An on-site personalized invitation letter generation method and system based on artificial intelligence

Abstract

The application discloses a kind of on-site personalized invitation generation method and system based on artificial intelligence, comprising: collecting user image and associated personal information and event information in the activity site;User image is processed by artificial intelligence image;After segmentation, the portrait is adapted and fused with the preset brand invitation template, and the portrait position and size are adjusted according to the key points, and the user text information is laid out to avoid blocking the key area of the portrait;Style processing and holographic rendering are applied to the fused invitation, to generate personalized digital invitation content containing user image and text;Machine-readable code or near field communication chip is bound with physical medium through output device, so that users can retrieve digital invitation by scanning or sensing;For weak network environment, use placeholder two-dimensional code and post-mapping mechanism to ensure reliable delivery of content. Realize the automatic, personalized invitation production and distribution in the activity site, improve the interactive experience and brand promotion effect of participants.

Description

Technical Field

[0001] This invention belongs to the fields of artificial intelligence, image processing and computer vision technology, and in particular relates to a method and system for generating personalized invitations on-site based on artificial intelligence. Background Technology

[0002] In the fields of events, exhibitions, and marketing, invitations are a crucial medium for conveying event information and brand image. With the development of social media and digital technology, people have higher demands for the form and experience of invitations. Contemporary invitation design increasingly emphasizes personalization and interactive creativity, using various visual and interactive elements to highlight brand personality. For example, some companies are already using augmented reality (AR) technology to create invitations, allowing invitees to scan the physical invitation with their mobile phones to view virtual AR effects. This demonstrates that combining physical media with digital content has become a significant trend in enhancing the invitation experience.

[0003] On the other hand, taking photos on-site has become a common interactive activity at events. Traditional photo booths or photographers typically take photos of guests and print them as souvenirs. With the widespread use of smart devices and QR code technology, some event photography services have begun to incorporate QR codes into photo prints, allowing participants to scan and obtain digital copies of the photos or access online albums. This type of solution uses QR codes to link physical photos with online digital content, improving the convenience of photo sharing and dissemination.

[0004] However, existing technological solutions still have significant limitations. Current on-site photo and content distribution systems are mainly limited to providing photos and their digital copies, lacking further processing and personalized utilization of photo content, and unable to directly generate invitations containing participants' personal information and brand customization elements. Furthermore, the creation of existing invitations typically requires pre-design and mass production, making it difficult to customize them for each user on-site. While AR-based invitations enhance the interactive experience, they are mostly pre-designed, generic content that fails to fully reflect the individual characteristics of each invitee.

[0005] In real-world applications, especially at events with poor network conditions, the real-time generation and transmission of rich media content faces risks of delays and failures. Ensuring the timely and reliable delivery of AR-enhanced invitations to users is a major challenge in technology implementation. Existing technologies lack effective fault-tolerance mechanisms and cannot guarantee reliable content delivery under unstable network conditions. These shortcomings limit the interactive experience and brand promotion effectiveness at events, necessitating a new technological solution to address these issues. Summary of the Invention

[0006] To address the aforementioned technical issues, this invention proposes a method and system for generating personalized invitations for events based on artificial intelligence. This system enables automated and personalized invitation creation and distribution at event sites, enhancing the interactive experience for attendees and improving brand promotion.

[0007] To achieve the above objectives, this invention provides a method for generating personalized on-site invitations based on artificial intelligence, comprising:

[0008] AI-powered photography devices are used to collect user images and associated personal and event information at the event site.

[0009] Artificial intelligence image processing is performed on the user image, including image segmentation to extract the foreground of the human figure, key point detection to locate the feature points of the human figure, and depth estimation to generate a depth map;

[0010] The segmented portrait is adapted and integrated with the preset brand invitation template, and the position and size of the portrait are adjusted according to key points. At the same time, the user text information is arranged to avoid obscuring the key areas of the portrait.

[0011] The merged invitation is stylized and rendered holographically to generate personalized digital invitation content that includes user images and text.

[0012] Assign a unique content identifier and retrieval link to the content of the digital invitation, and encode the retrieval link into a machine-readable code or write it into a near-field communication chip;

[0013] The machine-readable code or near-field communication chip is bound to a physical medium via an output device, enabling users to retrieve digital invitations by scanning or sensing.

[0014] Optionally, the AI-powered photo-taking device can be used to collect user images and associated personal and event information at the event site, including: capturing user portrait images or video frames using a high-definition camera; recording user names and company information via a touch screen or automatic scanning device; and simultaneously associating basic event information, including event time, location, and RSVP link, and storing it for later use.

[0015] Optionally, AI image processing of user images includes: calling a pre-trained deep learning model to perform image segmentation to generate a portrait foreground image with alpha channels; using a pose estimation model to detect portrait key points, including the positions of facial features and limb skeleton points; and applying a monocular depth estimation model to infer the depth map of the portrait scene, providing depth data for subsequent rendering.

[0016] Optionally, the segmented portrait can be adapted and integrated with the preset brand invitation template, including: selecting a template layout that corresponds to the brand's visual style from the template library; adjusting the size, position, and angle of the portrait in the template based on the key information of the portrait; integrating the portrait foreground image into the template background, and achieving natural boundary blending through feathering edges or color adjustments.

[0017] Optionally, applying stylization and holographic rendering to the merged invitation includes: adjusting the tone and texture of the invitation image through a style transfer algorithm to match the brand's visual style; using a depth map to perform layered rendering of the invitation image to generate dynamic content with parallax effects; and combining mobile device sensor data to present a 3D stereoscopic effect that changes with the user's viewing angle on the user's terminal.

[0018] Optionally, assigning a unique content identifier and retrieval link to the digital invitation content includes: generating a unique content identifier (CID) for each invitation content; establishing a mapping relationship between the CID and the invitation content in the server database; generating a corresponding access link URL, and converting the link into a QR code or writing it into an NFC / AFC chip via an encoding module.

[0019] Optionally, binding a machine-readable code or near-field communication chip to a physical medium via an output device includes: printing an overlaid QR code on a user's finished photo or photo frame using a photo printer; or embedding an NFC / AFC chip containing a retrieval link into a physical card using a card-making machine; outputting the physical medium allows the user to retrieve the digital invitation through that medium.

[0020] Optionally, the method further includes a placeholder and post-mapping process in a weak network environment: when the network is poor, a placeholder QR code is first generated and printed out to a physical medium; the placeholder QR code corresponds to a temporary link, pointing to a prompt page in the content generation process; after the invitation content is generated, the temporary link is mapped and updated to the actual invitation content link.

[0021] On the other hand, to achieve the above objectives, the present invention also provides an artificial intelligence-based on-site personalized invitation generation system, comprising:

[0022] The image acquisition module is used to collect user images and associated personal and event information at the event site through an AI photography device;

[0023] The image processing module is used to perform artificial intelligence image processing on the user image, including image segmentation to extract the foreground of the portrait, key point detection to locate the feature points of the portrait, and depth estimation to generate a depth map;

[0024] The template fusion module is used to adapt and merge the segmented portrait with the preset brand invitation template, and adjust the position and size of the portrait according to key points, while arranging user text information to avoid obscuring key areas of the portrait;

[0025] The rendering module is used to apply stylization and holographic rendering to the merged invitation to generate personalized digital invitation content that includes user images and text.

[0026] The content identification module is used to assign a unique content identifier and retrieval link to the content of the digital invitation letter;

[0027] The encoding module is used to encode the retrieved link into machine-readable code or write it into the near-field communication chip;

[0028] An output module is used to bind the machine-readable code or near-field communication chip to a physical medium via an output device.

[0029] Technical Effects of this Invention: This invention discloses a method and system for generating personalized invitations for events based on artificial intelligence. It enables real-time generation and distribution of personalized invitations at events. By integrating user portraits with brand templates using AI technology, it creates a unique personalized experience, enhancing participants' sense of participation and pride. The invitations incorporate the brand's visual style, combining physical photos or cards with digital content to deepen brand impression and facilitate social sharing, creating a secondary dissemination effect. Seamless connection between physical media and online H5 pages allows users to directly RSVP and interact, improving usability and interactivity. For weak network environments, placeholder QR codes and post-mapping mechanisms ensure reliable content delivery and improve system robustness. Intelligent image synthesis algorithms automatically complete portrait segmentation, template matching, and stylization processing, guaranteeing the visual quality and layout of each invitation, achieving a combination of professional design and mass production. The modular system architecture supports multi-device collaboration, adapts to event scenarios of different scales, and has good scalability and compatibility. Attached Figure Description

[0030] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0031] Figure 1 This is a flowchart illustrating a method for generating personalized on-site invitations based on artificial intelligence, according to an embodiment of the present invention.

[0032] Figure 2 This is a schematic diagram of the structural framework of an artificial intelligence-based on-site personalized invitation generation system according to an embodiment of the present invention;

[0033] Figure 3This is a schematic diagram of the structure of the AI ​​photography device according to an embodiment of the present invention, wherein 1-external flash, 2-industrial PC or embedded computer, 3-SLR camera, 4-touch screen, and 5-thermal printer;

[0034] Figure 4 This is a schematic diagram illustrating the effect of a personalized invitation generated according to an embodiment of the present invention. Detailed Implementation

[0035] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0036] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0037] like Figure 1 As shown, this embodiment provides a method for generating personalized on-site invitations based on artificial intelligence, including:

[0038] AI-powered photography devices are used to collect user images and associated personal and event information at the event site.

[0039] Artificial intelligence image processing is performed on the user image, including image segmentation to extract the foreground of the human figure, key point detection to locate the feature points of the human figure, and depth estimation to generate a depth map;

[0040] The segmented portrait is adapted and integrated with the preset brand invitation template, and the position and size of the portrait are adjusted according to key points. At the same time, the user text information is arranged to avoid obscuring the key areas of the portrait.

[0041] The merged invitation is stylized and rendered holographically to generate personalized digital invitation content that includes user images and text.

[0042] Assign a unique content identifier and retrieval link to the content of the digital invitation, and encode the retrieval link into a machine-readable code or write it into a near-field communication chip;

[0043] The machine-readable code or near-field communication chip is bound to a physical medium via an output device, enabling users to retrieve digital invitations by scanning or sensing.

[0044] Furthermore, the AI-powered photo-taking device collects user images and associated personal and event information at the event site, including: capturing user portrait images or video frames using a high-definition camera; recording user names and company information via touchscreens or automatic scanning devices; and simultaneously associating basic event information, including event time, location, and RSVP link, and storing it for later use.

[0045] Specifically, the implementation process of this embodiment includes:

[0046] AI-powered photo-taking devices set up at the event venue capture users' portraits or video frames, recording their names, companies, and other personal information, as well as corresponding event information (time, location, RSVP link, etc.). When a guest stands in front of the AI ​​photo-taking device, the device's camera captures their image, and the system obtains the guest's name, company, and other information via a touchscreen or other input device (this information can also be automatically obtained by scanning name tags, QR codes, etc.). Simultaneously, the system stores the event's basic information (event name, time and location, RSVP webpage link, etc.) for later use.

[0047] Furthermore, the AI ​​image processing of user images includes: calling a pre-trained deep learning model to perform image segmentation to generate a portrait foreground image with alpha channels; using a pose estimation model to detect key points of the portrait, including the positions of facial features and limb skeleton points; and applying a monocular depth estimation model to infer the depth map of the portrait scene, providing depth data for subsequent rendering.

[0048] Specifically, the implementation process of this embodiment includes:

[0049] Artificial intelligence image processing is performed on the collected user images, including but not limited to: segmenting the user portrait from the background and obtaining the transparency channel map of the foreground portrait area; detecting key points of the portrait (such as the position of facial features, limb skeleton, or posture) for subsequent template matching and element avoidance; performing depth estimation on the monocular portrait image to generate a depth map of the portrait scene, providing a basis for subsequent parallax effect generation; after shooting, the AI ​​photo-taking device transmits the obtained photo data to the image processing module. Depending on the device performance and network conditions, the image processing module can be deployed on a local device or a cloud server. In this embodiment, it is assumed that the on-site network is normal and the photo is uploaded to the cloud server for processing. The processing steps include: calling a pre-trained deep learning model (such as U^2-Net, portrait segmentation model, etc.) to automatically segment the foreground of the portrait and obtain a PNG image of the alpha channel. The background part can be discarded or blurred for later use. Calling a pose estimation model (such as OpenPose, MediaPipe, etc.) to obtain the skeletal key points and facial feature points of the portrait. These key points are used to determine the person's posture and facial orientation, thereby selecting the placement of the portrait in the invitation template. For example, if a person is detected standing upright, a full-body template might be used; if only the upper body is detected, a half-body layout would be used. Keypoint information can also be used to avoid placing text on the person's face in subsequent steps. A monocular depth estimation model (such as MiDaS) is used to infer a depth map from the single captured image of a person. The depth map represents the distance of each pixel in the image in grayscale. Since the background has already been segmented, the relative depth between the person and the background can be estimated by combining the background blur level. This step lays the foundation for generating AR parallax effects later.

[0050] Furthermore, adapting and integrating the segmented portrait with the preset brand invitation template includes: selecting a template layout that corresponds to the brand's visual style from the template library; adjusting the size, position, and angle of the portrait in the template based on the key information of the portrait; integrating the foreground image of the portrait into the template background, and achieving natural boundary blending through feathering edges or color adjustments.

[0051] Specifically, the implementation process of this embodiment includes:

[0052] The segmented portrait is adapted to a pre-set brand invitation template, including adjusting its size, position, and angle according to key points and layout requirements to ensure it blends harmoniously with the template background. A template layout corresponding to the event and brand image is selected from the stored invitation template library. The template predefines the position and size of the background pattern area, the portrait area, and the text area. For example, a brand's template might reserve space for the portrait on the left and the event title text on the right. Guided by the key point information, the size and position of the cropped portrait are adjusted so that the portrait or full-body image fits precisely within the reserved area of ​​the template. If necessary, the portrait can be rotated or mirrored for better composition. Finally, the foreground image of the portrait is pasted and blended onto the template background, using feathering edges or color adjustments to naturally blend the boundaries.

[0053] Furthermore, the stylized processing and holographic rendering of the merged invitation include: adjusting the tone and texture of the invitation image through a style transfer algorithm to match the brand's visual style; using a depth map to perform layered rendering of the invitation image to generate dynamic content with parallax effects; and combining mobile device sensor data to present a 3D stereoscopic effect that changes with the user's viewing angle on the user's terminal.

[0054] Specifically, the implementation process of this embodiment includes:

[0055] The brand's visual style is applied to the portrait and overall invitation, using style transfer or filters to ensure the invitation's visual effect aligns with the brand's tone. This includes adjusting color tones and adding brand-specific textures. The depth map is then used to render a holographic stereoscopic effect on the synthesized invitation image, generating images or animations with parallax effects, which can be displayed as 3D stereoscopic or AR effects on mobile devices.

[0056] Based on the obtained user name, company, and other information, the algorithm fills the corresponding text content into the text boxes in the template. To prevent text from overlapping with the person's image, the algorithm checks for collisions between the text area and the foreground of the person. If text is detected that will obscure the face or important parts, the algorithm can automatically adjust the position of the text box (e.g., move it up or down) or change the font size to fit the blank area. Simultaneously, the system retrieves the event name, date, location, etc., from the event information and fills them into the corresponding positions on the invitation. The RSVP link is usually presented as a QR code or displayed as a short link below the card.

[0057] The initially synthesized invitation image is input into a style transfer algorithm to apply the brand's unique visual style. For example, a neural network trained on the brand's promotional materials can be used to filter the image, making the tone and texture match the brand's style; or the overall color balance of the image can be simply adjusted to match the brand's main color scheme. The brand logo, watermark, and other elements can also be overlaid on appropriate locations on the invitation to enhance brand recognition.

[0058] Using previously generated depth maps of the characters and background, the invitation image is rendered in layers to create dynamic content with a parallax effect. Specifically, the invitation can be divided into a foreground layer (characters), a midground layer (partial UI decorations or text), and a background layer, with different offset coefficients applied based on depth. When a user views the invitation H5 page on their mobile phone, combined with the phone's gyroscope sensor data, each layer undergoes subtle displacement with the viewing angle, creating a 3D, holographic, floating effect. If dynamic effects are not required, a static 3D poster incorporating depth and lighting effects can also be generated directly.

[0059] Furthermore, assigning a unique content identifier and retrieval link to the digital invitation content includes: generating a unique content identifier (CID) for each invitation content; establishing a mapping relationship between the CID and the invitation content in the server database; generating a corresponding access link URL, and converting the link into a QR code or writing it into an NFC / AFC chip through an encoding module.

[0060] Specifically, the implementation process of this embodiment includes:

[0061] Each generated invitation is assigned a unique Content Identifier (CID) and a corresponding retrieval link URL. This retrieval link is then converted into a QR code image via an encoding module or written into an NFC / AFC chip. This establishes a unique binding between the invitation content and a physical carrier identifier (a QR code or an NFC chip).

[0062] Furthermore, binding machine-readable codes or near-field communication chips to physical media via output devices includes: printing overlaid QR codes on user photos or photo frames using a photo printer; or embedding NFC / AFC chips containing retrieval links into physical cards using a card-making machine; outputting physical media allows users to retrieve digital invitations through that medium.

[0063] Specifically, the implementation process of this embodiment includes:

[0064] The aforementioned QR code or chip is linked to the user's physical photo / card via an on-site output device: when using a photo, the QR code is printed and overlaid on the user's photo or photo frame; or an invitation card / ticket with a QR code is printed. When using NFC / AFC media, a chip containing the invitation link is embedded or affixed to the physical card. In this way, each user receives a physical medium (photo or card) containing a QR code or sensor chip for retrieving their digital invitation.

[0065] Furthermore, the method also includes a placeholder and post-mapping process in a weak network environment: when the network is poor, a placeholder QR code is first generated and printed out to a physical medium; the placeholder QR code corresponds to a temporary link, pointing to a prompt page in the content generation process; when the invitation content is generated, the temporary link is mapped and updated to the actual invitation content link.

[0066] Specifically, the implementation process of this embodiment includes:

[0067] Users can quickly retrieve the digital invitation content and open the corresponding H5 / RSVP page on their phone by scanning the QR code on their photo / card or by bringing an NFC-enabled phone close to the chip. This page displays the invitation's detailed content, allowing users to further browse event information, view dynamic AR effects within the invitation, and reply online (RSVP) or share it on social media. The AR parallax holographic effect in the invitation utilizes the phone's gyroscope / accelerometer and other sensors to achieve a displacement effect between the portrait and background that changes with the viewing angle, enhancing the immersive three-dimensional experience.

[0068] To address potential delays in invitation content uploads or user access due to poor network conditions, the system employs a placeholder mechanism. Specifically, after basic information collection, a placeholder QR code is immediately generated and printed / written onto a physical medium for the user to temporarily store. This placeholder QR code corresponds to a temporary link or ID, initially pointing to a "Content Generating" notification page. Once the backend completes the full invitation generation and upload, this temporary ID is updated to a link to the actual invitation content, allowing the user to scan the same QR code again later to view the final invitation content. This post-mapping mechanism ensures that users can still obtain the physical QR code promptly even under weak network conditions and retrieve the invitation after content generation, thus improving system robustness.

[0069] After completing the above steps, the system obtains a complete digital invitation, including a pre-formatted invitation image (or animation) file, as well as structured text and link data (event introduction, RSVP button link, etc.), which will be packaged together into an H5 page or mini-program.

[0070] like Figure 1The process is as follows: the system creates a unique Content ID (CID) for each generated invitation and stores the mapping relationship between this ID and the invitation content in the server database. Then, the system generates a corresponding access link URL, allowing the user's terminal to access the invitation. Next, the link is converted into a QR code image using a QR code generator; alternatively, the link is written into a chip tag with NFC functionality. In this embodiment, a QR code is used as an example: the system prints the generated QR code onto a small strip of photo paper using the printer built into the camera device. Simultaneously, the device prints the user's photo as a commemorative photo and attaches the QR code strip below the photo, giving the photo a QR code label corresponding to the invitation / photo. Figure 4 As shown. The entire printing and affixing process can be completed automatically by the machine or by on-site staff with simple operation. Ultimately, the user receives a personalized photo card with a QR code.

[0071] After the user receives the printed photo card, such as Figure 4 The diagram shows the interaction flow and system module principle of a user obtaining augmented reality (AR) experience through a mobile terminal in an embodiment of the present invention.

[0072] The system primarily involves user interaction with a smartphone (as the user terminal), a physical photograph, and a QR code label attached to the photograph. Its interaction and processing logic is as follows:

[0073] Image acquisition and recognition: This step is performed by an AI-powered camera (such as...). Figure 3 (As shown) This was performed at the event site. After the device's high-definition camera captured the user's image, the raw image data was transmitted to the core image processing module / processor.

[0074] Modular processing: The image processing module / processor acts as the system's control center, coordinating with the various functional modules on the right to generate and bind personalized content.

[0075] First, the image recognition module is called to perform feature analysis and recognition on the collected user photos to extract key information about the human image;

[0076] Next, the 3D rendering engine is invoked to adapt the corresponding elements and effects based on the human features;

[0077] The key is that the processor performs a "content binding" operation: it maps the processed user photo material, the generated 3D rendered elements, and the unique QR code tag (or ID) assigned to that user, storing this entire set of personalized data in the virtual content database. This step ensures that each QR code acts as a "unique key," pointing only to the digital content of that specific user. Rendering and Content Retrieval: After determining the content index, the processor invokes the 3D rendering engine. This engine has a bidirectional connection to the virtual content database. The virtual content database stores pre-generated model, animation, or holographic image data bound to specific photo IDs. The 3D rendering engine extracts the corresponding virtual materials from the database and performs real-time rendering calculations using the spatial coordinate information provided by the image recognition module.

[0078] AR overlay display: The rendered virtual image is fed back to the user's smartphone via a specific user experience link (or real-time rendering data stream). Ultimately, the AR overlay display is achieved on the smartphone screen, allowing the user to see virtual 3D objects (such as...). Figure 4 The AR overlay shown is precisely superimposed on the physical photo, achieving a visual experience that blends the virtual and the real.

[0079] Users can also share the e-invitation page with friends or on social media to further expand the event's reach. This personalized invitation, due to its uniqueness, often attracts more attention and discussion. Meanwhile, event organizers can use the system to obtain real-time RSVP feedback data and track how many people have viewed or shared the invitation, facilitating follow-up and adjustments to promotional strategies.

[0080] This embodiment also provides an artificial intelligence-based on-site personalized invitation generation system, including:

[0081] The image acquisition module is used to collect user images and associated personal and event information at the event site through an AI photography device;

[0082] The image processing module is used to perform artificial intelligence image processing on the user image, including image segmentation to extract the foreground of the portrait, key point detection to locate the feature points of the portrait, and depth estimation to generate a depth map;

[0083] The template fusion module is used to adapt and merge the segmented portrait with the preset brand invitation template, and adjust the position and size of the portrait according to key points, while arranging user text information to avoid obscuring key areas of the portrait;

[0084] The rendering module is used to apply stylization and holographic rendering to the merged invitation to generate personalized digital invitation content that includes user images and text.

[0085] The content identification module is used to assign a unique content identifier and retrieval link to the content of the digital invitation letter;

[0086] The encoding module is used to encode the retrieved link into machine-readable code or write it into the near-field communication chip;

[0087] An output module is used to bind the machine-readable code or near-field communication chip to a physical medium via an output device.

[0088] In the case of placeholder mechanisms in weak network environments:

[0089] In events with poor network conditions, the system of this invention can enable placeholder QR codes and post-mapping mechanisms to ensure that users can obtain physical QR codes in a timely manner and ultimately access the content.

[0090] like Figure 2 As shown, the system includes a placeholder link generation unit. When the system detects that the network upload of the invitation content is slow (e.g., exceeding a preset threshold time), it does not wait for the content upload to complete before generating the QR code. Instead, it immediately generates a URL and QR code pointing to a temporary placeholder page, as follows:

[0091] After completing the basic information collection and photo upload, if the subsequent AI processing takes a long time (e.g., tens of seconds), the system will assign the user a temporary CID and a corresponding QR code, and immediately print out the QR code for the user to claim. The page linked by the QR code will only display a message such as "Your exclusive invitation is being generated, please try again later" or a brief introduction to the event, and will not be blank.

[0092] Once the backend completes the generation of the user's invitation and successfully uploads the content to the server, it updates the link mapping on the server, redirecting the previously assigned temporary CID to the actual invitation content page. At this point, the placeholder QR code transforms into an entry point to the real content.

[0093] When users scan an earlier QR code, if they see a placeholder page before the invitation is fully generated, they can refresh the page later. Once the content is ready, scanning the code again or refreshing the page will display the complete invitation. Since the physical QR code held by the user remains unchanged, the system's mapping update is transparent to the user, eliminating the need for them to change their QR code and ensuring a smooth and continuous experience.

[0094] This placeholder-mapping mechanism ensures that even with poor network conditions, each user can immediately receive a physical QR code as an invitation after taking a photo, preventing disruption to the on-site process while waiting for content generation. Once network conditions allow for content transmission, users will ultimately receive the complete invitation, achieving a balance between efficiency and user experience.

[0095] System composition and hardware implementation:

[0096] like Figure 2 and Figure 3 As shown, the hardware and software of this invention can be implemented in various ways. The following describes a typical implementation architecture:

[0097] The AI ​​photography device (front-end terminal) consists of a DSLR camera 3, an external flash 1, an industrial PC or embedded computer 2, a touchscreen display 4, a thermal printer 5, and an NFC card writer. The DSLR camera 3 is responsible for imaging. The industrial PC or embedded computer 2 is pre-installed with the software of this invention to control the shooting process and perform simple local processing (e.g., segmentation and key point detection models can be pre-deployed locally to reduce the amount of data uploaded). The touchscreen display 4 is used to show users previews of their photos, AR effects, or UI prompts (such as entering a name, waiting progress, etc.). The thermal printer 5 is used to instantly output photos and QR codes, and the small NFC card writer is used to write links to NFC stickers or IC cards. If the device is powerful enough, all image processing and compositing can be performed locally, making cloud processing optional, thus allowing for completely independent operation. The device communicates with the backend server via wired / wireless network and can store data for batch uploading after the connection is restored, even in weak network conditions.

[0098] Backend Server: A high-performance computing server cluster running image processing and invitation page generation services. It receives photos and user information uploaded from front-end terminals, requests AI algorithm service modules to complete segmentation, depth estimation, etc., and calls the template engine to synthesize the results into the final invitation image and page. The server also handles database functions, storing CID mappings for each invitation content, user RSVP replies, and other data. For scenarios with high AR effects requirements, the server can pre-generate some 3D effects resources for end users to load. When generating H5 pages, the server packages images and scripts from different layers to support AR parallax rendering on the terminal. The entire backend system can be deployed in the cloud to support large-scale concurrency, or it can be built locally to enhance data security and real-time performance.

[0099] User terminal: Typically, this is the guest's own smartphone. No special app is needed; the invitation H5 page can be accessed by scanning a QR code using a built-in browser like WeChat. The script running on the page can utilize the phone's sensors to achieve parallax effects, or prompt the user to open the camera to view the physical card, thus triggering AR animation (if image recognition is used to trigger AR). The user terminal sends RSVP feedback or social sharing requests to the server, which records and processes them.

[0100] In practical applications, the invention can be expanded as needed. For example, a physical badge or pendant can be generated for each guest, embedding an NFC chip for the invitation. This serves not only as an entry ID card but also as an invitation; users can simply tap their phones to the badge to view invitation details and check in. This demonstrates the flexibility and compatibility of the invention.

[0101] This invention discloses an AI-based method and system for generating personalized invitations for events. It enables real-time generation and distribution of personalized invitations, integrating user portraits with brand templates using AI technology to create a unique personalized experience and enhance participants' sense of participation and pride. The invitations incorporate the brand's visual style, combining physical photos or cards with digital content to deepen brand impression and facilitate social sharing, creating a secondary dissemination effect. Seamless integration of physical media and online H5 pages allows users to directly RSVP and interact, improving usability and interactivity. For weak network environments, placeholder QR codes and post-mapping mechanisms ensure reliable content delivery and improve system robustness. Intelligent image synthesis algorithms automatically perform portrait segmentation, template matching, and stylization, guaranteeing the visual quality and layout of each invitation, combining professional design with mass production. The modular system architecture supports multi-device collaboration, adapts to event scenarios of different scales, and has good scalability and compatibility.

[0102] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for generating personalized on-site invitations based on artificial intelligence, characterized in that, include: AI-powered photography devices are used to collect user images and associated personal and event information at the event site. Artificial intelligence image processing is performed on the user image, including image segmentation to extract the foreground of the human figure, key point detection to locate the feature points of the human figure, and depth estimation to generate a depth map; The segmented portrait is adapted and integrated with the preset brand invitation template, and the position and size of the portrait are adjusted according to key points. At the same time, the user text information is arranged to avoid obscuring the key areas of the portrait. The merged invitation is stylized and rendered holographically to generate personalized digital invitation content that includes user images and text. Assign a unique content identifier and retrieval link to the content of the digital invitation, and encode the retrieval link into a machine-readable code or write it into a near-field communication chip; The machine-readable code or near-field communication chip is bound to a physical medium via an output device, so that the user can retrieve the digital invitation by scanning or sensing. The stylization and holographic rendering of the merged invitation include: adjusting the tone and texture of the invitation image through a style transfer algorithm to match the brand's visual style; using a depth map to render the invitation image in layers to generate dynamic content with parallax effects; and combining mobile device sensor data to present a 3D stereoscopic effect that changes with the user's viewing angle on the user's terminal. The method also includes a placeholder and post-mapping process in a weak network environment: when the network is poor, a placeholder QR code is first generated and printed out to a physical medium; the placeholder QR code corresponds to a temporary link, pointing to a prompt page in the content generation process; when the invitation content is generated, the temporary link is mapped and updated to the actual invitation content link.

2. The method for generating personalized on-site invitations based on artificial intelligence as described in claim 1, characterized in that, The AI-powered photo-taking device collects user images and associated personal and event information at the event site, including: capturing user portrait images or video frames using a high-definition camera; recording user names and company information via touch screens or automatic scanning devices; and simultaneously associating basic event information, including event time, location, and RSVP link, and storing it for later use.

3. The method for generating personalized on-site invitations based on artificial intelligence as described in claim 1, characterized in that, Artificial intelligence image processing of user images includes: calling a pre-trained deep learning model to perform image segmentation to generate a portrait foreground image with alpha channels; using a pose estimation model to detect key points of the portrait, including the positions of facial features and limb skeleton points; and applying a monocular depth estimation model to infer the depth map of the portrait scene, providing depth data for subsequent rendering.

4. The method for generating personalized on-site invitations based on artificial intelligence as described in claim 1, characterized in that, Adapting and integrating the segmented portrait with the preset brand invitation template includes: selecting a template layout that matches the brand's visual style from the template library; adjusting the size, position, and angle of the portrait in the template based on the portrait's key information; blending the portrait's foreground image into the template background, and achieving natural boundary blending through feathering edges or color adjustments.

5. The method for generating personalized on-site invitations based on artificial intelligence as described in claim 1, characterized in that, Assigning a unique content identifier and retrieval link to the digital invitation content includes: generating a unique content identifier (CID) for each invitation content; establishing a mapping relationship between the CID and the invitation content in the server database; generating a corresponding access link URL, and converting the link into a QR code or writing it into an NFC / AFC chip through an encoding module.

6. The method for generating personalized on-site invitations based on artificial intelligence as described in claim 1, characterized in that, Binding machine-readable codes or near-field communication chips to physical media via output devices includes: printing overlaid QR codes on user photos or frames using a photo printer; or embedding NFC / AFC chips containing retrieval links into physical cards using a card-making machine; outputting physical media that allows users to retrieve digital invitations.

7. A personalized on-site invitation generation system based on artificial intelligence, characterized in that, For implementing the AI-based on-site personalized invitation generation method as described in any one of claims 1-6, the system comprises: The image acquisition module is used to collect user images and associated personal and event information at the event site through an AI photography device; The image processing module is used to perform artificial intelligence image processing on the user image, including image segmentation to extract the foreground of the portrait, key point detection to locate the feature points of the portrait, and depth estimation to generate a depth map; The template fusion module is used to adapt and merge the segmented portrait with the preset brand invitation template, and adjust the position and size of the portrait according to key points, while arranging user text information to avoid obscuring key areas of the portrait; The rendering module is used to apply stylization and holographic rendering to the merged invitation to generate personalized digital invitation content that includes user images and text. The content identification module is used to assign a unique content identifier and retrieval link to the content of the digital invitation letter; The encoding module is used to encode the retrieved link into machine-readable code or write it into the near-field communication chip; An output module is used to bind the machine-readable code or near-field communication chip to a physical medium via an output device; The stylization and holographic rendering of the merged invitation include: adjusting the tone and texture of the invitation image through a style transfer algorithm to match the brand's visual style; using a depth map to render the invitation image in layers to generate dynamic content with parallax effects; and combining mobile device sensor data to present a 3D stereoscopic effect that changes with the user's viewing angle on the user's terminal. When the network is poor, a placeholder QR code is first generated and printed out to physical media; the placeholder QR code corresponds to a temporary link that points to a prompt page during content generation; once the invitation content is generated, the temporary link is updated to the actual invitation content link.

Images