Gamification creation system and method based on machine learning

Abstract

The invention relates to the technical field of machine learning computing systems, in particular to a gamification creation system and method based on machine learning, comprising a sensing data acquisition module, a gamification data identification module, a machine learning creation module, a game creation output module and a real-time feedback optimization module. According to the method, the personalized game image is generated based on the gamification data, style migration and detail enhancement are supported, network delay monitoring and equipment computing power margin evaluation are introduced, creation parameters are dynamically adjusted, the model resolution is reduced or local cache generation is started, creation fluency is ensured, and the creation efficiency of virtual image gamification is improved; the high-popularity public image is identified through page view analysis, an output result is corrected in combination with a similarity coefficient, overlapping of the image and the public image is avoided, and the creation efficiency and effect of virtual image gamification are improved by forming a'data-creation-feedback 'closed loop.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning computing systems, and in particular to a gamification creation system and method based on machine learning. Background Art

[0002] Regarding computing power generation in gamification creation fields such as the metaverse that require avatars, the introduction of distributed rendering architectures and neural radiation field technology has significantly improved rendering efficiency in recent years. AI-driven reinforcement learning algorithms have further enabled intelligent optimization of rendering parameters. Combined with NVIDIA DLSS technology, this reduces computing power requirements while ensuring image quality. However, significant latency still exists when capturing expressions with low computing power. Existing solutions mostly use the WebRTC protocol or fog computing deployment, but still suffer from delays exceeding 20ms. Emerging non-invasive brain-computer interface technology directly converts neural signals into virtual expressions, breaking through the latency bottleneck of traditional sensors and providing a new path for real-time interaction. However, it still requires significant computing power. In the field of public image identification, although blockchain NFT technology has been used to authenticate the uniqueness of avatars, and biometric watermarking and multimodal behavior recognition technologies have enhanced the security of identity authentication, after gamification of avatars based on user needs, it is impossible to identify and distinguish public images while satisfying users' creative freedom.

[0003] Chinese Patent Publication No. CN116882482A discloses a method and apparatus for training an avatar generation model and generating an avatar. These methods involve computer vision, augmented reality, virtual reality, deep learning, and other artificial intelligence technologies, and are applicable to scenarios such as the metaverse and digital humans. However, this solution fails to match the computing power of the avatar production process and cannot optimize the delay in capturing facial expressions during avatar generation. Summary of the Invention

[0004] To this end, the present invention provides a gamification creation system and method based on machine learning to overcome the problems in the prior art of low efficiency and poor effect in the gamification creation of virtual images caused by computing power mismatch, network delay and failure to identify and distinguish public images.

[0005] To achieve the above objectives, the present invention provides, on the one hand, a gamification creation system based on machine learning, comprising:

[0006] A sensor data acquisition module is used to acquire the creation source data through a group of sensor nodes;

[0007] A gamification data identification module is used to identify the source data according to the data identification model to obtain gamification data and non-gamification data, and is also used to adjust the acquisition process of the source data according to the number of non-gamification data items of the same source;

[0008] A machine learning creation module, which is used to create game characters based on gamification data, monitor network latency, and optimize the creation process of game characters in real time based on network latency. It is also used to adjust the real-time optimization process of the game character creation process based on the device's computing power margin;

[0009] A game creation output module is used to output game characters, identify public characters based on the number of visits to the game characters, calibrate the output results of the game characters based on the public similarity coefficient of the game characters, and optimize the public character recognition process based on the basic subject ratio of the game characters;

[0010] The real-time feedback optimization module is used to optimize the acquisition process of creative source data in real time according to the frequency of gamification data points.

[0011] Furthermore, the gamified data recognition module sets the loss function of the data recognition model to be a binary cross entropy function, N is the total number of samples in the data recognition test set, i Identify the sample number in the test set for the data, y i Identify the actual category of the i-th sample in the test set for the data. For the gamified data y i =1, for non-gamified data y i =0, is the predicted probability of the data recognition model, which indicates the predicted probability that the i-th sample is gamification data.

[0012] Furthermore, the gamification data identification module obtains the sensor node identifier of the non-gamification data, and counts the non-gamification data with the same sensor node identifier within the cluster monitoring period to obtain the number n of non-gamification data of the same source, and compares the number n of non-gamification data of the same source with the preset number n0 of non-gamification data of the same source, and judges the clustering status of the non-gamification data based on the comparison result.

[0013] Furthermore, the machine learning creation module preprocesses the gamification data to obtain preprocessed gamification data, and performs feature selection on the preprocessed gamification data to obtain gamification feature data, and selects a machine learning model for the gamification data according to modeling requirements to obtain an adaptive creation model.

[0014] Furthermore, the machine learning creation module sends an ICMP request to the target address, measures the round-trip time of the ICMP request, obtains the current network delay A, and compares the current network delay A with the maximum network delay Amax and the minimum network delay Amin, Amax ≥ 150ms, Amin ≤ 50ms, judges the network delay situation based on the comparison results, and optimizes the game image creation process in real time based on the judgment results.

[0015] Furthermore, the machine learning creation module obtains the current GPU real-time utilization rate z, and calculates the device computing power margin β based on the current GPU real-time utilization rate z and the theoretical maximum value zmax, compares the device computing power margin β with the preset device computing power margin β0, judges the current device computing power margin situation based on the comparison result, and adjusts the real-time optimization process of the game image creation process based on the judgment result.

[0016] Furthermore, the game creation output module obtains the game image of the machine learning creation module and outputs the game image. The game creation output module obtains the number of visits K of the network image through network retrieval, compares the number of visits K with the preset number of visits K0, K0 ≥ 10,000 times, and judges the category of the network image based on the comparison result.

[0017] Furthermore, the game creation output module inputs the marking feature data and the gamification feature data into a similarity comparison model, outputs the public similarity coefficient Q of the game image, and compares the public similarity coefficient Q of the game image with the preset similarity coefficient Q0, Q0≥70%, judges the public similarity degree of the game image based on the comparison result, and corrects the output result of the game image based on the judgment result.

[0018] Furthermore, the game creation output module inputs the game image into the basic body proportion model and outputs the basic body proportion, wherein the basic body proportion includes the basic body proportion r1 and the decoration proportion r2, and r1+r2=100% is set. The basic body proportion r1 is compared with the preset body proportion r0, r0≥65%, and the actual situation of the game image is judged based on the comparison result, and the public image recognition process is optimized based on the judgment result.

[0019] In another aspect, the present invention further provides a gamification creation method based on machine learning, comprising:

[0020] Step S1, acquiring the creation source data through the sensor node group;

[0021] Step S2: The gamification data identification module identifies the creative source data according to the data identification model to obtain gamification data and non-gamification data. The module also adjusts the acquisition process of the creative source data according to the number of non-gamification data items of the same source.

[0022] Step S3: Creating a game character based on the gamification data, monitoring network latency, and optimizing the game character creation process in real time based on the network latency. Furthermore, optimizing the adjustment method for obtaining the source data based on the real-time optimization method, and adjusting the real-time optimization process of the game character creation process based on the computing power margin of the device.

[0023] Step S4: Output the game avatar, identify the public image based on the number of visits to the game avatar, correct the output of the game avatar based on the public similarity coefficient of the game avatar, and optimize the public image identification process based on the basic subject ratio of the game avatar;

[0024] Step S5: Optimize the acquisition process of the creative source data in real time according to the frequency of the gamification data points.

[0025] Compared with the prior art, the beneficial effect of the present invention is that the system dynamically adjusts the collection strategy through data recognition models and non-gamified data, and combines the number of homologous non-gamified data to reduce redundant data interference and improve data quality. The system generates personalized game images based on gamified data through a machine learning creation module, supports style transfer and detail enhancement, introduces network delay monitoring and device computing power margin assessment, dynamically adjusts creation parameters, reduces model resolution or enables local cache generation to ensure creation fluency, and improves the efficiency of virtual image gamification creation. The system also identifies highly popular public images through visit volume analysis, corrects the output results in combination with similarity coefficients to avoid overlap between images and public images, optimizes public image recognition logic based on the proportion of basic subjects, avoids excessive homogenization, retains the creator's style, and improves the creation effect of virtual image gamification. The system also monitors the frequency of gamification data points and optimizes data collection strategies in real time to form a "data-creation-feedback" closed loop, thereby improving the efficiency and effect of virtual image gamification creation. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a schematic diagram of the structure of the gamification creation system based on machine learning in this embodiment;

[0027] Figure 2 Schematic diagram of the flow of the gamification creation method based on machine learning in this embodiment. DETAILED DESCRIPTION

[0028] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0029] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0030] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0031] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0032] See also Figure 1 As shown in FIG, which is a schematic diagram of the structure of the gamification creation system based on machine learning in this embodiment, the system includes:

[0033] A sensor data acquisition module is used to acquire the creation source data through a group of sensor nodes;

[0034] A gamification data identification module, configured to identify the source data according to the data identification model to obtain gamification data and non-gamification data, and to adjust the acquisition process of the source data according to the number of non-gamification data items of the same source. The gamification data identification module is connected to the sensor data acquisition module;

[0035] A machine learning creation module, configured to create game characters based on gamification data, monitor network latency, optimize the game character creation process in real time based on network latency, and adjust the real-time optimization of the game character creation process based on device computing power margin. The machine learning creation module is connected to the gamification data recognition module.

[0036] A game creation output module is used to output the game image, identify the public image based on the number of visits to the game image, correct the output result of the game image based on the public similarity coefficient of the game image, and optimize the public image identification process based on the basic subject ratio of the game image. The game creation output module is connected to the machine learning creation module;

[0037] The real-time feedback optimization module is used to optimize the acquisition process of the creation source data in real time according to the frequency of the gamification data points. The real-time feedback optimization module is connected to the game creation output module.

[0038] Specifically, the system is set up in a gamified creation terminal for an avatar. It acquires source data through a group of sensor nodes on the gamified creation terminal and creates and outputs a game avatar based on the source data. The system dynamically adjusts the collection strategy based on data recognition models and non-gamified data, combined with the number of homologous non-gamified data items, to reduce redundant data interference and improve data quality. The system generates personalized game avatars based on gamified data through a machine learning creation module, supports style transfer and detail enhancement, introduces network latency monitoring and device computing power margin assessment, dynamically adjusts creation parameters, reduces model resolution, or enables local cache generation to ensure smooth creation and improve the efficiency of gamified avatar creation. The system also identifies highly popular public images through visit volume analysis, corrects output results based on similarity coefficients to avoid overlap between images and public images, optimizes public image recognition logic based on the proportion of basic entities, avoids excessive homogeneity, preserves the creator's style, and improves the effectiveness of gamified avatar creation. The system also monitors the frequency of gamified data points to optimize data collection strategies in real time, forming a "data-creation-feedback" closed loop to improve the efficiency and effectiveness of gamified avatar creation.

[0039] Specifically, the sensor data acquisition module is connected to the sensor node group through a hybrid connection method, and acquires the creative source data through an event-driven method at a preset acquisition frequency P0. The sensor node group refers to a collection of sensor devices distributed at various positions of the wearable device. This embodiment does not limit the setting position of the sensor node group. Those skilled in the art can freely set it according to actual conditions, as long as it meets the sufficient acquisition requirements of various positions of the wearable device. For example, it can be set on the head of the wearable device, and an infrared imaging sensor facing the wearer's face is set. The creative source data refers to the original data collected by the sensor node group for game image creation. The creative source data is marked with the sensor node source as the sensor node identifier. This embodiment does not limit the marking method of the sensor node identifier. Those skilled in the art can mark the creative source data identifier according to actual conditions, such as setting the sensor device identification code as a suffix to the creative source data. This embodiment does not limit the specific type of the creative source data. Those skilled in the art can freely set it according to actual conditions, such as setting the creative source data to include visual data, audio data, motion data, environmental data and wearer behavior data. The visual data The data refers to images and videos collected by visual sensors, used to capture the wearer's facial expressions, movements, and environmental scenes. The audio data refers to sounds collected by microphones, used to analyze the wearer's voice commands and environmental sounds. The motion data refers to the wearer's motion information collected by sensors such as accelerometers and gyroscopes, used to capture the wearer's movements or postures. The environmental data refers to environmental information collected by sensors such as temperature, humidity, and light, used to adjust the performance of game scenes or characters. The wearer behavior data refers to wearer interaction data collected by interactive sensors, such as clicks and slides, used to analyze the wearer's habits and preferences. The hybrid connection mode refers to a connection scheme that combines wireless and wired connections, selected based on node characteristics. This embodiment does not limit the content of the hybrid connection mode. Those skilled in the art can freely configure it according to actual circumstances, as long as it meets the requirements for obtaining the original source data. For example, Wi-Fi can be configured for wireless connection and Ethernet for wired connection. The event-driven mode refers to each sensor node in the sensor node group actively sending a signal to the sensor data acquisition module when detecting a specific event. For example, when there is a fluctuation in the wearer's facial expression data in the visual data, the visual data can be configured to send the visual data to the sensor data acquisition module.

[0040] Specifically, the gamified data recognition module selects a convolutional neural network model as the basic framework of the data recognition model, and divides 70% of the data recognition learning sample data set into a data recognition training set, 15% of the data recognition learning sample data set into a data recognition verification set, and 15% of the data recognition learning sample data set into a data recognition test set. The convolutional neural network model is trained according to the data recognition training set, and the weight of the convolutional neural network model is updated through the backpropagation algorithm. After each epoch, the convolutional neural network model is verified using the data recognition verification set to obtain a recognition verification loss value and a recognition verification accuracy rate. When the verification loss value and the verification accuracy rate meet the preset verification conditions, the convolutional neural network model is tested according to the data recognition test set to obtain a recognition test accuracy rate. When the recognition test accuracy rate reaches the preset accuracy rate, the convolutional neural network model is output as the data recognition model. The gamified data recognition module converts the creation source data into image form and inputs it into the data recognition model, and obtains the data recognition result output by the data recognition model, which includes gamified data and non-gamified data.

[0041] Specifically, the data recognition learning sample data set refers to a learning data set stored in the form of creation source data - data recognition results, which is used to train the convolutional neural network model. The epoch refers to the process in which the convolutional neural network model completes a forward propagation and a backward propagation on the data recognition training set. The verification loss value refers to the loss value of the loss function in the convolutional neural network model when the data recognition verification set is input into the convolutional neural network model for verification. The verification accuracy rate refers to the ratio of the number of data recognition results output after the data recognition verification set is input into the convolutional neural network model that are consistent with the data recognition results in the data recognition verification set to the total number of samples in the data recognition verification set. The preset verification condition means that the verification loss value of 5 consecutive epochs has not decreased and the verification accuracy has not improved. The recognition test accuracy refers to the ratio of the number of data test results output after the data recognition test set is input into the convolutional neural network model and the data recognition results in the data recognition test set are consistent with the total number of samples in the data recognition test set. The preset accuracy refers to the preset value of the recognition test accuracy for judging whether the convolutional neural network model meets the output standard. For example, the preset accuracy can be set to 98%. The gamified data refers to the data in the creative source data used to create game characters, and the non-gamified data refers to the data in the creative source data that is not used to create game characters.

[0042] It is understandable that this embodiment does not limit the way in which the creative source data is converted into a picture form. Those skilled in the art can limit it according to actual conditions, and they only need to meet the feature expression requirements of the creative source data. For example, it can be set to extract features from visual data through a convolution layer, adjust the image in the visual data to a fixed size, such as 224x224, and perform normalization processing, convert the audio data into Mel-frequency cepstral coefficients MFCC, and treat it as an image input into the CNN, reconstruct the motion data, environmental data and wearer behavior data into images in time series, or use a one-dimensional convolutional network to process it, and process the environmental data through a simple fully connected layer, use feature engineering to extract the features of the wearer behavior data, and combine it into the convolutional neural network model.

[0043] Specifically, the gamified data recognition module sets the loss function of the data recognition model to be a binary cross entropy function, and sets the binary cross entropy function to be N is the total number of samples in the data recognition test set, i Identify the sample number in the test set for the data, y i Identify the actual category of the i-th sample in the test set for the data. For the gamified data y i =1, for non-gamified data y i =0, is the predicted probability of the data recognition model, which indicates the predicted probability that the i-th sample is gamification data.

[0044] Specifically, the gamification data identification module obtains the sensor node identifier of the non-gamification data, and counts the non-gamification data with the same sensor node identifier within the cluster monitoring period to obtain the number n of non-gamification data of the same source, and compares the number n of non-gamification data of the same source with the preset number n0 of non-gamification data of the same source, and determines the clustering state of the non-gamification data based on the comparison result, wherein:

[0045] When n≤n0, the gamification data identification module determines that the clustering state of the non-gamification data is not clustered;

[0046] When n>n0, the gamification data identification module determines that the clustering state of the non-gamification data is clustering;

[0047] The gamification data identification module adjusts the acquisition process of the creation source data according to the clustering status of the non-gamification data, wherein:

[0048] If the clustering status of the non-gamification data is not clustered, the gamification data identification module does not adjust the acquisition process of the creation source data;

[0049] If the clustering state of the non-gamification data is clustering, the gamification data identification module adjusts the acquisition process of the creation source data, and adjusts the preset acquisition frequency P0 of the sensor node corresponding to the non-gamification data according to the clustering adjustment coefficient α. The adjusted preset acquisition frequency is Pa0, and Pa0 is set to α×P0, and 1<α<1.2 is set.

[0050] Specifically, the preset acquisition frequency P0 refers to the frequency of acquiring the creative source data in an event-driven manner in the sensor data acquisition module. This embodiment does not specifically limit the value of the preset acquisition frequency P0. Those skilled in the art can freely set it according to actual conditions, as long as it meets the sufficient acquisition requirements of the creative source data. For example, the preset acquisition frequency P0 can be set to 1s / time. The cluster monitoring period refers to monitoring whether the clustering state of the non-gamified data reaches the preset period length of the cluster. For example, the cluster monitoring period can be set to 48 hours. The same sensor node identifier refers to the same character expression content of the sensor node identifier. This embodiment does not set the cluster monitoring period. The content of the counting method for non-gamified data with the same sensor node identifier within a period is specifically limited. Those skilled in the art can freely set it according to actual conditions, as long as the counting requirements for non-gamified data with the same sensor node identifier are met. For example, it can be set to count non-gamified data with the same character expression content of the sensor node identifier as one piece. The preset number of same-source non-gamified data refers to a preset value of the number of same-source non-gamified data that reflects the clustering state of the non-gamified data to achieve clustering. For example, the preset number of same-source non-gamified data n0 can be set to 30. The sensor node corresponding to the non-gamified data refers to the sensor node that serves as the source of the non-gamified data.

[0051] Specifically, the machine learning creation module preprocesses the gamification data to obtain preprocessed gamification data, performs feature selection on the preprocessed gamification data to obtain gamification feature data, selects a machine learning model for the gamification data according to modeling requirements, and obtains an adaptive creation model, wherein:

[0052] When the modeling requirement is to create a new game character, the machine learning creation module selects a generative adversarial network model as the adaptive creation model;

[0053] When the modeling requirement is to reconstruct a game image, the machine learning creation module selects a variational autoencoder model as the adaptive creation model;

[0054] The machine learning creation module inputs the gamification feature data into the adaptive creation model for creation, and obtains the game image output by the adaptive creation model.

[0055] Specifically, this embodiment does not limit the method for preprocessing the gamification data. Those skilled in the art can freely set it according to actual conditions, as long as the preprocessing requirements of the gamification data are met. For example, this embodiment converts the gamification data into a unified JSON format to obtain gamification data in a unified format, inputs the gamification data in the unified format into a multimodal learning framework to obtain fused feature gamification data, and uses the fused feature gamification data as preprocessed gamification data. This embodiment performs feature selection on the preprocessed gamification data according to the adaptive creation model, wherein:

[0056] When the machine learning creation module selects a generative adversarial network model as the adaptive creation model, this embodiment converts non-image features in the preprocessed gamification data into image channels to obtain image channel gamification data, aligns the image channel gamification data using a cross-modal attention CLIP style, and performs weighted feature extraction to obtain gamification feature data.

[0057] When the machine learning creation module selects the variational autoencoder model as the adaptive creation model, this embodiment performs semantic group coding on the preprocessed gamification data to obtain semantic group coded gamification data, and isolates sensitive attributes of the semantic group coded gamification data to obtain gamification feature data;

[0058] The modeling requirements refer to the requirements for constructing game images, including creating new game images and reconstructing game images. This embodiment does not limit the method for obtaining modeling requirements. Those skilled in the art can freely set them according to actual conditions, and only need to meet the actual reflection requirements of the modeling requirements. For example, the modeling requirements can be obtained through the user's operation path in the stay interface. The machine learning model of gamification data refers to the infrastructure model of the adaptive creation model. The adaptive creation model is a machine learning model with gamification feature data as input and game images as output, including a generative adversarial network model and a variational autoencoder model. This embodiment does not limit the content of the generative adversarial network model. It only needs to meet the generation requirements of the game image. For example, the gamification feature data can be set to be normalized as an RGB image and then input into the generative adversarial network model for training and game image generation. This embodiment does not limit the content of the variational autoencoder model. It only needs to meet the generation requirements of the game image. For example, the gamification feature data can be set to be Gaussian distributed and then input into the generative adversarial network model for training and game image generation.

[0059] Specifically, the machine learning creation module sends an ICMP request to the target address, measures the round-trip time of the ICMP request, obtains the current network delay A, and compares the current network delay A with the maximum network delay Amax and the minimum network delay Amin, where Amax is ≥ 150ms and Amin is ≤ 50ms. The network delay is judged based on the comparison results, and the game character creation process is optimized in real time based on the judgment results, wherein:

[0060] When A<Amin, the machine learning creation module determines that the network delay is excellent and does not perform real-time optimization on the creation process of the game character;

[0061] When Amin≤A≤Amax, the machine learning creation module determines that the network delay is good and does not perform real-time optimization on the creation process of the game character;

[0062] When A>Amax, the machine learning creation module determines that the network delay is network congestion, and optimizes the game image creation process in real time, extracts historical gamification feature data of historical game images, and uses the gamification feature that appears most frequently in the historical gamification feature data as creation habit data to supplement the gamification feature data currently input into the adaptive creation model, and obtains the game image output by the adaptive creation model, wherein the proportion a1 of the creation habit data input into the adaptive creation model is 30%, and the proportion b1 of the gamification feature data is 70%.

[0063] Specifically, the ICMP request refers to a data packet type used for network diagnosis and status reporting in the Internet Control Message Protocol, and is used to monitor network delay. The maximum network delay Amax refers to the maximum preset value used to judge the network delay situation. The minimum network delay Amin refers to the minimum preset value used to judge the network delay situation. The network delay situation refers to a situation used to describe the quality of the network delay. The network delay situation includes the network delay situation of excellent network, the network delay situation of good network and the network delay situation of congested network. The historical game image refers to the game image created historically, and the historical gamification feature data refers to the gamification feature data of the historical game image.

[0064] Specifically, the machine learning creation module obtains the current network status by monitoring the current network delay, and optimizes the game image creation process in real time according to the current network status. When optimizing the game image creation process in real time, the creation habit data is input into the adaptive creation model together with the gamification feature data as a supplement, and the game image is quickly produced according to the user's historical habits, thereby reducing the waiting time for the user to produce the game image.

[0065] Specifically, the machine learning creation module obtains the current GPU real-time usage rate z, and calculates the device computing power margin β based on the current GPU real-time usage rate z and the theoretical maximum value zmax, setting β=(1-z / zmax), and compares the device computing power margin β with the preset device computing power margin β0. Based on the comparison result, the current device computing power margin is judged, and the real-time optimization process of the game image creation process is adjusted based on the judgment result, wherein:

[0066] When β≥β0, the machine learning creation module determines that the computing power margin of the current device is sufficient, and does not adjust the real-time optimization process of the game character creation process;

[0067] When β<β0, the machine learning creation module determines that the current device computing power margin is insufficient, and adjusts the real-time optimization process of the game image creation process according to the adjustment coefficient w=1.3-0.3×e -(β0-β) The maximum network delay Amax is adjusted, e is the base of the natural logarithm, and the adjusted maximum network delay Amax' is obtained. Amax'=W×Amax is set, and the maximum network delay Amax is replaced by the adjusted maximum network delay Amax' to re-judge the network delay situation.

[0068] Specifically, the current GPU real-time utilization rate refers to the usage ratio of the GPU under the current circumstances. The GPU refers to a processor specially designed for efficiently processing graphics rendering and parallel computing tasks. The Chinese name of the GPU is graphics processing unit. The device computing power margin refers to a margin used to describe the usage ratio of the GPU. The preset device computing power margin refers to a preset value used to judge the current device computing power margin situation. The current device computing power margin situation refers to a description of the device computing power margin situation. The current device computing power margin situation includes the current device computing power margin situation being insufficient computing power and the current device computing power margin situation being sufficient computing power.

[0069] Specifically, the machine learning creation module compares the device computing power margin β with the preset device computing power margin β0. If β is less than β0, the adjustment coefficient w=1.3-0.3×e -(β0-β) Adjust the maximum network delay Amax to increase the maximum network delay and avoid misjudgment of network delay due to insufficient computing power.

[0070] Specifically, the game creation output module obtains the game image of the machine learning creation module and outputs the game image.

[0071] Specifically, the game creation output module obtains the number of visits K of the online image through network search, compares the number of visits K with the preset number of visits K0, where K0 is ≥ 10,000 times, and determines the category of the online image based on the comparison result, wherein:

[0072] When K<K0, the game creation output module determines that the category of the network image is an ordinary image;

[0073] When K≥K0, the game creation output module determines that the image category of the network image is a public image, and marks the feature data of the network image to obtain marked feature data.

[0074] Specifically, the number of visits refers to the total number of times the network image is clicked to visit, the network image refers to the character image that appears on the network, the preset number of visits refers to the preset value used to judge the category of the network image, the category of the network image refers to the classification of the network image, the category of the network image includes the category of the network image being an ordinary image and the image category of the network image being a public image, the public image refers to the image of a well-known character, the characteristic data of the network image refers to the image parameters obtained by scanning the network image through 3D modeling software, and the image parameters include facial organ parameters and face size parameters.

[0075] Specifically, the game creation output module classifies the network images through network retrieval so as to distinguish the public images, which facilitates the game creation output module to correct the output game images.

[0076] Specifically, the game creation output module inputs the marked feature data and the gamification feature data into the similarity comparison model, outputs the public similarity coefficient Q of the game image, and compares the public similarity coefficient Q of the game image with the preset similarity coefficient Q0, Q0 ≥ 70%, judges the public similarity of the game image based on the comparison result, and corrects the output result of the game image based on the judgment result, wherein:

[0077] When Q<Q0, the game creation output module determines that the public similarity of the game image is not similar, and does not correct the output result of the game image;

[0078] When Q≥Q0, the game creation output module determines that the public similarity of the game image is similar, and corrects the output result of the game image, uses the marked feature data as the error training set to retrain the adaptive creation model, and outputs the corrected game image. The game creation output module replaces the game image with the corrected game image for output.

[0079] Specifically, the similarity comparison model refers to a neural network learning model that takes the marked feature data and the gamification feature data as input and outputs the public similarity coefficient of the game character. The similarity comparison model is constructed by a similarity comparison model construction method, which includes:

[0080] 70% of the historical comparison training set is divided into a simulated training set and 30% of the historical comparison training set is divided into a simulated validation set. A recurrent neural network model is selected as the neural network architecture of the similarity comparison model. The Adam optimizer and cross-entropy loss function are selected to train the recurrent neural network model. The simulated training set is loaded into the recurrent neural network model. The forward propagation is performed through the recurrent neural network model to calculate the output value of the similarity comparison model. The loss function value is calculated based on the output value of the recurrent neural network model and the true value. The gradient is calculated through the backpropagation algorithm, and the weights and bias of the recurrent neural network model are updated. The process of forward propagation, loss calculation and backpropagation is repeated until the preset training rounds are reached. The accuracy of the recurrent neural network model is verified through the simulated validation set, and the recurrent neural network model with an accuracy rate of 90% is output as the similarity comparison model.

[0081] The historical comparison training set refers to a training set used to train a similarity comparison model. The historical comparison training set is obtained by historical marking feature data and historical gamification feature data. The public similarity coefficient refers to a specific numerical value of the similarity between the game image and the public image. The preset similarity coefficient refers to a preset value for judging the public similarity of the game image. The public similarity of the game image refers to the similarity between the game image and the public image. The public similarity of the game image includes the public similarity of the game image being dissimilar and the public similarity of the game image being similar.

[0082] Specifically, the game creation output module compares the public similarity coefficient of the game image output by the similarity comparison model with the preset similarity coefficient. If Q is greater than Q0, the output result of the game image is corrected to avoid disputes caused by overlap between the game image and the public image.

[0083] Specifically, the game creation output module inputs the game image into the basic body proportion model and outputs the basic body proportion, wherein the basic body proportion includes the basic body proportion r1 and the decoration proportion r2, and r1+r2=100%. The basic body proportion r1 is compared with the preset body proportion r0, and r0 ≥ 65%. The actual situation of the game image is judged based on the comparison result, and the public image recognition process is optimized based on the judgment result, wherein:

[0084] When r1≤r0, the game creation output module determines that the real situation of the public game image is not real and does not optimize the recognition process of the public image;

[0085] When r1>r0, the game creation output module determines that the real situation of the game image is real, and optimizes the recognition process of the public image, according to the recognition optimization coefficient γ=1+0.4× e-(r1-r0) The preset access amount K0 is optimized, e is the base of the natural logarithm, and the optimized preset access amount K0' is obtained. The optimized preset access amount K0' is set to K0×γ, and the preset access amount K0 is replaced by the optimized preset access amount K0'.

[0086] Specifically, the basic subject proportion model refers to a neural network learning model that takes game characters as input and basic subject proportions as output. The basic subject proportion model is constructed using a basic subject proportion model construction method, which includes:

[0087] 70% of the basic subject proportion parsing data set is divided into a parsing training set, 15% of the basic subject proportion parsing data set is divided into a parsing verification set, and 15% of the basic subject proportion parsing data set is divided into a parsing test set. A convolutional neural network model is trained according to the basic subject proportion parsing data set, and the weight of the convolutional neural network model is updated by a back-propagation algorithm. After each epoch, the convolutional neural network model is verified using the parsing verification set to obtain a recognition verification loss value and a recognition verification accuracy rate. When the verification loss value and the verification accuracy rate meet preset verification conditions, the convolutional neural network model is tested according to the parsing test set to obtain a parsing test accuracy rate. When the parsing test accuracy rate reaches a preset accuracy rate, the convolutional neural network model is output as a basic subject proportion parsing model.

[0088] The basic subject proportion analysis data set refers to a learning data set used to train the basic subject proportion analysis model, the epoch refers to the process in which the convolutional neural network model completes a forward propagation and a backward propagation on the data recognition training set, the verification loss value refers to the loss value of the loss function in the convolutional neural network model when the data recognition verification set is input into the convolutional neural network model for verification, the verification accuracy rate refers to the ratio of the number of parsing results output after the data recognition verification set is input into the convolutional neural network model that are consistent with the parsing results in the parsing verification set to the total number of samples in the parsing verification set, the preset verification condition refers to the verification loss value of 7 consecutive epochs that has not decreased and the verification accuracy rate has not increased, the parsing test accuracy rate refers to the parsing test set input into the convolutional neural network model. The ratio of the number of data test results outputted by the convolutional neural network model that are consistent with the data recognition results in the parsing test set to the total number of samples in the parsing test set, the preset accuracy rate refers to the preset value of the parsing accuracy rate for judging whether the convolutional neural network model meets the output standard, such as the preset accuracy rate can be set to 95%, the basic body proportion refers to the proportion of the body part of the game image to the overall part of the game image, the decoration proportion refers to the proportion of the decorative part of the game image to the overall part of the game image, the preset body proportion refers to the preset value used to judge the real situation of the game image, the real situation of the game image refers to the similarity between the game image and the real person, and the real situation of the game image includes the real situation of the game image being real and the real situation of the game image being unreal.

[0089] Specifically, by optimizing the preset access volume K0 and reducing the value of the preset access volume K0, when the basic body ratio is greater than the preset body ratio, the game image is closer to the real state. The value of the preset access volume K0 should be lowered to make the judgment requirements of the public image more stringent and avoid the overlap of the game image and the public image.

[0090] Specifically, the real-time feedback optimization module obtains the gamification data point frequency c1 through the sensor node group, and compares the gamification data point frequency c1 with the preset gamification point frequency c0, where c0 is ≥ 7 times. The importance of the gamification data point is judged based on the comparison result, and the acquisition process of the creative source data is optimized in real time based on the judgment result, wherein:

[0091] When c1≥c0, the real-time feedback optimization module determines that the gamification data point is important and does not perform real-time optimization on the acquisition process of the creative source data;

[0092] When c1<c0, the real-time feedback optimization module determines that the importance of the gamification data point is unimportant, and optimizes the acquisition process of the creative source data in real time, canceling the acquisition of the creative source data for the gamification data point.

[0093] Specifically, the gamification data point frequency refers to the frequency at which the gamification data point obtains data, the preset gamification point frequency refers to the preset value used to judge the importance of the gamification data point, and the gamification data point importance refers to the degree of importance of the gamification data point. The gamification data point importance includes the gamification data point importance being important and the gamification data point importance being unimportant.

[0094] Specifically, the real-time feedback optimization module compares the gamification data point frequency c1 with the preset gamification point frequency c0 to obtain the importance of the gamification data point, and optimizes the acquisition process of the creative source data in real time according to the importance. If c1 is less than c0, the acquisition process of the creative source data is optimized in real time, and the real-time optimization is performed by canceling the acquisition of the creative source data for the gamification data point, thereby reducing the data collection of unimportant game points, reducing the generation time of the game image and the computing power burden of the GPU.

[0095] See also Figure 2 , which is a flow chart of the gamification creation method based on machine learning in this embodiment, the method includes:

[0096] Step S1, acquiring the creation source data through the sensor node group;

[0097] Step S2: The gamification data identification module identifies the creative source data according to the data identification model to obtain gamification data and non-gamification data. The module also adjusts the acquisition process of the creative source data according to the number of non-gamification data items of the same source.

[0098] Step S3: Creating a game character based on the gamification data, monitoring network latency, and optimizing the game character creation process in real time based on the network latency. Furthermore, optimizing the adjustment method for obtaining the source data based on the real-time optimization method, and adjusting the real-time optimization process of the game character creation process based on the computing power margin of the device.

[0099] Step S4: Output the game avatar, identify the public image based on the number of visits to the game avatar, correct the output of the game avatar based on the public similarity coefficient of the game avatar, and optimize the public image identification process based on the basic subject ratio of the game avatar;

[0100] Step S5: Optimize the acquisition process of the creative source data in real time according to the frequency of the gamification data points.

[0101] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A gamification creation system based on machine learning, characterized in that: include: A sensor data acquisition module is used to acquire the creation source data through a group of sensor nodes; A gamification data identification module is used to identify the source data according to the data identification model to obtain gamification data and non-gamification data, and is also used to adjust the acquisition process of the source data according to the number of non-gamification data items of the same source; A machine learning creation module, which is used to create game characters based on gamification data, monitor network latency, and optimize the creation process of game characters in real time based on network latency. It is also used to adjust the real-time optimization process of the game character creation process based on the device's computing power margin; A game creation output module is used to output game characters, identify public characters based on the number of visits to the game characters, calibrate the output results of the game characters based on the public similarity coefficient of the game characters, and optimize the public character recognition process based on the basic subject ratio of the game characters; The real-time feedback optimization module is used to optimize the acquisition process of creative source data in real time according to the frequency of gamification data points.

2. The gamification creation system based on machine learning according to claim 1, characterized in that: The gamified data recognition module sets the loss function of the data recognition model to be a binary cross entropy function, and N is the total number of samples in the data recognition test set. i Identify the sample number in the test set for the data, y i Identify the actual category of the i-th sample in the test set for the data. For the gamified data y i =1, for non-gamified data y i =0, is the predicted probability of the data recognition model, which indicates the predicted probability that the i-th sample is gamification data.

3. The gamification creation system based on machine learning according to claim 2, characterized in that: The gamification data identification module obtains the sensor node identifier of the non-gamification data, counts the non-gamification data with the same sensor node identifier within the cluster monitoring period, obtains the number n of non-gamification data of the same source, compares the number n of non-gamification data of the same source with the preset number n0 of non-gamification data of the same source, and determines the clustering state of the non-gamification data based on the comparison result.

4. The gamification creation system based on machine learning according to claim 3, characterized in that: The machine learning creation module preprocesses the gamification data to obtain preprocessed gamification data, performs feature selection on the preprocessed gamification data to obtain gamification feature data, selects a machine learning model for the gamification data according to modeling requirements, and obtains an adaptive creation model.

5. The gamification creation system based on machine learning according to claim 4, characterized in that: The machine learning creation module sends an ICMP request to the target address, measures the round-trip time of the ICMP request, obtains the current network delay A, and compares the current network delay A with the maximum network delay Amax and the minimum network delay Amin, Amax ≥ 150ms, Amin ≤ 50ms. The network delay situation is judged according to the comparison result, and the game image creation process is optimized in real time according to the judgment result.

6. The gamification creation system based on machine learning according to claim 5, characterized in that: The machine learning creation module obtains the current GPU real-time usage rate z, and calculates the device computing power margin β based on the current GPU real-time usage rate z and the theoretical maximum value zmax, compares the device computing power margin β with the preset device computing power margin β0, judges the current device computing power margin situation based on the comparison result, and adjusts the real-time optimization process of the game image creation process based on the judgment result.

7. The gamification creation system based on machine learning according to claim 6, characterized in that: The game creation output module obtains the game image of the machine learning creation module and outputs the game image. The game creation output module obtains the number of visits K of the network image through network retrieval, compares the number of visits K with the preset number of visits K0, K0 ≥ 10,000 times, and judges the category of the network image based on the comparison result.

8. The gamification creation system based on machine learning according to claim 7, characterized in that: The game creation output module inputs the marking feature data and the gamification feature data into the similarity comparison model, outputs the public similarity coefficient Q of the game image, and compares the public similarity coefficient Q of the game image with the preset similarity coefficient Q0, Q0≥70%, judges the public similarity degree of the game image based on the comparison result, and corrects the output result of the game image based on the judgment result.

9. The gamification creation system based on machine learning according to claim 8, characterized in that: The game creation output module inputs the game image into the basic body proportion model and outputs the basic body proportion. The basic body proportion includes the basic body proportion r1 and the decoration proportion r2. Set r1+r2=100%, compare the basic body proportion r1 with the preset body proportion r0, r0≥65%, and judge the actual situation of the game image based on the comparison result, and optimize the public image recognition process based on the judgment result.

10. A method for the gamification creation system based on machine learning according to any one of claims 1 to 9, characterized in that: include: Step S1, acquiring the creation source data through the sensor node group; Step S2: The gamification data identification module identifies the creative source data according to the data identification model to obtain gamification data and non-gamification data. The module also adjusts the acquisition process of the creative source data according to the number of non-gamification data items of the same source. Step S3: Creating a game character based on the gamification data, monitoring network latency, and optimizing the game character creation process in real time based on the network latency. Furthermore, optimizing the adjustment method for obtaining the source data based on the real-time optimization method, and adjusting the real-time optimization process of the game character creation process based on the computing power margin of the device. Step S4: Output the game avatar, identify the public image based on the number of visits to the game avatar, correct the output of the game avatar based on the public similarity coefficient of the game avatar, and optimize the public image identification process based on the basic subject ratio of the game avatar; Step S5: Optimize the acquisition process of the creative source data in real time according to the frequency of the gamification data points.

Images