Adaptive parameter adjustment method and system for multi-mode switching of facial beautification camera

Abstract

The invention provides a parameter adaptive adjustment method and system for multi-mode switching of a facial beautification camera, and relates to the technical field of camera adjustment, and the method comprises the steps: obtaining a photographing target feature point cloud, carrying out the data fusion of the photographing target feature point cloud and photographing environment information, generating a photographing feature scene, inputting the photographing feature scene into a facial beautification demand analysis model, and obtaining a facial beautification feature demand; multi-mode shooting decision making is carried out according to the shooting feature scene and the beauty feature requirement, a decision space is established, optimization analysis is carried out in combination with a beauty prediction evaluation model, and a multi-mode shooting optimization strategy is determined; and performing beautification preview feedback correction on the multi-mode shooting optimization strategy, generating a multi-mode switching adjustment strategy, and executing adaptive switching adjustment. According to the method and the device, the technical problem that the facial beautification effect is distorted or does not conform to the user expectation due to the fact that facial beautification mode switching depends on fixed parameters and manual adjustment of the user in the prior art is solved, the facial beautification camera mode switching adaptability is improved, and then the facial beautification effect and the user experience feeling are improved.

Description

Technical Field

[0001] The present application relates to the field of camera adjustment technology, and in particular to a parameter adaptive adjustment method and system for multi-mode switching of a beauty camera. Background Art

[0002] Beauty cameras improve the visual effects of photos by identifying facial features and combining beautification algorithms such as skin smoothing, whitening, and face thinning. At present, most beauty cameras rely on preset fixed parameters and manual adjustments by users. Users need to manually select beauty modes (such as natural, exquisite, retro, etc.) and adjust parameters (such as skin smoothing intensity, whitening degree, etc.) according to different shooting scenes and goals. Although some high-end cameras support intelligent recommendation modes, they still mainly rely on lighting recognition or skin color analysis and cannot achieve truly intelligent adjustments. Users usually need to repeatedly try different modes or manually fine-tune parameters to obtain the ideal beauty effect, which not only increases the operating cost, but may also cause the beauty effect to be unstable, distorted or not meet user expectations under different lighting conditions or different character features, thereby affecting user satisfaction. Summary of the invention

[0003] The present application provides a parameter adaptive adjustment method and system for multi-mode switching of a beauty camera, which solves the technical problem in the prior art that the beauty mode switching relies on fixed parameters and manual adjustment by the user, resulting in a lack of intelligent adaptability of the beauty mode under different shooting targets and environments, causing the beauty effect to be distorted or not meeting user expectations. The application achieves the technical effect of improving the adaptability of the beauty camera mode switching in different shooting scenes, thereby improving the beauty effect and user experience.

[0004] In view of the above problems, on the one hand, the present application provides a parameter adaptive adjustment method for multi-mode switching of a beauty camera, the method comprising: obtaining a user shooting instruction of the beauty camera, wherein the user shooting instruction includes a shooting target and shooting environment information; performing feature detection according to the shooting target to obtain a shooting target feature point cloud, and performing data fusion on the shooting target feature point cloud and the shooting environment information to generate a shooting feature scene; inputting the shooting feature scene into a beauty requirement parsing model to obtain a beauty feature requirement; performing multi-mode shooting decisions on the beauty camera according to the shooting feature scene and the beauty feature requirement, and establishing a multi-mode shooting decision space; based on the shooting feature scene, performing optimization analysis on the multi-mode shooting decision space according to a beauty prediction and evaluation model to determine a multi-mode shooting optimization strategy; performing beauty preview feedback correction on the multi-mode shooting optimization strategy according to the preview interface of the beauty camera, generating a multi-mode switching adjustment strategy, and executing adaptive switching adjustment of the beauty camera according to the multi-mode switching adjustment strategy.

[0005] On the other hand, the present application also provides a parameter adaptive adjustment system for multi-mode switching of a beauty camera, the system comprising: a shooting instruction acquisition module, used to obtain a user shooting instruction of the beauty camera, wherein the user shooting instruction includes a shooting target and shooting environment information; a feature scene generation module, used to perform feature detection according to the shooting target, obtain a shooting target feature point cloud, and perform data fusion on the shooting target feature point cloud and the shooting environment information to generate a shooting feature scene; a beauty requirement analysis module, used to input the shooting feature scene into a beauty requirement analysis model to obtain a beauty feature requirement; a shooting decision module, used to make a multi-mode shooting decision for the beauty camera according to the shooting feature scene and the beauty feature requirement, and establish a multi-mode shooting decision space; an optimization analysis module, used to perform an optimization analysis on the multi-mode shooting decision space based on the shooting feature scene and a beauty prediction evaluation model to determine a multi-mode shooting optimization strategy; a switching adjustment decision module, used to perform beauty preview feedback correction on the multi-mode shooting optimization strategy according to the preview interface of the beauty camera, generate a multi-mode switching adjustment strategy, and execute adaptive switching adjustment of the beauty camera according to the multi-mode switching adjustment strategy.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: By obtaining the user's shooting instructions, including shooting target and shooting environment information, basic data is provided for subsequent feature detection and beauty demand analysis, ensuring that the technical solution can be targeted according to the user's actual needs and shooting scenes. Feature detection is performed on the shooting target to generate the feature point cloud of the shooting target, and these feature point clouds are fused with the shooting environment information to generate shooting feature scenes, providing comprehensive and accurate input data for beauty demand analysis, so that the subsequent beauty demand analysis can more accurately reflect the user's actual needs and the characteristics of the shooting scene. The shooting feature scene is input into the beauty demand analysis model to intelligently identify the user's beauty needs, providing a basis for subsequent multi-mode shooting decisions, avoiding universal and fixed beauty settings, and improving the targeting and personalization of beauty. According to the shooting feature scene and beauty feature requirements, a multi-mode shooting decision space is established, and multiple possible decision plans are determined for the shooting mode of the beauty camera. Through optimization analysis, the shooting strategy that best suits the current shooting scene and beauty needs is automatically selected, avoiding the tedious operation of users manually switching modes, while improving the beauty effect. Through the preview interface of the beauty camera, real-time feedback and correction are performed on the multi-mode shooting optimization strategy, a multi-mode switching adjustment strategy is generated, and adaptive switching adjustment is performed according to the strategy, so that the final adjustment strategy can better adapt to the actual shooting effect, meet user expectations, and further improve the user experience.

[0007] In summary, this application realizes the adaptive adjustment of the beauty camera in different shooting scenes through intelligent feature detection, data fusion, beauty demand analysis, multi-mode shooting decision-making and real-time feedback correction mechanism. This solution can not only automatically select the optimal beauty mode according to the shooting target and environment, but also ensure the naturalness and consistency of the beauty effect through real-time feedback correction, reduce the need for manual adjustment by users, and significantly improve the intelligence level of the beauty camera and the user experience and satisfaction.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of a parameter adaptive adjustment method for multi-mode switching of a beauty camera provided in an embodiment of the present application.

[0010] Figure 2 A schematic diagram of a process for establishing a multi-mode shooting decision space in a parameter adaptive adjustment method for multi-mode switching of a beauty camera provided in an embodiment of the present application.

[0011] Figure 3 A schematic diagram of a process for determining a multi-mode shooting optimization strategy in a parameter adaptive adjustment method for multi-mode switching of a beauty camera provided in an embodiment of the present application.

[0012] Figure 4 A schematic diagram of the structure of a parameter adaptive adjustment system for multi-mode switching of a beauty camera provided in an embodiment of the present application.

[0013] Explanation of the reference numerals: shooting instruction acquisition module 10 , feature scene generation module 20 , beauty requirement analysis module 30 , shooting decision module 40 , optimization analysis module 50 , switching adjustment decision module 60 . DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a parameter adaptive adjustment method and system for multi-mode switching of a beauty camera, thereby solving the technical problem in the prior art that the beauty mode lacks intelligent adaptability in different shooting targets and environments, resulting in distorted beauty effects or not meeting user expectations, due to the reliance on fixed parameters and manual adjustment by the user for beauty mode switching. The present application achieves the technical effect of improving the adaptability of mode switching of the beauty camera in different shooting scenes, thereby improving the beauty effect and user experience.

[0015] Embodiment 1, as Figure 1As shown, the embodiment of the present application provides a parameter adaptive adjustment method for multi-mode switching of a beauty camera, the method comprising: Step S1: obtaining a user shooting instruction of a beauty camera, wherein the user shooting instruction includes shooting target and shooting environment information.

[0016] Specifically, the user shooting command is the operation command issued by the user when using the beauty camera, which contains information related to the user's intention to shoot. When the user turns on the beauty camera and prepares to shoot, the camera system relies on the camera, light sensor, user interface, etc. combined with the software algorithm to receive and parse the user shooting command. This command contains the shooting target and shooting environment information. Among them, the shooting target refers to the object being photographed, such as people, scenery, objects, etc.; the shooting environment information includes factors such as the surrounding light conditions (such as strong light, weak light, natural light, etc.) and background (such as indoor background, outdoor scenery, etc.) during shooting. For example, the shooting target is obtained and identified through the camera and computer vision methods (such as OpenCV), the camera's automatic exposure (AE), automatic white balance (AWB) and autofocus (AF) are used to detect information such as light intensity, color temperature, depth of field, and the shooting mode selected by the user is obtained through the user interface.

[0017] By obtaining the user's shooting instructions of the beauty camera, basic information is provided for subsequent beauty processing, so that the beauty camera can perform targeted processing according to the specific shooting target and environment.

[0018] Step S2: performing feature detection according to the shooting target to obtain a shooting target feature point cloud, and performing data fusion on the shooting target feature point cloud and the shooting environment information to generate a shooting feature scene.

[0019] Specifically, for the shooting target, the beauty camera uses an image recognition algorithm to perform feature detection and identify specific features of the shooting target, such as detecting the position of facial features, contour shape and other features in human shooting. A feature point cloud is constructed based on the identified shooting target features. The feature points in the feature point cloud can represent the shape, structure and other information of the shooting target. Then, the feature point cloud is fused with the shooting environment information (such as light intensity, background color, etc.) to generate a shooting feature scene containing comprehensive information of the shooting target and environment. Exemplarily, the face detection algorithm of OpenCV is used to extract the key points of the face (such as eyes, nose, mouth, etc.) to generate a feature point cloud. At the same time, combined with the light intensity information detected by the light sensor, a shooting feature scene containing human features and light conditions is generated.

[0020] By integrating the shooting target and environmental information, a more comprehensive shooting feature scene is constructed, providing a basis for more accurate analysis of beauty needs, making the beauty effect more in line with the actual shooting situation.

[0021] Step S3: inputting the shooting feature scene into the beauty requirement analysis model to obtain beauty feature requirements.

[0022] Specifically, the beauty requirement analysis model is a pre-built algorithm model that can analyze the required beauty features based on the input shooting feature scene, such as the degree of skin smoothing, whitening, and facial beautification direction. This model can be trained based on a large amount of sample data, for example, by collecting beauty effect data under different shooting scenes and using machine learning algorithms (such as neural network algorithms) to build it.

[0023] The shooting feature scene generated in step S2 is input into the beauty requirement parsing model. The model outputs the user's beauty feature requirements by analyzing the feature point cloud and shooting environment information in the shooting feature scene. When the shooting feature scene is input, the model will output the corresponding beauty feature requirements based on its internal algorithms and parameters, that is, the user's specific requirements for beauty effects, such as the degree of skin smoothing and whitening. For example, if the shooting feature scene is a close-up of a person's face in a dark light environment, the model may parse that a higher degree of skin smoothing is required to cover up skin blemishes that may appear in the dark light, and appropriate whitening is required to brighten the skin tone.

[0024] Through the beauty demand analysis model, it can intelligently identify users' beauty needs, provide a basis for subsequent multi-mode shooting decisions, and achieve personalized customization of beauty effects.

[0025] Step S4: performing a multi-mode shooting decision on the beauty camera according to the shooting feature scene and the beauty feature requirement, and establishing a multi-mode shooting decision space.

[0026] Specifically, based on the shooting feature scene obtained in step S2 and the beauty feature requirement parsed in step S3, a pre-set rule or algorithm is used to make a multi-mode shooting decision to determine the various shooting modes that the beauty camera can adopt, such as different filter modes, different beauty intensity modes, etc. For example, if the beauty feature requirement is high skin grinding and appropriate whitening, and the shooting feature scene is a person in an outdoor landscape, then the system may consider adopting a fresh-style filter mode, a medium-intensity beauty mode, and other decisions. All possible multi-mode shooting decisions are combined to construct a multi-mode shooting decision space, which includes the possibility of different shooting mode combinations and parameter settings.

[0027] By establishing a multi-mode shooting decision space, a variety of possible shooting mode options are provided for the beauty camera, which increases the flexibility to cope with different shooting scenes and beauty needs and improves the possibility of finding the optimal shooting mode.

[0028] Step S5: Based on the shooting feature scene, the multi-mode shooting decision space is optimized and analyzed according to the beauty prediction evaluation model to determine the multi-mode shooting optimization strategy.

[0029] Specifically, based on the shooting feature scene in step S2, various decisions in the multi-mode shooting decision space are input into the beauty prediction and evaluation model one by one. This beauty prediction and evaluation model is constructed based on aesthetic principles and a large amount of beauty effect data. It can predict the beauty effect that each decision may produce according to various decisions in the multi-mode shooting decision space, and evaluate it, for example, from the aspects of beauty, naturalness, etc. According to the evaluation results of the beauty prediction and evaluation model, an optimization algorithm (such as genetic algorithm, particle swarm optimization, etc.) is used to perform optimization analysis, and the most optimized strategy is selected from a large number of shooting mode decisions, and determined as a multi-mode shooting optimization strategy, so as to improve the quality of the beauty effect of the final shooting, so that the photos or videos taken are more in line with aesthetic standards.

[0030] Step S6: performing beauty preview feedback correction on the multi-mode shooting optimization strategy according to the preview interface of the beauty camera, generating a multi-mode switching adjustment strategy, and performing adaptive switching adjustment of the beauty camera according to the multi-mode switching adjustment strategy.

[0031] Specifically, through the preview interface of the beauty camera, feedback information from the user after preview is received (such as sliding the adjustment bar, clicking the confirmation button, etc.), and the previously determined multi-mode shooting optimization strategy is adjusted and corrected according to the user's feedback, thereby generating a multi-mode switching adjustment strategy, and automatically adjusting the beauty mode and parameters according to the strategy.

[0032] Through preview feedback correction, the accuracy of the beautification effect and user satisfaction are further improved, so that the final beautification effect can better meet user expectations.

[0033] Further, such as Figure 2 As shown, step S4 includes: Step S41: obtaining a plurality of preset shooting modes of the beauty camera.

[0034] Step S42: collecting the shooting parameter constraint information of the plurality of preset shooting modes to obtain a multi-mode shooting constraint domain.

[0035] Step S43: performing association optimization on the multi-mode shooting constraint domain according to the shooting feature scene and the beauty feature requirement to establish a multi-mode shooting optimization domain.

[0036] Step S44: performing shooting parameter decisions on the plurality of preset shooting modes according to the multi-mode shooting optimization domain to generate the multi-mode shooting decision space.

[0037] Specifically, the beauty camera system stores multiple preset shooting modes with different filter effects, beauty styles, etc. For example, one preset shooting mode is "fresh style", which will make the overall tone of the photo light green and the beauty effect more natural; there is also an "old photo" mode, which will add retro tones and specific beauty effects to the photo to simulate the feeling of old photos. The relevant information of these preset shooting modes is read from the storage module of the beauty camera system to provide basic shooting mode options for subsequent optimization decisions based on shooting scenes and beauty needs, so that the camera can be adjusted based on the existing mode instead of building a shooting mode from scratch.

[0038] Each preset shooting mode has some information that limits its parameter adjustment range, that is, shooting parameter constraint information. For example, in the "fresh style" shooting mode, the skin grinding parameters may be limited to a smaller range (30% to 50%) to maintain the natural texture of the skin; the whitening parameters also have a reasonable upper limit (less than 50%) to avoid excessive whitening. These parameter constraint information is the shooting parameter constraint information. Traverse each preset shooting mode, find the corresponding shooting parameter constraint information, such as the constraint values ​​of parameters such as skin grinding, whitening, and filter strength, and then integrate these values ​​to form a multi-mode shooting constraint domain. This multi-mode shooting constraint domain is a set of parameter constraint ranges composed of the shooting parameter constraint information of multiple preset shooting modes, which clarifies the parameter adjustment restriction range of each preset shooting mode, which helps to optimize the parameters reasonably according to actual needs in the subsequent steps, avoid unreasonable parameter settings, and ensure that the beauty effect is within a certain reasonable range.

[0039] The parameter constraints in the multi-mode shooting constraint domain are optimized by associating shooting feature scenes (such as human faces under strong light, indoor scenery, etc.) and beauty feature requirements (such as high skin smoothing, specific color tones, etc.), so that the parameter constraints of each preset shooting mode are more in line with the current shooting situation. The set of parameter ranges and constraints after these associative optimizations is the multi-mode shooting optimization domain. For example, if the beauty feature requirement is high skin smoothing and the shooting environment is dark light, the constraint range of the skin smoothing parameters of the preset shooting mode that originally has tight skin smoothing parameter constraints may be appropriately relaxed. Through associative optimization, the parameter constraints of the preset shooting mode are more in line with the actual shooting scenes and beauty requirements, which improves the adaptability of the preset shooting mode in the current shooting situation and provides conditions for generating a more reasonable shooting decision space.

[0040] According to the parameter range and constraints in the multi-mode shooting optimization domain, parameters are configured for each preset shooting mode, multiple possible shooting parameter combination schemes under each preset shooting mode are determined, all preset shooting modes and their possible parameter combinations are summarized, and a multi-mode shooting decision space is constructed. Exemplarily, the step size is set according to the minimum parameter adjustment interval of the beauty camera (e.g., the minimum parameter adjustment range is 1%, and the adjustment step size is set to 2 times, i.e., 2%), and each parameter range is taken according to the step size (e.g., the skin grinding value range is 30% to 50%, then within the range of 30% to 50%, a parameter value is taken every 2%). The multiple shooting values ​​of the determined multiple shooting parameters are randomly combined, and verified with constraints, unreasonable parameter combinations are excluded, multiple shooting parameter combination schemes are generated, and these shooting parameter combination schemes are summarized to construct a multi-mode shooting decision space. The multi-mode shooting decision space contains various possibilities of different preset shooting modes under optimized shooting parameters, providing an initial solution set for subsequent optimization analysis.

[0041] Further, step S43 includes: Step S431: taking the shooting feature scene and the beauty feature requirement as shooting retrieval constraints, performing shooting parameter record retrieval on the plurality of preset shooting modes to obtain a multi-mode shooting parameter retrieval set.

[0042] Step S432: performing data cleaning according to the multi-mode shooting parameter retrieval set to obtain a multi-mode shooting record set.

[0043] Step S433: performing trigger interval calibration according to the multi-mode shooting record set to generate a multi-mode shooting association domain.

[0044] Step S434: performing intersection identification on the multi-mode shooting constraint domain and the multi-mode shooting association domain to generate the multi-mode shooting optimization domain.

[0045] Specifically, the shooting feature scene and beauty feature requirements are set as retrieval conditions, and the retrieval algorithm is used to find parameter records that meet the conditions (i.e., meet the shooting feature scene and beauty feature requirements) in the parameter records of the preset shooting mode to obtain a multi-mode shooting parameter retrieval set. Shooting parameter record retrieval can quickly screen out shooting parameter records that may meet the current shooting scene and beauty requirements, providing a preliminary candidate set for further optimization of the shooting mode, reducing the amount of data to be processed later.

[0046] Each record in the multi-mode shooting parameter retrieval set is checked, corrected, deleted, etc. to remove duplicate, incomplete or non-compliant data to obtain a multi-mode shooting record set. Some data validation rules can be set, such as the reasonable range of dermabrasion parameters is 0% to 100%, whitening parameters cannot exceed a certain value, etc., and records that do not meet these rules are corrected or deleted to obtain a multi-mode shooting record set. The multi-mode shooting record set obtained through the data cleaning operation is a set of relatively accurate and useful multi-mode shooting parameter records obtained after data cleaning, which improves the quality of the data.

[0047] For each shooting parameter (such as skin smoothing, whitening, filter strength, etc.) in the multi-mode shooting record set, the trigger interval is calibrated according to preset rules or based on a large amount of data statistics. For example, by analyzing the distribution of skin smoothing parameters in a large number of user-satisfied beauty photos, the trigger intervals corresponding to different skin smoothing effects are determined. The calibration process can use data statistical analysis algorithms, such as calculating the mean, standard deviation, etc., to determine a reasonable trigger interval. After calibrating the trigger intervals of all parameters, these interval relationships are combined to form a multi-mode shooting association domain. The multi-mode shooting association domain is a set of trigger interval relationships of different shooting mode parameters, which clarifies the trigger interval relationships of different shooting mode parameters, so that in subsequent operations, the shooting mode can be adjusted more accurately according to the shooting scene and beauty requirements, and the matching degree between the shooting mode and the beauty effect is improved.

[0048] Compare and analyze the parameters in the multi-mode shooting constraint domain and the multi-mode shooting association domain, traverse each parameter in the two domains (such as skin grinding parameters, whitening parameters, filter strength, etc.), compare the parameter value ranges of the corresponding shooting modes in the two domains one by one, find the same parts and combine them to generate a multi-mode shooting optimization domain. The parameters in this multi-mode shooting optimization domain not only meet the original constraints of the shooting mode, but also meet the trigger interval relationship calibrated according to the shooting scene and beauty requirements. It is more reasonable and accurate to make shooting parameter decisions in this optimization domain, which helps to generate a multi-mode shooting decision space that better meets actual needs.

[0049] Further, such as Figure 3 As shown, step S5 includes: Step S51: extracting a first mode shooting decision space according to the multi-mode shooting decision space.

[0050] Step S52: Based on the shooting feature scene, the first mode shooting decision space is evaluated and optimized according to the beauty prediction evaluation model to obtain the first mode shooting decision optimization space.

[0051] Step S53: constructing a beauty fitness evaluation network, wherein the beauty fitness evaluation network includes beauty naturalness weight, beauty transition naturalness weight and visual beautification quality weight.

[0052] Step S54: maximizing the beauty fitness of the first mode shooting decision space according to the beauty fitness evaluation network to obtain a first mode shooting optimization strategy.

[0053] Step S55: adding the first mode shooting optimization strategy to the multi-mode shooting optimization strategy, and continuing to perform optimization analysis on the multi-mode shooting decision space according to the shooting feature scene, the beauty prediction evaluation model and the beauty fitness evaluation network to generate the multi-mode shooting optimization strategy.

[0054] Specifically, from the large set of the entire multi-mode shooting decision space, a decision space related to a specific mode is randomly extracted, namely the first mode shooting decision space. Based on the determined shooting feature scene, various decision parameters in the first mode shooting decision space are input into the beauty prediction evaluation model. This model predicts the beauty effect based on the input decision parameters, and evaluates and optimizes these parameter combinations according to the preset evaluation criteria, thereby obtaining the first mode shooting decision optimization space. This first mode shooting decision optimization space is a shooting parameter combination range that is more suitable for the current shooting feature scene and shooting mode after evaluating and optimizing the first mode shooting decision space.

[0055] Construct a beauty fitness evaluation network that includes the weights of naturalness of beauty, naturalness of transition of beauty, and quality of visual beautification. Naturalness of beauty measures whether the effect after beauty looks natural, for example, whether the skin texture of the face is preserved naturally after beauty; naturalness of transition of beauty measures whether the transition between different areas of the face is natural, such as the transition between the cheek and the chin; and quality of visual beautification measures the overall visual beautification effect, such as the overall beauty after beauty. Set the value of each weight according to actual needs, for example, the weight of naturalness of beauty is 0.4, the weight of naturalness of transition of beauty is 0.3, and the weight of quality of visual beautification is 0.3. The beauty fitness evaluation network can comprehensively evaluate the naturalness, naturalness of transition, and quality of visual beautification of the beauty effect under each combination of shooting parameters to obtain a comprehensive beauty fitness score.

[0056] The shooting parameter combinations in the first mode shooting decision space are input into the beauty fitness evaluation network one by one. The network calculates these shooting parameter combinations according to the beauty naturalness weight, beauty transition naturalness weight and visual beautification quality weight, and obtains multiple beauty fitness corresponding to each shooting parameter combination. Combined with the optimization algorithm (such as genetic algorithm, particle swarm optimization), the parameter combination with the highest beauty fitness is found and determined as the first mode shooting optimization strategy, which can achieve the best beauty effect in the current shooting mode.

[0057] The obtained first-mode shooting optimization strategy is added to the multi-mode shooting optimization strategy. Then, based on the shooting feature scene, the beauty prediction evaluation model and the beauty fitness evaluation network are used to optimize the decision parameters of each shooting mode in the multi-mode shooting decision space one by one, and the parameter combination in the multi-mode shooting decision space is continuously adjusted. Finally, a multi-mode shooting optimization strategy is generated, which provides comprehensive guidance for the adaptive switching adjustment of the beauty camera and further improves the beauty effect.

[0058] Further, step S52 includes: Step S521: extracting a first shooting decision according to the first mode shooting decision space.

[0059] Step S522: Input the shooting feature scene and the first shooting decision into the beauty prediction evaluation model to obtain a first beauty prediction evaluation result, wherein the beauty prediction evaluation model includes multi-dimensional beauty prediction evaluation indicators, and the multi-dimensional beauty prediction evaluation indicators include beauty naturalness, beauty transition naturalness and visual beautification quality.

[0060] Step S523: determining whether the first beauty prediction evaluation result satisfies the beauty prediction evaluation constraint.

[0061] Step S524: if the first beauty prediction evaluation result satisfies the beauty prediction evaluation constraint, adding the first shooting decision to the first mode shooting decision optimization space.

[0062] Step S525: if the first beauty prediction evaluation result does not satisfy the beauty prediction evaluation constraint, the first shooting decision is eliminated.

[0063] Step S526: Based on the shooting feature scene and the beauty prediction evaluation model, the first mode shooting decision space is continuously evaluated and optimized according to the beauty prediction evaluation constraint to obtain the first mode shooting decision optimization space.

[0064] Specifically, from the first mode shooting decision space set, a shooting decision is selected according to certain rules, which is defined as the first shooting decision. It can be selected randomly or in a certain order (such as in the order of parameter values, etc.).

[0065] The beauty prediction evaluation model is a model based on machine learning, such as support vector machines (SVM) or neural networks (such as multi-layer perceptrons). When training this model, a large amount of annotated before-and-after beauty image data is used. The annotation information includes the scores of multi-dimensional beauty prediction evaluation indicators such as beauty naturalness, beauty transition naturalness, and visual beautification quality. The shooting feature scene and the first shooting decision are input into the beauty prediction evaluation model. The model calculates the shooting parameters in the first shooting decision based on its internal algorithm and pre-trained parameters, based on multi-dimensional beauty prediction evaluation indicators such as beauty naturalness, beauty transition naturalness, and visual beautification quality, and obtains the first beauty prediction evaluation result.

[0066] The beauty prediction evaluation constraint is a pre-set restriction condition on the beauty prediction evaluation result. For example, specific numerical restrictions such as requiring the beauty naturalness score to be above 80 points, the beauty transition naturalness score to be above 75 points, and the visual beautification quality score to be above 85 points, or restrictions such as the proportional relationship between various indicators. The first beauty prediction evaluation result obtained is compared with the pre-set beauty prediction evaluation constraint. For example, check whether the scores of beauty naturalness, beauty transition naturalness, and visual beautification quality meet the set minimum requirements or proportional relationship.

[0067] If the first beauty prediction evaluation result meets the beauty prediction evaluation constraint, then the first shooting decision is added to the first mode shooting decision optimization space. This optimization space is constructed step by step, and decisions that meet the requirements will be included in it. If the first beauty prediction evaluation result does not meet the beauty prediction evaluation constraint, the first shooting decision is eliminated and no longer considered in the subsequent optimization process.

[0068] Based on the shooting feature scenes and the beauty prediction evaluation model, and in accordance with the requirements of the beauty prediction evaluation constraints, other shooting decisions in the first mode shooting decision space are continuously evaluated and optimized, and shooting decisions that meet the requirements are continuously screened out and added to the first mode shooting decision optimization space, so as to finally obtain the complete first mode shooting decision optimization space.

[0069] By comprehensively evaluating and optimizing all shooting decisions in the first-mode shooting decision space, we obtain a first-mode shooting decision optimization space that meets the beauty prediction evaluation constraints. The shooting decisions in this space have undergone rigorous multi-dimensional evaluation and screening, providing high-quality candidate solutions for the subsequent beauty fitness maximization optimization, further improving the adaptability of the beauty effect.

[0070] Further, step S6 includes: Step S61: executing the beauty preview of the multi-mode shooting optimization strategy according to the preview interface, and receiving multi-mode preview feedback data.

[0071] Step S62: Correcting the multi-mode shooting optimization strategy according to the multi-mode preview feedback data to obtain a multi-mode shooting correction strategy.

[0072] Step S63: performing multi-mode switching preference prediction according to the shooting feature scene to obtain a multi-mode switching preference prediction result, and adjusting the multi-mode shooting correction strategy according to the multi-mode switching preference prediction result to obtain the multi-mode switching adjustment strategy.

[0073] Specifically, on the preview interface of the beauty camera, the beauty preview is performed according to the settings in the multi-mode shooting optimization strategy, and the beauty effect and other pictures when shooting according to the strategy are displayed to the user through the preview interface. At the same time, the multi-mode preview feedback data provided by the user through the interactive interface (such as touch operation, button operation, etc.) is received. These preview feedback data include user evaluation or adjustment information on beauty effects, shooting modes, picture quality, etc. For example, the user can adjust the beauty intensity through the slider on the preview interface and feedback whether the current beauty effect is satisfactory, or switch shooting modes and feedback the experience in different modes.

[0074] The multi-mode shooting optimization strategy is corrected according to the received multi-mode preview feedback data. If the feedback data indicates that the beautification is excessive, the beautification parameters are reduced in the multi-mode shooting optimization strategy; if the feedback indicates that the color effect in a certain mode is not good, the color correction parameters in the mode are adjusted, etc. By analyzing the feedback data, the various parameters in the multi-mode shooting optimization strategy are modified in a targeted manner, thereby obtaining a multi-mode shooting correction strategy.

[0075] Based on the shooting feature scene, the multi-mode switching preference prediction is performed using the relevant prediction algorithm. The multi-mode switching preference prediction can use machine learning algorithms, such as decision tree algorithms or neural network algorithms, which establish prediction models by learning a large amount of shooting scenes and user behavior data. Exemplarily, the algorithm can be trained based on a large amount of shooting scenes and user behavior data. For example, if the shooting scene is an outdoor scenery with sufficient light, based on the previous user behavior data in similar scenes, it is predicted that the user may be more inclined to use the landscape mode and may adjust parameters such as color saturation. Then, the multi-mode shooting correction strategy is adjusted according to the multi-mode switching preference prediction results to obtain a multi-mode switching adjustment strategy. For example, if it is predicted that the user will switch to the landscape mode, the parameters in the multi-mode shooting correction strategy are adjusted to adapt to the optimized shooting of the landscape mode, thereby obtaining a multi-mode switching adjustment strategy.

[0076] The shooting strategy is further optimized through feedback adjustment and multi-mode switching preference prediction, which not only ensures that the beautification effect can be adjusted in real time and meet user expectations, but also further improves the adaptability of the beautification effect and user experience by intelligently predicting user preferences.

[0077] Furthermore, the steps of constructing the beauty requirement analysis model in step S3 include: Step A: Obtain a shooting feature scene sample set and a beauty feature requirement sample set.

[0078] Step B: using the shooting feature scene sample set as input information and the beauty feature requirement sample set as output information, training the BP neural network, and obtaining the beauty requirement analysis loss coefficient after each predetermined number of trainings.

[0079] Step C: if the beauty requirement analysis loss coefficient is less than the beauty requirement analysis loss threshold, the beauty requirement analysis model is generated.

[0080] Specifically, the shooting feature scene sample set is a collection of a series of shooting scene features, and these samples contain various feature information of different shooting scenes. For example, the samples may include different lighting conditions (such as strong light, weak light, backlight, etc.), the distribution of people or objects in the scene (such as single-person close-up, multi-person group photo, large-scene scenery, etc.), the color characteristics of the scene (such as bright flower sea scene, monotonous indoor scene, etc.), and other information. The beauty feature demand sample set is a collection of beauty effect requirements in different shooting scenes corresponding to the shooting feature scene sample set. For example, in a single-person close-up scene under strong light, the possible beauty feature requirements are slight skin smoothing to reduce oiliness, moderate three-dimensional beautification of facial features, etc.; in a multi-person group photo scene, uniform skin color adjustment, natural beauty effect, etc. may be required. Collect the actual shooting data of different users in various shooting scenes, analyze and annotate these data, extract the shooting feature scenes and the corresponding beauty feature requirements, and thus construct the shooting feature scene sample set and the beauty feature demand sample set.

[0081] The shooting feature scene sample set is input as input information into the BP neural network, and the beauty feature requirement sample set is used as the expected output information. Then the network training begins. During the training process, the shooting feature scene information input during forward propagation is calculated by each layer of neurons in the network to obtain the output result. This output result is then compared with the actual result in the beauty feature requirement sample set, and the weight in the network is adjusted according to the difference back propagation error. After each predetermined number of trainings (for example, every 100 trainings), the beauty requirement parsing loss coefficient between the current BP neural network output and the beauty feature requirement sample set is calculated. The parsing loss coefficient is used to measure the error degree between the current network output result and the actual beauty feature requirement, reflecting the accuracy of the BP neural network in parsing the beauty requirement.

[0082] During the training process, it is constantly checked whether the beauty requirement analysis loss coefficient is less than the beauty requirement analysis loss threshold. This beauty requirement analysis loss threshold is a pre-set value used to determine whether the beauty requirement analysis loss coefficient has reached an acceptable level. If the beauty requirement analysis loss coefficient is less than this threshold at a certain training stage, it is considered that the BP neural network has been able to analyze the beauty requirement well. At this time, the training is stopped and the trained network is output as a beauty requirement analysis model. This model can predict beauty requirements in the subsequent shooting process, thereby providing a basis for the formulation of shooting strategies, etc.

[0083] Through the above model training steps, the beauty requirement analysis model can learn the beauty requirements in different shooting scenarios, providing technical support for the intelligent beauty requirement analysis of the beauty camera.

[0084] In summary, the parameter adaptive adjustment method for multi-mode switching of a beauty camera provided in the embodiment of the present application has the following technical effects: By obtaining the user's shooting instructions, including shooting target and shooting environment information, basic data is provided for subsequent feature detection and beauty demand analysis, ensuring that the technical solution can be targeted according to the user's actual needs and shooting scenes. Feature detection is performed on the shooting target to generate the feature point cloud of the shooting target, and these feature point clouds are fused with the shooting environment information to generate shooting feature scenes, providing comprehensive and accurate input data for beauty demand analysis, so that the subsequent beauty demand analysis can more accurately reflect the user's actual needs and the characteristics of the shooting scene. The shooting feature scene is input into the beauty demand analysis model to intelligently identify the user's beauty needs, providing a basis for subsequent multi-mode shooting decisions, avoiding universal and fixed beauty settings, and improving the targeting and personalization of beauty. According to the shooting feature scene and beauty feature requirements, a multi-mode shooting decision space is established, and multiple possible decision plans are determined for the shooting mode of the beauty camera. Through optimization analysis, the shooting strategy that best suits the current shooting scene and beauty needs is automatically selected, avoiding the tedious operation of users manually switching modes, while improving the beauty effect. Through the preview interface of the beauty camera, real-time feedback and correction are performed on the multi-mode shooting optimization strategy, and a multi-mode shooting correction strategy is generated. Through multi-mode switching preference prediction, the beauty strategy is further adjusted and a multi-mode switching adjustment strategy is generated. Real-time optimization and adaptive switching of the beauty effect are achieved, so that the final adjustment strategy can better adapt to the actual shooting effect, meet user expectations, and further enhance the user experience.

[0085] Overall, the embodiments of the present application achieve adaptive adjustment of the beauty camera in different shooting scenes through intelligent feature detection, data fusion, beauty demand analysis, multi-mode shooting decision-making and real-time feedback correction mechanism. This method can not only automatically select the optimal beauty mode according to the shooting target and environment, but also ensure the naturalness and consistency of the beauty effect through real-time feedback correction, reduce the need for manual adjustment by users, and significantly improve the intelligence level of the beauty camera and the user experience and satisfaction.

[0086] Embodiment 2, as Figure 4 As shown, based on the same inventive concept as the aforementioned embodiment 1, the embodiment of the present application provides a parameter adaptive adjustment system for multi-mode switching of a beauty camera, the system comprising: The shooting instruction acquisition module 10 is used to obtain the user shooting instruction of the beauty camera, wherein the user shooting instruction includes shooting target and shooting environment information.

[0087] The feature scene generation module 20 is used to perform feature detection according to the shooting target, obtain a shooting target feature point cloud, and perform data fusion on the shooting target feature point cloud and the shooting environment information to generate a shooting feature scene.

[0088] The beauty requirement parsing module 30 is used to input the shooting feature scene into the beauty requirement parsing model to obtain beauty feature requirements.

[0089] The shooting decision module 40 is used to make a multi-mode shooting decision for the beauty camera according to the shooting feature scene and the beauty feature requirement, and establish a multi-mode shooting decision space.

[0090] The optimization analysis module 50 is used to perform optimization analysis on the multi-mode shooting decision space based on the shooting feature scene and according to the beauty prediction evaluation model to determine the multi-mode shooting optimization strategy.

[0091] The switching adjustment decision module 60 is used to perform beauty preview feedback correction on the multi-mode shooting optimization strategy according to the preview interface of the beauty camera, generate a multi-mode switching adjustment strategy, and perform adaptive switching adjustment of the beauty camera according to the multi-mode switching adjustment strategy.

[0092] Furthermore, the shooting decision module 40 in the embodiment of the present application is used to perform the following steps: A plurality of preset shooting modes of the beauty camera are obtained; shooting parameter constraint information of the plurality of preset shooting modes is collected to obtain a multi-mode shooting constraint domain; the multi-mode shooting constraint domain is optimized in association with the shooting feature scene and the beauty feature requirement to establish a multi-mode shooting optimization domain; shooting parameter decisions are made for the plurality of preset shooting modes according to the multi-mode shooting optimization domain to generate the multi-mode shooting decision space.

[0093] Furthermore, the shooting decision module 40 in the embodiment of the present application is also used to perform the following steps: Taking the shooting feature scene and the beauty feature requirement as shooting retrieval constraints, the shooting parameter records of the multiple preset shooting modes are retrieved to obtain a multi-mode shooting parameter retrieval set; data cleaning is performed according to the multi-mode shooting parameter retrieval set to obtain a multi-mode shooting record set; trigger interval calibration is performed according to the multi-mode shooting record set to generate a multi-mode shooting association domain; intersection recognition is performed on the multi-mode shooting constraint domain and the multi-mode shooting association domain to generate the multi-mode shooting optimization domain.

[0094] Furthermore, the optimization analysis module 50 of the embodiment of the present application is used to perform the following steps: According to the multi-mode shooting decision space, extract the first mode shooting decision space; based on the shooting feature scene, evaluate and optimize the first mode shooting decision space according to the beauty prediction evaluation model to obtain the first mode shooting decision optimization space; construct a beauty fitness evaluation network, wherein the beauty fitness evaluation network includes beauty naturalness weight, beauty transition naturalness weight and visual beautification quality weight; according to the beauty fitness evaluation network, optimize the first mode shooting decision space to maximize beauty fitness and obtain the first mode shooting optimization strategy; add the first mode shooting optimization strategy to the multi-mode shooting optimization strategy, and continue to optimize the multi-mode shooting decision space according to the shooting feature scene, the beauty prediction evaluation model and the beauty fitness evaluation network to generate the multi-mode shooting optimization strategy.

[0095] Furthermore, the optimization analysis module 50 in the embodiment of the present application is also used to perform the following steps: Extract the first shooting decision according to the first mode shooting decision space; input the shooting feature scene and the first shooting decision into the beauty prediction evaluation model to obtain a first beauty prediction evaluation result, wherein the beauty prediction evaluation model includes multidimensional beauty prediction evaluation indicators, and the multidimensional beauty prediction evaluation indicators include beauty naturalness, beauty transition naturalness and visual beautification quality; judge whether the first beauty prediction evaluation result meets the beauty prediction evaluation constraint; if the first beauty prediction evaluation result meets the beauty prediction evaluation constraint, add the first shooting decision to the first mode shooting decision optimization space; if the first beauty prediction evaluation result does not meet the beauty prediction evaluation constraint, eliminate the first shooting decision; based on the shooting feature scene and the beauty prediction evaluation model, continue to evaluate and optimize the first mode shooting decision space according to the beauty prediction evaluation constraint to obtain the first mode shooting decision optimization space.

[0096] Furthermore, the switching adjustment decision module 60 in the embodiment of the present application is used to perform the following steps: A beauty preview of the multi-mode shooting optimization strategy is performed according to the preview interface, and multi-mode preview feedback data is received; the multi-mode shooting optimization strategy is corrected according to the multi-mode preview feedback data to obtain a multi-mode shooting correction strategy; a multi-mode switching preference prediction is performed according to the shooting feature scene to obtain a multi-mode switching preference prediction result, and the multi-mode shooting correction strategy is adjusted according to the multi-mode switching preference prediction result to obtain the multi-mode switching adjustment strategy.

[0097] Furthermore, the system described in the embodiment of the present application further includes a demand analysis model construction module, and the demand analysis model construction module is used to perform the following steps: Obtain a shooting feature scene sample set and a beauty feature requirement sample set; use the shooting feature scene sample set as input information and the beauty feature requirement sample set as output information to train the BP neural network, and obtain a beauty requirement analysis loss coefficient after each predetermined number of trainings; if the beauty requirement analysis loss coefficient is less than a beauty requirement analysis loss threshold, generate the beauty requirement analysis model.

[0098] Through the detailed description of the parameter adaptive adjustment method for multi-mode switching of the beauty camera in the foregoing specification, those skilled in the art can clearly understand the parameter adaptive adjustment system for multi-mode switching of the beauty camera in this embodiment. For the system disclosed in Example 2, since it corresponds to the method disclosed in Example 1 and has corresponding functional modules and beneficial effects, the relevant parts can be referred to the description of the method part.

[0099] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A parameter adaptive adjustment method for multi-mode switching of a beauty camera, characterized in that: The method comprises: Obtaining a user shooting instruction of the beauty camera, wherein the user shooting instruction includes shooting target and shooting environment information; Performing feature detection according to the shooting target to obtain a shooting target feature point cloud, and performing data fusion on the shooting target feature point cloud and the shooting environment information to generate a shooting feature scene; Inputting the shooting feature scene into a beauty requirement parsing model to obtain beauty feature requirements; Performing a multi-mode shooting decision on the beauty camera according to the shooting feature scene and the beauty feature requirement, and establishing a multi-mode shooting decision space; Based on the shooting feature scene, performing optimization analysis on the multi-mode shooting decision space according to the beauty prediction evaluation model to determine the multi-mode shooting optimization strategy; The multi-mode shooting optimization strategy is corrected by beauty preview feedback according to the preview interface of the beauty camera, a multi-mode switching adjustment strategy is generated, and adaptive switching adjustment of the beauty camera is performed according to the multi-mode switching adjustment strategy.

2. The parameter adaptive adjustment method for multi-mode switching of a beauty camera according to claim 1, characterized in that: According to the shooting feature scene and the beauty feature requirement, a multi-mode shooting decision is made for the beauty camera, and a multi-mode shooting decision space is established, including: Obtaining multiple preset shooting modes of the beauty camera; Collecting shooting parameter constraint information of the plurality of preset shooting modes to obtain a multi-mode shooting constraint domain; According to the shooting feature scene and the beauty feature requirement, the multi-mode shooting constraint domain is optimized in association to establish a multi-mode shooting optimization domain; Shooting parameter decisions are made for the plurality of preset shooting modes according to the multi-mode shooting optimization domain to generate the multi-mode shooting decision space.

3. The parameter adaptive adjustment method for multi-mode switching of a beauty camera according to claim 2, characterized in that: The multi-mode shooting constraint domain is optimized according to the shooting feature scene and the beauty feature requirement to establish a multi-mode shooting optimization domain, including: Taking the shooting feature scene and the beauty feature requirement as shooting retrieval constraints, performing shooting parameter record retrieval on the multiple preset shooting modes to obtain a multi-mode shooting parameter retrieval set; Perform data cleaning according to the multi-mode shooting parameter retrieval set to obtain a multi-mode shooting record set; Perform trigger interval calibration according to the multi-mode shooting record set to generate a multi-mode shooting association domain; An intersection recognition is performed on the multi-mode shooting constraint domain and the multi-mode shooting association domain to generate the multi-mode shooting optimization domain.

4. The parameter adaptive adjustment method for multi-mode switching of a beauty camera according to claim 1, characterized in that: Based on the shooting feature scene, the multi-mode shooting decision space is optimized and analyzed according to the beauty prediction evaluation model to determine the multi-mode shooting optimization strategy, including: Extracting a first mode shooting decision space according to the multi-mode shooting decision space; Based on the shooting feature scene, evaluating and optimizing the first mode shooting decision space according to the beauty prediction evaluation model to obtain the first mode shooting decision optimization space; Constructing a beauty fitness evaluation network, wherein the beauty fitness evaluation network includes a beauty naturalness weight, a beauty transition naturalness weight, and a visual beautification quality weight; According to the beauty fitness evaluation network, the beauty fitness maximization optimization is performed on the first mode shooting decision space to obtain the first mode shooting optimization strategy; The first mode shooting optimization strategy is added to the multi-mode shooting optimization strategy, and the multi-mode shooting decision space is continued to be optimized and analyzed according to the shooting feature scene, the beauty prediction evaluation model and the beauty fitness evaluation network to generate the multi-mode shooting optimization strategy.

5. The parameter adaptive adjustment method for multi-mode switching of a beauty camera according to claim 4, characterized in that: Based on the shooting feature scene, the first mode shooting decision space is evaluated and optimized according to the beauty prediction evaluation model to obtain the first mode shooting decision optimization space, including: Extracting a first shooting decision according to the first mode shooting decision space; Inputting the shooting feature scene and the first shooting decision into the beauty prediction evaluation model to obtain a first beauty prediction evaluation result, wherein the beauty prediction evaluation model includes multi-dimensional beauty prediction evaluation indicators, and the multi-dimensional beauty prediction evaluation indicators include beauty naturalness, beauty transition naturalness, and visual beautification quality; Determining whether the first beauty prediction evaluation result satisfies the beauty prediction evaluation constraint; If the first beauty prediction evaluation result satisfies the beauty prediction evaluation constraint, adding the first shooting decision to the first mode shooting decision optimization space; If the first beauty prediction evaluation result does not satisfy the beauty prediction evaluation constraint, eliminating the first shooting decision; Based on the shooting feature scene and the beauty prediction evaluation model, the first mode shooting decision space is continuously evaluated and optimized according to the beauty prediction evaluation constraint to obtain the first mode shooting decision optimization space.

6. The parameter adaptive adjustment method for multi-mode switching of a beauty camera according to claim 1, characterized in that: The multi-mode shooting optimization strategy is corrected by beauty preview feedback according to the preview interface of the beauty camera to generate a multi-mode switching adjustment strategy, including: Executing the beauty preview of the multi-mode shooting optimization strategy according to the preview interface, and receiving multi-mode preview feedback data; Correcting the multi-mode shooting optimization strategy according to the multi-mode preview feedback data to obtain a multi-mode shooting correction strategy; A multi-mode switching preference prediction is performed according to the shooting feature scene to obtain a multi-mode switching preference prediction result, and the multi-mode shooting correction strategy is adjusted according to the multi-mode switching preference prediction result to obtain the multi-mode switching adjustment strategy.

7. The parameter adaptive adjustment method for multi-mode switching of a beauty camera according to claim 1, characterized in that: The steps of constructing the beauty requirement analysis model include: Obtain a sample set of shooting feature scenes and a sample set of beauty feature requirements; Taking the shooting feature scene sample set as input information and the beauty feature requirement sample set as output information, the BP neural network is trained, and after each training for a predetermined number of times, a beauty requirement analysis loss coefficient is obtained; If the beauty requirement analysis loss coefficient is less than the beauty requirement analysis loss threshold, the beauty requirement analysis model is generated.

8. A parameter adaptive adjustment system for multi-mode switching of a beauty camera, characterized in that: The system is used to execute the parameter adaptive adjustment method for multi-mode switching of a beauty camera according to any one of claims 1 to 7, comprising: A shooting instruction acquisition module is used to obtain a user shooting instruction of the beauty camera, wherein the user shooting instruction includes shooting target and shooting environment information; A feature scene generation module is used to perform feature detection according to the shooting target, obtain a feature point cloud of the shooting target, and perform data fusion on the feature point cloud of the shooting target and the shooting environment information to generate a shooting feature scene; A beauty requirement parsing module, used for inputting the shooting feature scene into a beauty requirement parsing model to obtain beauty feature requirements; A shooting decision module, used for making a multi-mode shooting decision for the beauty camera according to the shooting feature scene and the beauty feature requirement, and establishing a multi-mode shooting decision space; An optimization analysis module is used to perform optimization analysis on the multi-mode shooting decision space based on the shooting feature scene and according to the beauty prediction evaluation model to determine the multi-mode shooting optimization strategy; The switching adjustment decision module is used to perform beauty preview feedback correction on the multi-mode shooting optimization strategy according to the preview interface of the beauty camera, generate a multi-mode switching adjustment strategy, and perform adaptive switching adjustment of the beauty camera according to the multi-mode switching adjustment strategy.

Images