wallpaper cues and the target color range; generating icon cues based on color cues, content cues, and preset icon information; and generating icon data using a theme generation model based on the icon cues and the target color range.

[0191] In some embodiments, the theme generation module 1104 is specifically used to: generate wallpaper data using a theme generation model based on wallpaper tips, target color range, and style feature vector, wherein the style feature vector is determined based on a theme reference image; and generate icon data using a theme generation model based on icon tips, target color range, and style feature vector.

[0192] In some embodiments, the theme generation module 1104 is specifically used to: generate wallpaper data using a theme generation model based on wallpaper hints, target color range, style feature vector, and preset quality hints; and generate icon data using the theme generation model based on icon hints, target color range, style feature vector, and preset quality hints.

[0193] In some embodiments, the theme generation module 1104 is further configured to: generate a theme preview image based on wallpaper data, icon data, and display parameters of the vehicle's infotainment system.

[0194] In some embodiments, the theme generation module 1104 is further configured to: perform cluster analysis on the wallpaper data after generating the wallpaper data to determine the theme color of the wallpaper data; compare the theme color with candidate colors to determine the highlight color; determine the text color based on the brightness of the theme color; and generate color value data based on the theme color, highlight color, and text color.

[0195] In some embodiments, the theme generation module 1104 is specifically used to generate a theme preview image based on wallpaper data, icon data, display parameters of the vehicle system, and color value data.

[0196] In some embodiments, the prompt word generation module 1103 is specifically used to perform color association on the theme keywords to obtain color prompt words; and to expand the theme keywords to obtain content prompt words.

[0197] In some embodiments, the receiving module 1101 is further configured to receive a theme download instruction sent by the vehicle. The theme acquisition module is configured to, in response to the theme download instruction, determine theme resource data matching the theme download instruction. The sending module is configured to send the theme resource data to the vehicle, so that the vehicle updates the theme of the vehicle's infotainment system based on the theme resource data.

[0198] The electronic device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0199] Figure 12 A schematic diagram of the structure of the electronic device provided in this application. Figure 12 As shown, the electronic device 120 provided in this embodiment includes at least one processor 1201 and a memory 1202. Optionally, the electronic device 120 further includes a communication component 1203. The processor 1201, the memory 1202, and the communication component 1203 are connected via a bus 1204.

[0200] In a specific implementation, at least one processor 1201 executes computer execution instructions stored in memory 1202, causing at least one processor 1201 to perform the above-described method.

[0201] The specific implementation process of processor 1201 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0202] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0203] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0204] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0205] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0206] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0207] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0208] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0209] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0210] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0211] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0212] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0213] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0214] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A topic generation method characterized by comprising: Applied to a server, the method comprises: receiving theme generation instructions, environment images of an environment in which a vehicle is located, and driving data of the vehicle, the environment images being used to represent a visual environment in which the vehicle is currently located; performing semantic feature extraction on the theme generation instructions to obtain a semantic feature vector, performing visual feature extraction on the environment images to obtain a visual feature vector, and performing alignment and fusion processing on the semantic feature vector, the visual feature vector, and a data feature vector corresponding to the driving data to obtain a multi-modal fusion feature, the visual feature vector being used to represent environment information of the environment in which the vehicle is located, different environment information corresponding to different saturation and lightness of a theme color; determining theme parameters based on the multi-modal fusion feature, the theme parameters including a reference hue range; extracting theme keywords from the theme generation instructions, and generating prompt words based on the theme keywords, the prompt words including color prompt words and content prompt words, the reference hue range including a value range of R, G, and B channels of a theme to be generated; generating theme resource data using a theme generation model based on the color prompt words, the content prompt words, the reference hue range, and preset theme generation information.

2. The method of claim 1, wherein, The determination of the theme parameters based on the multi-modal fusion feature comprises: determining the theme parameters based on the multi-modal fusion feature and a pre-trained parameter generation model.

3. The method of claim 1, wherein, The preset theme generation information includes preset wallpaper information and preset icon information, the theme resource data includes wallpaper data and icon data, and the generation of the theme resource data using the theme generation model based on the color prompt words, the content prompt words, the reference hue range, and the preset theme generation information comprises: determining a target hue range based on the reference hue range and the color prompt words, the target hue range being included in the reference hue range; generating wallpaper prompt words based on the color prompt words, the content prompt words, and the preset wallpaper information; generating the wallpaper data using the theme generation model based on the wallpaper prompt words and the target hue range; generating icon prompt words based on the color prompt words, the content prompt words, and the preset icon information; generating the icon data using the theme generation model based on the icon prompt words and the target hue range.

4. The method of claim 3, wherein, The generation of the wallpaper data using the theme generation model based on the wallpaper prompt words and the target hue range comprises: generating the wallpaper data using the theme generation model based on the wallpaper prompt words, the target hue range, and a style feature vector, the style feature vector being determined based on a theme reference image; The generation of the icon data using the theme generation model based on the icon prompt words and the target hue range comprises: generating the icon data using the theme generation model based on the icon prompt words, the target hue range, and the style feature vector.

5. The method of claim 4, wherein, The generating the wallpaper data according to the wallpaper prompt word, the target color range, and the style feature vector by using the topic generation model comprises: The generating the wallpaper data according to the wallpaper prompt word, the target color range, the style feature vector, and a preset quality prompt word by using the topic generation model; The generating the icon data according to the icon prompt word, the target color range, and the style feature vector by using the topic generation model comprises: The generating the icon data according to the icon prompt word, the target color range, the style feature vector, and the preset quality prompt word by using the topic generation model.

6. The method of claim 4, wherein, The method further comprises: receiving the theme reference image sent by the vehicle; extracting the style feature vector from the theme reference image.

7. The method of claim 3, wherein, The theme resource data further comprises a theme preview image, and the method further comprises: generating the theme preview image according to the wallpaper data, the icon data, and display parameters of a vehicle infotainment system of the vehicle.

8. The method of claim 7, wherein, The theme resource data further comprises color value data, and the method further comprises: after generating the wallpaper data, performing cluster analysis on the wallpaper data to determine a theme color of the wallpaper data; comparing the theme color with a candidate color to determine a highlight color; determining a text color according to the brightness of the theme color; generating the color value data according to the theme color, the highlight color, and the text color; The generating the theme preview image according to the wallpaper data and the icon data comprises: generating the theme preview image according to the wallpaper data, the icon data, display parameters of the vehicle infotainment system, and the color value data.

9. The method of claim 1, wherein, The generating the prompt word according to the theme keyword comprises: performing color association on the theme keyword to obtain the color prompt word; performing expansion on the theme keyword to obtain the content prompt word.

10. The method of claim 1, wherein, The method further comprises: receiving a theme download instruction sent by the vehicle; in response to the theme download instruction, determining theme resource data matched by the theme download instruction; sending the theme resource data to the vehicle to enable the vehicle to update a theme of a vehicle infotainment system according to the theme resource data.

11. A theme generation apparatus characterized by comprising: The device configured on a server comprises: a receiving module configured to receive a theme generation instruction, an environment image of an environment in which a vehicle is located, and driving data of the vehicle, the environment image being used to represent a visual environment in which the vehicle is currently located; a multi-modal fusion module configured to perform semantic feature extraction on the theme generation instruction to obtain a semantic feature vector, perform visual feature extraction on the environment image to obtain a visual feature vector, and perform alignment and fusion processing on the semantic feature vector, the visual feature vector, and a data feature vector corresponding to the driving data to obtain a multi-modal fusion feature, the visual feature vector being used to represent environment information of the environment in which the vehicle is located, and different environment information corresponding to different saturation and brightness of a theme color; The parameter determination module is configured to determine a theme parameter based on the multi-modal fusion feature, the theme parameter comprising a reference color tone range, the reference color tone range comprising a value interval of an R, G and B channel of a theme to be generated. The prompt word generation module is configured to extract a theme keyword from the theme generation instruction, and generate a prompt word based on the theme keyword, the prompt word comprising a color prompt word and a content prompt word. The theme generation module is configured to generate theme resource data based on the color prompt word, the content prompt word, the reference color tone range and preset theme generation information by using a theme generation model.

12. An electronic device, comprising: Comprise: a memory, a processor; The memory stores computer execution instructions; The processor executes the computer execution instructions stored in the memory, so that the processor executes the method of any one of claims 1-10.

13. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the method of any one of claims 1-10.

Images