Artificial intelligence-based image processing methods, devices, computer equipment, and media

Abstract

This application belongs to the field of artificial intelligence technology and relates to an AI-based image processing method, apparatus, computer device, and storage medium. The method includes: acquiring input text description and a reference image; extracting features from the text description and reference image using a hierarchical encoder to obtain a text embedding corresponding to the text description and a pose embedding corresponding to the reference image; performing semantic alignment processing on the text embedding and pose embedding using a hierarchical alignment module to obtain alignment features; performing semantic optimization processing on the alignment features using a cross-modal adapter to obtain target features; performing image generation processing corresponding to the target features using a pose condition generator to obtain a target image; and outputting the target image. Furthermore, the target image of this application can be stored in a blockchain. This application can be applied to text-to-image scenarios in the financial and medical fields, achieving precise pose control from text to image and improving the quality of the generated target image.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and can be applied to fields such as fintech and digital healthcare, particularly to image processing methods, devices, computer equipment, and storage media based on artificial intelligence. Background Technology

[0002] In traditional text-to-image generation techniques, the content of the generated image mainly relies on the semantic information of the text description, but the ability to precisely control specific attributes (such as human pose) is limited. Existing methods typically employ unsupervised or weakly supervised learning strategies, such as inferring pose information from unlabeled images through self-supervised learning. However, these methods lack the ability to model the pose-text relationship at a fine-grained level, making it difficult to achieve precise pose control. Specifically, traditional methods usually treat pose information as implicit features and generate images through global semantic matching. This results in insufficient matching between the human pose and the text description in the generated image, as well as problems such as pose blurring and structural distortion. This coarse-grained generation approach is difficult to meet the needs of high-precision applications, and the quality of the generated images is low.

[0003] For example, in remote insurance claims in the financial sector, customers might describe their accidental injury via text (e.g., "I fell and braced myself with my right hand") and expect an image to be generated to aid in damage assessment. However, traditional text-to-image generation methods cannot accurately control the human posture in the image (e.g., the joint angle when bracing oneself with the right hand, the direction of body tilt), resulting in generated images that do not match the actual injury posture, affecting the efficiency and accuracy of claims processing. This technological deficiency not only increases the cost of manual verification but may also lead to claims disputes.

[0004] Therefore, there is an urgent need to provide a novel text-to-image generation method to achieve fine-grained control over specific attributes such as human posture, improve the quality of generated images, and meet the needs of related fields for accurate image generation. Summary of the Invention

[0005] The purpose of this application is to propose an image processing method, apparatus, computer device, and storage medium based on artificial intelligence, so as to solve the technical problem that existing text-to-image generation technologies are difficult to achieve precise posture control and have low image quality.

[0006] Firstly, an image processing method based on artificial intelligence is provided, including:

[0007] Obtain the input text description and reference image;

[0008] Based on a preset hierarchical encoder, features are extracted from the text description and the reference image to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image.

[0009] Based on a preset hierarchical alignment module, the text embedding and the pose embedding are semantically aligned to obtain the corresponding alignment features.

[0010] The alignment features are semantically optimized based on a preset cross-modal adapter to obtain the corresponding target features;

[0011] Based on a preset pose condition generator, image generation processing corresponding to the target features is performed to obtain the corresponding target image;

[0012] The target image is then processed for output.

[0013] Secondly, an image processing device based on artificial intelligence is provided, comprising:

[0014] The acquisition module is used to acquire the input text description and reference image;

[0015] The extraction module is used to extract features from the text description and the reference image based on a preset hierarchical encoder to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image.

[0016] An alignment module is used to perform semantic alignment processing on the text embedding and the pose embedding based on a preset hierarchical alignment module to obtain corresponding alignment features;

[0017] The optimization module is used to perform semantic optimization processing on the alignment features based on a preset cross-modal adapter to obtain the corresponding target features;

[0018] The generation module is used to perform image generation processing corresponding to the target features based on a preset pose condition generator to obtain the corresponding target image;

[0019] The output module is used to perform output processing on the target image.

[0020] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described artificial intelligence-based image processing method.

[0021] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described artificial intelligence-based image processing method.

[0022] In the aforementioned scheme implemented by the AI-based image processing method, apparatus, computer device, and storage medium, the input text description and reference image are first acquired; then, feature extraction is performed on the text description and reference image based on a preset hierarchical encoder to obtain a text embedding corresponding to the text description and a pose embedding corresponding to the reference image; subsequently, semantic alignment processing is performed on the text embedding and pose embedding based on a preset hierarchical alignment module to obtain corresponding alignment features; subsequently, semantic optimization processing is performed on the alignment features based on a preset cross-modal adapter to obtain corresponding target features; further, image generation processing corresponding to the target features is performed based on a preset pose condition generator to obtain the corresponding target image; finally, the target image is output. This application extracts features from the input text description and reference image using a hierarchical encoder to obtain corresponding text embeddings and pose embeddings; then, semantic alignment processing is performed on the text embeddings and pose embeddings using a hierarchical alignment module to obtain alignment features; then, semantic optimization processing is performed on the alignment features using a cross-modal adapter to obtain target features; and finally, image generation processing corresponding to the target features is performed using a pose condition generator to obtain and output the target image. This application processes the input text description and reference image by combining a hierarchical encoder, a hierarchical alignment module, a cross-modal adapter, and a pose condition generator. This enables precise pose control from text to image, thereby efficiently and accurately generating target images that meet pose conditions and improving the quality of the generated target images. Attached Figure Description

[0023] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0025] Figure 2 This is a flowchart of an embodiment of the artificial intelligence-based image processing method according to this application;

[0026] Figure 3 This is a schematic diagram of a structure of an embodiment of an artificial intelligence-based image processing apparatus according to this application;

[0027] Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0029] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0030] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0031] like Figure 1 As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Terminal device 101 may be a laptop 1011, tablet 1012, or mobile phone 1013. Network 102 is used as a medium to provide a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0032] Users can use terminal device 101 to interact with server 103 via network 102 to receive or send messages, etc. Various communication client applications can be installed on terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0033] Terminal device 101 can be various electronic devices with a display screen and support web browsing. In addition to laptops 1011, tablets 1012, or mobile phones 1013, terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer, and a desktop computer, etc.

[0034] Server 103 can be a server that provides various services, such as a backend server that provides support for the pages displayed on terminal device 101.

[0035] It should be noted that the AI-based image processing method provided in this application embodiment is generally executed by a server / terminal device, and correspondingly, the AI-based image processing device is generally located in the server / terminal device.

[0036] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0037] Continue to refer to Figure 2 This document illustrates a flowchart of an embodiment of the AI-based image processing method according to this application. The order of steps in the flowchart can be changed, and some steps can be omitted, depending on different requirements. The AI-based image processing method provided in this application can be applied to any scenario requiring image generation, and thus can be applied to products in these scenarios, such as text-to-image (TPI) scenarios in the financial and medical fields. The AI-based image processing method includes the following steps:

[0038] Step S201: Obtain the input text description and reference image.

[0039] In this embodiment, the artificial intelligence-based image processing method runs on an electronic device (e.g., Figure 1The server / terminal device shown can acquire the input text description and reference image via wired or wireless connection. It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-wideband) connections, and other currently known or future-developed wireless connection methods. The executing entity of this application is an image generation system, which can be simply referred to as the system. The aforementioned text description can be the text description information matching the desired generated image input according to the user's actual needs. The aforementioned reference image is an image providing initial pose information, which can be used to extract the initial pose.

[0040] The system described in this application comprises four main modules: a hierarchical encoder, a hierarchical alignment module, a cross-modal adapter, and a pose condition generator. The system takes natural language text and a reference image (used to extract the initial pose) as input and outputs a generated image that conforms to the text description and has a controllable pose. Specifically, the hierarchical encoder first processes the text input using a pre-trained language model, while simultaneously extracting 2D keypoints from the reference image using an open-source pose estimator. The data from these two modalities, after feature extraction, are fed into the hierarchical alignment module. The cross-modal adapter receives the output from the hierarchical alignment module and optimizes the feature space through contrastive learning. Finally, the pose condition generator (based on a diffusion model or GAN architecture) uses the adapted multimodal features as conditional input to progressively synthesize the target image.

[0041] Furthermore, this application can be applied to business scenarios of text-based graphics in fields such as finance and insurance, and healthcare. For example, in the finance and insurance field, the content described above may include: using gesture recognition technology to analyze a customer's hand movements (such as pen grip angle, signature trajectory stability) and body posture (such as sitting uprightness, head tilt angle) when signing electronic contracts to assess whether their signing behavior is abnormal or risky (such as forced signing, lack of concentration), thereby assisting in insurance fraud detection or contract validity verification. Explanation: Gesture recognition is used to analyze a customer's body and hand movements when signing a contract. Objective: To detect abnormal behavior (such as forced signing) and assist in insurance fraud detection or contract validity verification.

[0042] In the medical field, the content described above could include: utilizing posture recognition technology to monitor the movements and postures of elderly patients during home rehabilitation training in real time (such as joint flexion angles, body balance, and accuracy of movement repetition), and automatically generating personalized rehabilitation suggestions (such as adjusting movement range and increasing training frequency) by comparing them with a standard rehabilitation movement library, in order to improve rehabilitation outcomes and reduce the risk of secondary injury. Explanation: Posture recognition is used to monitor patients' rehabilitation movements (such as joint angles and balance). Objective: To improve rehabilitation outcomes by generating personalized rehabilitation suggestions through comparison with a standard movement library.

[0043] Step S202: Based on a preset hierarchical encoder, feature extraction is performed on the text description and the reference image to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image.

[0044] In this embodiment, the specific implementation process of extracting features from the text description and the reference image based on the preset hierarchical encoder to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0045] Step S203: Based on the preset hierarchical alignment module, perform semantic alignment processing on the text embedding and the pose embedding to obtain the corresponding alignment features.

[0046] In this embodiment, the specific implementation process of performing semantic alignment processing on the text embedding and the pose embedding based on the preset hierarchical alignment module to obtain the corresponding alignment features will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0047] Step S204: Perform semantic optimization processing on the alignment features based on the preset cross-modal adapter to obtain the corresponding target features.

[0048] In this embodiment, the specific implementation process of performing semantic optimization processing on the alignment features based on the preset cross-modal adapter to obtain the corresponding target features will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0049] Step S205: Based on the preset pose condition generator, perform image generation processing corresponding to the target features to obtain the corresponding target image.

[0050] In this embodiment, the specific implementation process of the above-mentioned image generation process based on the preset pose condition generator to obtain the corresponding target image is described in more detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0051] Step S206: Output processing is performed on the target image.

[0052] In this embodiment, the specific implementation process of outputting the target image described above will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0053] This application first acquires input text description and reference image; then, based on a preset hierarchical encoder, it extracts features from the text description and reference image to obtain text embeddings corresponding to the text description and pose embeddings corresponding to the reference image; subsequently, based on a preset hierarchical alignment module, it performs semantic alignment processing on the text embeddings and pose embeddings to obtain corresponding alignment features; subsequently, based on a preset cross-modal adapter, it performs semantic optimization processing on the alignment features to obtain corresponding target features; further, based on a preset pose condition generator, it performs image generation processing corresponding to the target features to obtain the corresponding target image; finally, it outputs the target image. This application extracts features from input text description and reference image using a hierarchical encoder to obtain corresponding text embeddings and pose embeddings; then, based on a hierarchical alignment module, it performs semantic alignment processing on the text embeddings and pose embeddings to obtain alignment features; then, based on a cross-modal adapter, it performs semantic optimization processing on the alignment features to obtain target features; and finally, based on a pose condition generator, it performs image generation processing corresponding to the target features to obtain the target image and output it. This application processes the input text description and reference image by combining a hierarchical encoder, a hierarchical alignment module, a cross-modal adapter, and a pose condition generator. This enables precise pose control from text to image, thereby efficiently and accurately generating target images that meet pose conditions and improving the quality of the generated target images.

[0054] In some alternative implementations, step S203 includes the following steps:

[0055] The hierarchical alignment module performs global alignment processing on the text embedding and the pose embedding to obtain the corresponding global features.

[0056] In this embodiment, the hierarchical alignment module includes a global alignment layer, the goal of which is to capture the overall semantic correspondence between text and pose. Global alignment processing of the text embedding and pose embedding can be performed based on the use of the global alignment layer. The specific implementation process includes:

[0057] Input: Text embedding E t Posture embedding E p .

[0058] Operation: 1. Flattening operation: Flatten E t and E p Flattened into a one-dimensional vector:

[0059] 2. Features after splicing and flattening: Splice the flattened text headers and posture features together:

[0060] E falt =[Flatten(E t Flatten (E) p )]

[0061] 3. Capture global semantics through the Transformer encoder:

[0062] E global =Transformer([Flatten(E t Flatten (E) p )])

[0063] Among them, E global These are global alignment features (abbreviated as global features), used to capture the overall meaning of an action, such as "dancing" or "running".

[0064] The text embedding and the pose embedding are aligned at the part level to obtain the corresponding part-level features.

[0065] In this embodiment, the implementation process of part-level alignment includes:

[0066] Input: Text embedding E t Posture embedding E p .

[0067] Operation: 1. Divide the human body into K predefined regions (e.g., upper body, left arm, etc.). For the kth region, construct a binary mask M. k ∈{0,1}^{n×m}, used to identify related text words and joints:

[0068] 2. Establishing local correspondences through attention mechanisms:

[0069]

[0070] Where Attention is scaled dot product attention, used to calculate text words. and The relevance. i represents a joint, j represents a text word. E k These are location-level features used to establish a correspondence between actions such as "waving the right hand" and the right arm joint.

[0071] The text embedding and the pose embedding are aligned at the joint level to obtain the corresponding joint-level features.

[0072] In this embodiment, the implementation process of joint-level alignment includes:

[0073] Input: Text embedding E t Posture embedding E p .

[0074] Operation: 1. Through a learnable correlation matrix Achieving fine-grained matching:

[0075]

[0076] Where σ is the sigmoid function, W a A is a learnable parameter. ij This indicates the relevance between joint i and text word j.

[0077] 2. Calculate joint-word pair features:

[0078]

[0079] Here, || represents vector concatenation, used to fuse fine-grained features of joints and text.

[0080] The global features, the location-level features, and the joint-level features are integrated to obtain the corresponding integrated features.

[0081] In this embodiment, the integrated feature is a feature set that includes the global feature, the location-level feature, and the joint-level feature.

[0082] The integration feature is used as the alignment feature.

[0083] In this embodiment, the alignment features obtained based on hierarchical alignment processing are used as input for subsequent cross-modal optimization, enabling contrastive learning to optimize the embedding space of text and pose at different semantic levels.

[0084] This application performs global alignment processing on the text embedding and the pose embedding based on the hierarchical alignment module to obtain corresponding global features; performs part-level alignment processing on the text embedding and the pose embedding to obtain corresponding part-level features; and performs joint-level alignment processing on the text embedding and the pose embedding to obtain corresponding joint-level features; then, the global features, the part-level features, and the joint-level features are integrated to obtain corresponding integrated features; subsequently, the integrated features are used as the alignment features. This application, by using the hierarchical alignment module to perform global, part-level, and joint-level alignment on text embedding and pose embedding, can efficiently and accurately complete the semantic alignment processing of text embedding and pose embedding, thereby establishing a semantic correspondence between text and pose from global to local, and from coarse-grained to fine-grained, improving the richness and accuracy of the generated alignment features, and facilitating the provision of multi-level feature support for subsequent cross-modal optimization and image generation.

[0085] In some optional implementations of this embodiment, step S204 includes the following steps:

[0086] Obtain a pre-defined comparative learning strategy.

[0087] In this embodiment, the above-mentioned contrastive learning strategy is a strategy based on the momentum contrastive learning framework. The strategy includes: enabling cross-modal optimization to force semantically similar text and pose features to be close in the embedding space through contrastive learning, thereby enhancing the semantic consistency of hierarchical alignment.

[0088] Based on the contrastive learning strategy, the cross-modal adapter is used to enhance the semantic consistency of the aligned features to obtain the corresponding processed features.

[0089] In this embodiment, the optimization process corresponding to the semantic consistency enhancement processing based on cross-modal adapter alignment features includes:

[0090] 1. Comparative learning frameworks

[0091] Objective: To optimize the embedding space of text and pose through contrastive learning, bringing semantically similar features closer together.

[0092] Input: A batch size of N pairs of samples (z p i ,z t i ), where z p i and z t i These are positive sample pairs (from the same instance). p i: The pose embedding of the i-th sample. z t i : The text embedding of the i-th sample.

[0093] operate:

[0094] Contrast loss calculation: Using the InfoNCE loss function, positive sample pairs are forced to be close together in the embedding space, while negative sample pairs are forced to be far apart.

[0095]

[0096] Where τ is the temperature hyperparameter, used to control the sharpness of the similarity distribution in the contrastive loss. N is the batch size, i.e., the number of sample pairs. exp is the exponential function used to calculate the weights of similarity.

[0097] 2. Hierarchical Sampling Strategy: Global Alignment Layer: Calculates contrastive loss using complete sample pairs (all N samples). Location-level and Joint-level: Sample negative samples within the corresponding partitions (e.g., sample negative samples only within the "right arm" partition) to enhance local semantic consistency.

[0098] Cross-modal optimization, through contrastive learning, forces these features to be semantically consistent in the embedding space, thereby improving the accuracy and robustness of text-to-pose generation. Hierarchical sampling further enhances the consistency of local semantics (e.g., the correspondence between the text "right arm" and the joint).

[0099] The processed features are used as the target features.

[0100] In this embodiment, image generation utilizes a latent diffusion model and a cross-attention mechanism to inject semantically aligned features into the denoising process, generating images that meet the specified conditions. A gating mechanism dynamically fuses features at different levels, achieving coarse-to-fine conditional control and further improving the quality of the generated images.

[0101] This application acquires a preset contrastive learning strategy; then, based on the contrastive learning strategy, it uses the cross-modal adapter to enhance the semantic consistency of the alignment features, obtaining corresponding processed features; subsequently, these processed features are used as the target features. This application, by using the acquired contrastive learning strategy and the cross-modal adapter to enhance the semantic consistency of the alignment features, forces text features and pose features to be semantically consistent in the embedding space, and can accurately generate adapted multimodal target features, effectively improving the accuracy and robustness of text-to-pose generation. Furthermore, the generated target features provide conditional input for subsequent image generation processing, enabling the generated target image to simultaneously satisfy text description and pose constraints, thereby improving the quality of the generated target image.

[0102] In some alternative implementations, step S205 includes the following steps:

[0103] Obtain the preset cross-attention mechanism.

[0104] In this embodiment, the pose condition generator can specifically employ a diffusion model. The goal of the pose condition generator is to generate an image that meets the pose conditions based on the adapted multimodal features (target features).

[0105] The architecture of the pose condition generator (or simply generator): taking the diffusion model as an example, the image is generated step by step through the denoising process.

[0106] Conditional injection: In the l-th layer of the network, hierarchical features are injected into the denoising process through cross-attention.

[0107]

[0108] Among them, Q l K represents the query features of the current layer (features from the generator's denoising process). c and V c The keys and values ​​are the output from the hierarchical alignment module (i.e., the features output after hierarchical alignment processing), and d is the feature dimension used to scale the dot product attention.

[0109] Gating mechanism: used for dynamically fusing features at different levels (global, part-level, joint-level), the formula is as follows:

[0110]

[0111] Among them, H l This represents the hidden state of the current layer. k The gating weights, predicted from the current noise level, control the fusion ratio of features at different levels. k (H l H is the cross-attention output of the features of the k-th layer. l ′ represents the updated hidden state.

[0112] Output: Generate an image that matches the text description and pose conditions.

[0113] Based on the cross-attention mechanism, the target features are injected as conditional information into the denoising process of the pose condition generator.

[0114] In this embodiment, the denoising process of injecting the target features as conditional information into the attitude condition generator can be completed based on the above-described conditional injection process.

[0115] Based on the pose condition generator with injected conditional information, image generation processing is performed according to the target features to obtain the corresponding generated image.

[0116] In this embodiment, after the conditional information injection is completed, the pose condition generator is used to generate images step by step through the denoising process, and a generated image that meets the pose conditions is generated based on the adapted multimodal features (target features), which is then used as the final target image.

[0117] The generated image is used as the target image.

[0118] In this embodiment, image generation utilizes a latent diffusion model and a cross-attention mechanism to inject semantically aligned features into the denoising process, generating images that meet pose conditions. A gating mechanism dynamically fuses features at different levels, achieving coarse-to-fine condition control and further improving the quality of the generated images.

[0119] This application obtains a preset cross-attention mechanism; then, based on the cross-attention mechanism, it injects the target features as conditional information into the denoising process of the pose condition generator; subsequently, based on the pose condition generator with injected conditional information, it performs image generation processing according to the target features to obtain a corresponding generated image; finally, it uses the generated image as the target image. This application, by using a cross-attention mechanism to inject target features as conditional information into the denoising process of the pose condition generator, and then using the pose condition generator to perform image generation processing according to the target features, can generate a target image that meets the pose conditions, effectively improving the quality of the generated target image.

[0120] In some alternative implementations, step S206 includes the following steps:

[0121] The target image is processed to optimize its quality based on a preset quality optimization strategy, resulting in an optimized image.

[0122] In this embodiment, the specific implementation process of performing quality optimization processing on the target image based on the preset quality optimization strategy to obtain the corresponding optimized image will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0123] The optimized image is semantically validated based on a preset semantic validation strategy.

[0124] In this embodiment, the semantic verification strategy described above includes the following steps for performing semantic verification on the optimized image: segmenting the optimized image using a pre-trained semantic segmentation model (such as DeepLabv3+), checking whether it matches the pose conditions of the input text description and the reference image, thereby completing the semantic verification of the optimized image and generating the corresponding semantic verification result. If the optimized image is found to match the pose conditions of the input text description and the reference image, the optimized image is determined to have passed the semantic verification; otherwise, the optimized image is determined to have failed the semantic verification.

[0125] If the optimized image passes semantic verification, then the target output method corresponding to the optimized image is obtained.

[0126] In this embodiment, user information can be obtained, and the user's preferred target data push method can be queried using this information. This target data push method can then be used as the target output method. The user can be the one who inputs the text description above. Users can be tagged with matching data push methods based on their registration options, historical behavior data, or questionnaires. These data push methods may include at least pop-up notifications, interface displays, email sending, and in-site messaging.

[0127] The optimized image is output based on the target output method.

[0128] In this embodiment, the generated optimized image can be sent to the user according to the determined target output method to complete the output processing of the optimized image.

[0129] This application optimizes the target image using a preset quality optimization strategy to obtain an optimized image. Then, it performs semantic verification on the optimized image using a preset semantic verification strategy. If the optimized image passes the semantic verification, a target output method corresponding to the optimized image is obtained. Subsequently, the optimized image is output based on the target output method. This application optimizes the target image using a quality optimization strategy, performs semantic verification on the optimized image using a semantic verification strategy, and after detecting that the optimized image passes the semantic verification, obtains the target output method corresponding to the optimized image, and then outputs the optimized image based on the target output method. This effectively ensures the quality and accuracy of the sent optimized image, improves the intelligence of the optimized image output, and enhances the user experience.

[0130] In some optional implementations of this embodiment, the step of performing quality optimization processing on the target image based on a preset quality optimization strategy to obtain a corresponding optimized image includes the following steps:

[0131] The target image is subjected to noise suppression processing to obtain the corresponding first processed image.

[0132] In this embodiment, the goal of the noise suppression processing described above is to suppress noise and artifacts in the generated image, making it smoother and more natural. The noise suppression processing includes: using the Non-Local Means (NLM) algorithm to suppress noise; and performing adaptive thresholding on high-frequency components (such as detail coefficients after wavelet transform).

[0133] The first processed image is style-adjusted to obtain the corresponding second processed image.

[0134] In this embodiment, the goal of the style adjustment process is to transfer the style of one image to a generated image to change its visual effect. The implementation process of the style adjustment process includes: selecting a style image containing the desired visual style; using the generated target image as the initial image; extracting features from the initial image and the style image using a pre-trained style transfer model (such as Neural Style Transfer, AdaIN, etc.); fusing the style features into the initial image through an optimization process, applying the style features while maintaining the content structure; and outputting the style-transferred image, which has the structure of the initial image and the visual effect of the style image.

[0135] The second processed image is subjected to detail enhancement processing to obtain the corresponding third processed image.

[0136] In this embodiment, the goal of the above-mentioned detail enhancement processing is to enhance the details of the generated image (such as texture and edge sharpness) to make it more realistic and refined. The implementation process of detail enhancement processing includes: using a lightweight super-resolution network to enhance the details of the generated image; and using a Laplacian operator or an unsharpened mask to enhance the image edges.

[0137] The third processed image is used as the optimized image.

[0138] This application performs noise suppression processing on the target image to obtain a corresponding first processed image; then performs style adjustment processing on the first processed image to obtain a corresponding second processed image; subsequently, it performs detail enhancement processing on the second processed image to obtain a corresponding third processed image; and finally, it uses the third processed image as the optimized image. This application, by performing noise suppression, style adjustment, and detail enhancement processing on the target image, can automatically and accurately complete the quality optimization processing of the target image, resulting in a significantly improved optimized image in terms of noise, style, and detail, thereby effectively enhancing the visual quality, smoothness, and naturalness of the generated optimized image.

[0139] In some optional implementations of this embodiment, the hierarchical encoder includes a text encoder and a pose estimator; step S202 includes the following steps:

[0140] The text description is encoded using the text encoder to obtain word embeddings, and the word embeddings are used as the text embeddings.

[0141] In this embodiment, the hierarchical encoder described above employs a two-stream architecture (text encoder and pose estimator) to process text input and pose input. Specifically, the text encoder can be a pre-trained language model (such as BERT). Specifically, for a text sequence (text description) of length n, T = {t1, t2, ..., t...} n First, word embeddings are obtained through a pre-trained language model:

[0142]

[0143] Where d is the embedding dimension, representing the semantic features of each word.

[0144] The pose estimator is used to estimate the pose of the reference image to obtain the corresponding coordinate data.

[0145] In this embodiment, the pose estimator can specifically be an open-source pose estimator. A reference image can be input into the pose estimator to estimate the pose of the reference image and output joint coordinates that match the reference image. This represents the two-dimensional coordinates of m joints, where m is the number of joints. The obtained joint coordinates can be used as the coordinate data mentioned above.

[0146] Linear projection is performed on the coordinate data to obtain the pose embedding corresponding to the reference image.

[0147] In this embodiment, the coordinate data, i.e., the joint coordinates, can be linearly projected onto the same dimension as the text embedding to obtain the corresponding pose embedding. The specific formula is as follows:

[0148]

[0149] Among them, W p and b p These are learnable parameters used to align the pose and text embedding space.

[0150] This application encodes the text description using a text encoder to obtain word embeddings, and uses these word embeddings as the text embeddings. Then, it performs pose estimation on the reference image using a pose estimator to obtain corresponding coordinate data. Subsequently, it performs linear projection on the coordinate data to obtain the pose embedding corresponding to the reference image. This application achieves efficient and accurate feature extraction from the text description and reference image using a hierarchical encoder, improving feature extraction efficiency and ensuring the accuracy of the extracted text and pose embeddings.

[0151] In some alternative implementations, the user information obtained is subject to user consent and complies with relevant laws and policies.

[0152] Furthermore, any software tools or components not belonging to our company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.

[0153] Furthermore, the technological breakthroughs of this application are mainly reflected in three dimensions: First, the hierarchical encoder designed to extract pose features from macro to micro levels can simultaneously capture the overall action intent and local joint details; second, the innovative cross-modal adapter optimizes the embedding space of text and pose features through contrastive learning, enabling the model to understand the relationship between action descriptions such as "raising hands" and specific joint positions; finally, the proposed dynamic injection mechanism ensures the synergistic effect of pose information and text semantics during the generation process, maintaining image quality while achieving precise control.

[0154] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0155] It should be emphasized that, to further ensure the privacy and security of the target images, they can also be stored in a blockchain node.

[0156] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0157] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0158] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0159] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware with computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0160] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0161] Further reference Figure 3As a response to the above Figure 2 The implementation of the method shown in this application provides an embodiment of an image processing device based on artificial intelligence, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0162] like Figure 3 As shown, the AI-based image processing device 300 described in this embodiment includes: an acquisition module 301, an extraction module 302, an alignment module 303, an optimization module 304, a generation module 305, and an output module 306. Wherein:

[0163] The acquisition module 301 is used to acquire the input text description and reference image;

[0164] The extraction module 302 is used to extract features from the text description and the reference image based on a preset hierarchical encoder to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image.

[0165] Alignment module 303 is used to perform semantic alignment processing on the text embedding and the pose embedding based on a preset hierarchical alignment module to obtain corresponding alignment features;

[0166] The optimization module 304 is used to perform semantic optimization processing on the alignment features based on a preset cross-modal adapter to obtain the corresponding target features;

[0167] The generation module 305 is used to perform image generation processing corresponding to the target features based on a preset pose condition generator to obtain the corresponding target image;

[0168] The output module 306 is used to perform output processing on the target image.

[0169] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0170] In some optional implementations of this embodiment, the alignment module 303 includes:

[0171] The first processing submodule is used to perform global alignment processing on the text embedding and the pose embedding based on the hierarchical alignment module to obtain the corresponding global features;

[0172] The second processing submodule is used to perform part-level alignment processing on the text embedding and the pose embedding to obtain the corresponding part-level features;

[0173] The third processing submodule is used to perform joint-level alignment processing on the text embedding and the pose embedding to obtain the corresponding joint-level features;

[0174] The integration submodule is used to integrate the global features, the part-level features, and the joint-level features to obtain the corresponding integrated features;

[0175] The first determining submodule is used to use the integrated feature as the alignment feature.

[0176] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0177] In some optional implementations of this embodiment, the optimization module 304 includes:

[0178] The first acquisition submodule is used to acquire the preset contrastive learning strategy;

[0179] An enhancement submodule is used to perform semantic consistency enhancement processing on the aligned features based on the contrastive learning strategy using the cross-modal adapter to obtain the corresponding processed features;

[0180] The second determining submodule is used to use the processed features as the target features.

[0181] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0182] In some optional implementations of this embodiment, the generation module 305 includes:

[0183] The second acquisition submodule is used to acquire the preset cross-attention mechanism;

[0184] The injection submodule is used to inject the target features as conditional information into the denoising process of the pose condition generator based on the cross-attention mechanism.

[0185] The execution submodule is used to perform image generation processing based on the target features and obtain the corresponding generated image, based on the pose condition generator with injected condition information.

[0186] The third determining submodule is used to use the generated image as the target image.

[0187] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0188] In some optional implementations of this embodiment, the output module 306 includes:

[0189] The optimization submodule is used to perform quality optimization processing on the target image based on a preset quality optimization strategy to obtain a corresponding optimized image.

[0190] The verification submodule is used to perform semantic verification on the optimized image based on a preset semantic verification strategy.

[0191] The third acquisition submodule is used to acquire the target output method corresponding to the optimized image if the optimized image passes the semantic verification.

[0192] The output submodule is used to perform output processing on the optimized image based on the target output method.

[0193] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0194] In some optional implementations of this embodiment, the optimized submodule includes:

[0195] The first processing unit is used to perform noise suppression processing on the target image to obtain a corresponding first processed image;

[0196] The second processing unit is used to perform style adjustment processing on the first processed image to obtain the corresponding second processed image;

[0197] The third processing unit is used to perform detail enhancement processing on the second processed image to obtain the corresponding third processed image;

[0198] A determining unit is used to use the third processed image as the optimized image.

[0199] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0200] In some optional implementations of this embodiment, the hierarchical encoder includes a text encoder and a pose estimator; the extraction module 302 includes:

[0201] An encoding submodule is used to encode the text description based on the text encoder to obtain word embeddings, and to use the word embeddings as the text embeddings;

[0202] The fourth processing submodule is used to perform pose estimation on the reference image based on the pose estimator to obtain the corresponding coordinate data;

[0203] The projection submodule is used to perform linear projection on the coordinate data to obtain the pose embedding corresponding to the reference image.

[0204] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based image processing method in the aforementioned embodiments, and will not be repeated here.

[0205] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0206] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected via a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0207] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.

[0208] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 4. Of course, the memory 41 may include both the internal storage unit and its external storage device of the computer device 4. In this embodiment, the memory 41 is typically used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions for image processing methods based on artificial intelligence. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.

[0209] In some embodiments, the processor 42 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to execute computer-readable instructions stored in the memory 41 or to process data, for example, to execute computer-readable instructions of the artificial intelligence-based image processing method.

[0210] The network interface 43 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 4 and other electronic devices.

[0211] Compared with the prior art, the embodiments of this application have the following beneficial effects:

[0212] In this embodiment, firstly, input text description and reference image are acquired; then, feature extraction is performed on the text description and reference image based on a preset hierarchical encoder to obtain text embeddings corresponding to the text description and pose embeddings corresponding to the reference image; subsequently, semantic alignment processing is performed on the text embeddings and pose embeddings based on a preset hierarchical alignment module to obtain corresponding alignment features; subsequently, semantic optimization processing is performed on the alignment features based on a preset cross-modal adapter to obtain corresponding target features; further, image generation processing corresponding to the target features is performed based on a preset pose condition generator to obtain the corresponding target image; finally, the target image is output. This application extracts features from input text description and reference image using a hierarchical encoder to obtain corresponding text embeddings and pose embeddings, then performs semantic alignment processing on the text embeddings and pose embeddings using a hierarchical alignment module to obtain alignment features, then performs semantic optimization processing on the alignment features using a cross-modal adapter to obtain target features, and finally performs image generation processing corresponding to the target features using a pose condition generator to obtain and output the target image. This application processes the input text description and reference image by combining a hierarchical encoder, a hierarchical alignment module, a cross-modal adapter, and a pose condition generator. This enables precise pose control from text to image, thereby efficiently and accurately generating target images that meet pose conditions and improving the quality of the generated target images.

[0213] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the artificial intelligence-based image processing method described above.

[0214] Compared with the prior art, the embodiments of this application have the following main advantages:

[0215] In this embodiment, firstly, input text description and reference image are acquired; then, feature extraction is performed on the text description and reference image based on a preset hierarchical encoder to obtain text embeddings corresponding to the text description and pose embeddings corresponding to the reference image; subsequently, semantic alignment processing is performed on the text embeddings and pose embeddings based on a preset hierarchical alignment module to obtain corresponding alignment features; subsequently, semantic optimization processing is performed on the alignment features based on a preset cross-modal adapter to obtain corresponding target features; further, image generation processing corresponding to the target features is performed based on a preset pose condition generator to obtain the corresponding target image; finally, the target image is output. This application extracts features from input text description and reference image using a hierarchical encoder to obtain corresponding text embeddings and pose embeddings, then performs semantic alignment processing on the text embeddings and pose embeddings using a hierarchical alignment module to obtain alignment features, then performs semantic optimization processing on the alignment features using a cross-modal adapter to obtain target features, and finally performs image generation processing corresponding to the target features using a pose condition generator to obtain and output the target image. This application processes the input text description and reference image by combining a hierarchical encoder, a hierarchical alignment module, a cross-modal adapter, and a pose condition generator. This enables precise pose control from text to image, thereby efficiently and accurately generating target images that meet pose conditions and improving the quality of the generated target images.

[0216] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0217] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. An image processing method based on artificial intelligence, characterized in that, Includes the following steps: Obtain the input text description and reference image; Based on a preset hierarchical encoder, features are extracted from the text description and the reference image to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image. Based on a preset hierarchical alignment module, the text embedding and the pose embedding are semantically aligned to obtain the corresponding alignment features. The alignment features are semantically optimized based on a preset cross-modal adapter to obtain the corresponding target features; Based on a preset pose condition generator, image generation processing corresponding to the target features is performed to obtain the corresponding target image; The target image is then processed for output. The step of performing semantic alignment processing on the text embedding and the pose embedding based on a preset hierarchical alignment module to obtain the corresponding alignment features specifically includes: The hierarchical alignment module performs global alignment processing on the text embedding and the pose embedding to obtain the corresponding global features. The text embedding and the pose embedding are aligned at the part level to obtain the corresponding part-level features; The text embedding and the pose embedding are aligned at the joint level to obtain the corresponding joint-level features; The global features, the location-level features, and the joint-level features are integrated to obtain the corresponding integrated features. The integration feature is used as the alignment feature; The step of performing image generation processing corresponding to the target features based on a preset pose condition generator to obtain the corresponding target image specifically includes: Obtain the preset cross-attention mechanism; Based on the cross-attention mechanism, the target features are injected as conditional information into the denoising process of the pose condition generator; Based on the pose condition generator with injected conditional information, image generation processing is performed according to the target features to obtain the corresponding generated image; The generated image is used as the target image.

2. The image processing method based on artificial intelligence according to claim 1, characterized in that, The step of performing semantic optimization processing on the alignment features based on a preset cross-modal adapter to obtain the corresponding target features specifically includes: Obtain a pre-defined comparative learning strategy; Based on the contrastive learning strategy, the cross-modal adapter is used to enhance the semantic consistency of the aligned features to obtain the corresponding processed features; The processed features are used as the target features.

3. The image processing method based on artificial intelligence according to claim 1, characterized in that, The step of outputting the target image specifically includes: The target image is subjected to quality optimization processing based on a preset quality optimization strategy to obtain a corresponding optimized image. The optimized image is semantically validated based on a preset semantic validation strategy; If the optimized image passes the semantic verification, then the target output method corresponding to the optimized image is obtained; The optimized image is output based on the target output method.

4. The image processing method based on artificial intelligence according to claim 3, characterized in that, The step of performing quality optimization processing on the target image based on a preset quality optimization strategy to obtain a corresponding optimized image specifically includes: The target image is subjected to noise suppression processing to obtain the corresponding first processed image; The first processed image is style-adjusted to obtain the corresponding second processed image; The second processed image is subjected to detail enhancement processing to obtain the corresponding third processed image; The third processed image is used as the optimized image.

5. The image processing method based on artificial intelligence according to claim 1, characterized in that, The hierarchical encoder includes a text encoder and a pose estimator; the step of extracting features from the text description and the reference image based on the preset hierarchical encoder to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image specifically includes: The text description is encoded using the text encoder to obtain word embeddings, and the word embeddings are used as the text embeddings. Based on the pose estimator, the pose of the reference image is estimated to obtain the corresponding coordinate data; Linear projection is performed on the coordinate data to obtain the pose embedding corresponding to the reference image.

6. An image processing device based on artificial intelligence, characterized in that, include: The acquisition module is used to acquire the input text description and reference image; The extraction module is used to extract features from the text description and the reference image based on a preset hierarchical encoder to obtain the text embedding corresponding to the text description and the pose embedding corresponding to the reference image. An alignment module is used to perform semantic alignment processing on the text embedding and the pose embedding based on a preset hierarchical alignment module to obtain corresponding alignment features; The optimization module is used to perform semantic optimization processing on the alignment features based on a preset cross-modal adapter to obtain the corresponding target features; The generation module is used to perform image generation processing corresponding to the target features based on a preset pose condition generator to obtain the corresponding target image; The output module is used to process the target image for output. The alignment module includes: The first processing submodule is used to perform global alignment processing on the text embedding and the pose embedding based on the hierarchical alignment module to obtain the corresponding global features; The second processing submodule is used to perform part-level alignment processing on the text embedding and the pose embedding to obtain the corresponding part-level features; The third processing submodule is used to perform joint-level alignment processing on the text embedding and the pose embedding to obtain the corresponding joint-level features; The integration submodule is used to integrate the global features, the part-level features, and the joint-level features to obtain the corresponding integrated features; A first determining submodule is used to use the integrated feature as the alignment feature; The generation module includes: The second acquisition submodule is used to acquire the preset cross-attention mechanism; The injection submodule is used to inject the target features as conditional information into the denoising process of the pose condition generator based on the cross-attention mechanism. The execution submodule is used to perform image generation processing based on the target features and obtain the corresponding generated image, based on the pose condition generator with injected condition information. The third determining submodule is used to use the generated image as the target image.

7. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the artificial intelligence-based image processing method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the artificial intelligence-based image processing method as described in any one of claims 1 to 5.

Images