A deep learning-based VR simulation scene control method and system

Abstract

The application provides a VR simulation scene control method and system based on deep learning. The interest evaluation description content of the user experience data to be parsed is obtained, so that the user experience data parsing thread can more accurately explain the description of the region in the user experience data to be parsed with the help of the interest evaluation description content, the VR interaction preference parsed by the user experience data parsing thread can be more perfectly associated with the region of the user experience data to be parsed, and the parsing accuracy of the VR interaction preference is improved. The interest evaluation description content of the user experience data to be parsed is obtained by means of the artificial intelligence evaluation thread of the user experience data parsing thread. The user experience data parsing thread can obtain reliable interest evaluation description content, improve the correlation degree of the subsequent parsed VR interaction preference and the user experience data to be parsed, and thus guarantee the pertinence and precision of the VR simulation scene control strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of VR control, in particular to a VR simulation scene control method and system based on deep learning. BACKGROUND

[0002] Virtual reality technology (VR) is a brand-new practical technology developed in the 20th century. Virtual reality technology covers computer, electronic information, and simulation technology, and its basic implementation is to simulate a virtual environment by a computer to give people a sense of environmental immersion. With the continuous development of social productivity and scientific technology, the demand for VR technology in various industries is increasing. VR technology has made great progress and has gradually become a new scientific and technological field.

[0003] With the increasingly wide application of VR technology, VR technology gradually changes people's life experience and improves the richness of people's life. Under the background of people's increasingly high requirements for VR technology, how to individualize the VR simulation scene to increase the game stickiness and sense of achievement is currently a technical problem to be solved. SUMMARY

[0004] To improve the technical problems existing in the related art, the present application provides a VR simulation scene control method and system based on deep learning.

[0005] In a first aspect, a VR simulation scene control method based on deep learning is provided, applied to a control system, the method at least comprising: obtaining user experience data to be analyzed; obtaining interest evaluation description content of the user experience data to be analyzed by means of an artificial intelligence evaluation thread; based on the interest evaluation description content, analyzing the user experience data to be analyzed by means of a user experience data analysis thread to obtain a VR interaction preference of the user experience data to be analyzed; and determining a VR simulation scene control strategy for the user experience data to be analyzed based on the VR interaction preference.

[0006] In an independently implemented embodiment, the VR interaction preference includes perception experience data; based on the interest evaluation description content, the user experience data to be analyzed is analyzed by means of a user experience data analysis thread to obtain a VR interaction preference of the user experience data to be analyzed, including: the user experience data to be analyzed is processed by means of a user experience data analysis thread to obtain regional perception requirement data of the user experience data to be analyzed; and the regional perception requirement data and the interest evaluation description content are combined to obtain perception experience data of the user experience data to be analyzed.

[0007] In an independently implemented embodiment, the area-aware requirement data is a best description set of interest evaluation descriptions best describing content of interest evaluation descriptions covering experience events in which the user experience data to be parsed differs; the combining the area-aware requirement data and the interest evaluation description content to obtain the awareness experience data of the user experience data to be parsed comprises: performing a binary operation on the best description set of interest evaluation descriptions and the cluster of interest evaluation descriptions to obtain the awareness experience data of the user experience data to be parsed.

[0008] In an independently implemented embodiment, the user experience data parsing thread comprises a multiplex feature mining layer and an awareness translation unit; the processing the user experience data to be parsed by means of the user experience data parsing thread to obtain the area-aware requirement data of the user experience data to be parsed comprises: performing key semantic mining on the user experience data to be parsed by means of the multiplex feature mining layer to obtain user experience data key descriptions, and clustering the user experience data key descriptions and the first area distribution key descriptions output by the interest evaluation description feature mining layer of the artificial intelligence evaluation thread to obtain first clustered key descriptions; and translating the first clustered key descriptions by means of the awareness translation unit to obtain the area-aware requirement data of the user experience data to be parsed.

[0009] In an independently implemented embodiment, the multiplex feature mining layer comprises one or more feature mining layers arranged in a certain order, each of the feature mining layers comprising a best description unit of interest evaluation descriptions; the clustering the user experience data key descriptions and the first area distribution key descriptions output by the interest evaluation description feature mining layer of the artificial intelligence evaluation thread to obtain first clustered key descriptions comprises: loading the user experience data key descriptions to the first feature mining layer; for each of the feature mining layers: clustering the key descriptions output by the previous feature mining layer and the first area distribution key descriptions by means of the best description unit of interest evaluation descriptions to obtain second clustered key descriptions bound to the feature mining layer; wherein the key words in the area distribution key descriptions bound to each of the feature mining layers differ in completeness; and obtaining the first clustered key descriptions based on the second clustered key descriptions of the last feature mining layer.

[0010] In an independently implemented embodiment, before the step of clustering the key descriptions output by the previous feature mining layer and the regional distribution key descriptions by the interest evaluation description optimal description unit to obtain the second clustered key descriptions of the feature mining layer, the method further comprises: compressing the key descriptions output by the previous feature mining layer; and / or the step of clustering the key descriptions output by the previous feature mining layer and the regional distribution key descriptions by the interest evaluation description optimal description unit to obtain the second clustered key descriptions of the feature mining layer comprises: adjusting the regional distribution key descriptions to the regional distribution key descriptions within the previous perception range, integrating the adjusted regional distribution key descriptions with the key descriptions output by the previous feature mining layer, and performing feature extraction to obtain the second clustered key descriptions of the feature mining layer.

[0011] In an independently implemented embodiment, the step of translating the first clustered key descriptions by the perception degree translation unit to obtain the regional perception requirement data of the user experience data to be parsed comprises: translating the first clustered key descriptions and one or more second clustered key descriptions of the interest evaluation description optimal description unit by the perception degree translation unit to obtain the regional perception requirement data of the user experience data to be parsed.

[0012] In an independently implemented embodiment, the user experience data parsing thread further comprises a somatic simulation degree translation unit; and the step of parsing the user experience data to be parsed by the user experience data parsing thread based on the interest evaluation description content to obtain the VR interaction preference of the user experience data to be parsed comprises: translating the first clustered key descriptions by the somatic simulation degree translation unit to obtain the somatic simulation degree user experience data of the user experience data to be parsed.

[0013] In an independently implemented embodiment, the step of translating the first clustered key descriptions by the somatic simulation degree translation unit to obtain the somatic simulation degree user experience data of the user experience data to be parsed comprises: translating the first clustered key descriptions and one or more second clustered key descriptions of the interest evaluation description optimal description unit by the somatic simulation degree translation unit to obtain the somatic simulation degree user experience data of the user experience data to be parsed.

[0014] In an independently implemented embodiment, the artificial intelligence evaluation thread comprises an interest evaluation description feature mining layer, an interest evaluation description translation unit, and a CNN thread; the interest evaluation description content of the user experience data to be analyzed is obtained by means of the artificial intelligence evaluation thread, comprising: feature mining of the user experience data to be analyzed by means of the interest evaluation description feature mining layer to obtain first regional distribution key descriptions; translation of the first regional distribution key descriptions by means of the interest evaluation description translation unit to obtain translated key descriptions; clustering of the first regional distribution key descriptions and the translated key descriptions by means of the CNN thread to obtain the interest evaluation description content of the user experience data to be analyzed.

[0015] In an independently implemented embodiment, the artificial intelligence evaluation thread and the user experience data analysis thread are configured respectively.

[0016] In an independently implemented embodiment, before the interest evaluation description content of the user experience data to be analyzed is obtained by means of the artificial intelligence evaluation thread, the method further comprises: configuring the artificial intelligence evaluation thread by means of a first example cluster, wherein the user experience data in the first example cluster is labeled with interest evaluation description content; obtaining example interest evaluation description content of user experience data in a second example cluster by means of the configured artificial intelligence evaluation thread, and configuring the user experience data analysis thread by means of the second example cluster and the example interest evaluation description content.

[0017] In an independently implemented embodiment, the second example cluster comprises a first local example cluster and a second local example cluster, and configuring the user experience data analysis thread by means of the second example cluster and the example interest evaluation description content comprises: configuring the user experience data analysis thread by means of the first local example cluster and the example interest evaluation description content bound to the first local example cluster, to debug variables of a multiplexed feature mining layer and a perceptual degree translation unit in the user experience data analysis thread; configuring the user experience data analysis thread by means of the second local example cluster and the example interest evaluation description content bound to the second local example cluster, to debug variables of a multiplexed feature mining layer and a somatosensory simulation degree translation unit in the user experience data analysis thread.

[0018] In a second aspect, a VR simulation scene control system based on deep learning is provided, comprising a processor and a memory in communication with each other, the processor being configured to retrieve a computer program from the memory and implement the above method by running the computer program.

[0019] The VR simulation scene control method and system based on deep learning provided by the embodiment of the application can make the user experience data analysis thread more accurately explain the description of the region in the user experience data to be analyzed by means of the interest evaluation description content, so that the VR interaction preference obtained by the user experience data analysis thread can be more perfectly associated with the region of the user experience data to be analyzed, and the analysis accuracy of the VR interaction preference is improved. In addition, the interest evaluation description content of the user experience data to be analyzed is obtained by means of the artificial intelligence evaluation thread of the user experience data analysis thread. The user experience data analysis thread can obtain reliable interest evaluation description content, further improve the association degree between the VR interaction preference obtained by subsequent analysis and the user experience data to be analyzed, and thus guarantee the pertinence and precision of the VR simulation scene control strategy. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0021] Figure 1 The flowchart of the VR simulation scene control method based on deep learning provided by the embodiment of the application.

[0022] Figure 2 The block diagram of the VR simulation scene control device based on deep learning provided by the embodiment of the application.

[0023] Figure 3 The architecture diagram of the VR simulation scene control system based on deep learning provided by the embodiment of the application. DETAILED DESCRIPTION

[0024] In order to better understand the above technical solutions, the technical solutions of the application will be described in detail below by means of the drawings and specific embodiments. It should be understood that the specific features of the embodiments of the application and the embodiments are detailed descriptions of the technical solutions of the application, and are not limitations on the technical solutions of the application. In the case of no conflict, the technical features in the embodiments of the application and the embodiments can be combined with each other.

[0025] Please refer to Figure 1 , which shows a VR simulation scene control method based on deep learning. The method can include the technical solutions described in the following step11-step14.

[0026] step11: obtaining the user experience data to be parsed.

[0027] The user experience data to be parsed is the user experience data as the initial transmission, which is used to parse the VR interaction preference bound therewith. The user experience data to be parsed can be the first user experience data, or the second user experience data, etc.

[0028] step12: obtaining the interest evaluation description content of the user experience data to be parsed by means of the artificial intelligence evaluation thread.

[0029] The artificial intelligence evaluation thread is an artificial intelligence evaluation thread constructed based on the interest evaluation description, which is used to filter the key data of the user experience data to be parsed to obtain the interest evaluation description content of the user experience data to be parsed. The artificial intelligence evaluation thread filters the key data of the user experience data to be parsed to obtain a plurality of key descriptions. The interest evaluation description content is, for example, the interest evaluation description content of each experience event node in the user experience data to be parsed. Through the interest evaluation description content, the keywords in the user experience data to be transmitted can be obtained, such as the distribution data of the region in the user experience data to be parsed.

[0030] For a possible embodiment, the artificial intelligence evaluation thread is a global feature extraction artificial intelligence evaluation thread, which can be composed of a plurality of analysis threads. The plurality of analysis threads can cluster the key descriptions (number of key descriptions with differences, degree of differentiation of user experience data with differences) to obtain a higher degree of differentiation and reliable VR interaction preference in the user experience data.

[0031] step13: parsing the user experience data to be parsed by means of the user experience data parsing thread based on the interest evaluation description content to obtain the VR interaction preference of the user experience data to be parsed.

[0032] In this way, after obtaining the interest evaluation description content of the user experience data to be parsed, the user experience data parsing thread can parse the user experience data to be transmitted by means of the interest evaluation description content. For example, the user experience data parsing thread can parse the user experience data to be parsed based on the interest evaluation description content of each experience event in the interest evaluation description content and the distribution data of the region covered in the interest evaluation description content to obtain the VR interaction preference, that is, the perceptual experience data and the somatosensory simulation degree user experience data.

[0033] For a possible embodiment, the user experience data parsing thread is a global feature extraction artificial intelligence evaluation thread.

[0034] step 14: determining a VR simulation scene control strategy for the user experience data to be parsed based on the VR interaction preference.

[0035] For example, the VR simulation scene control strategy can be understood as an intelligent control method of the VR simulation scene. Different VR simulation scenes can be built according to the preferences of different users, so that the perception in the VR can be optimized (the perception in the VR can include visual, auditory, tactile, force perception, and motion perception), thereby improving the realism of the VR simulation scene. Therefore, the needs of different users can be met as much as possible, and the interaction, immersion, and imagination of the users can be improved.

[0036] In this way, by obtaining the interest evaluation description content of the user experience data to be parsed, the user experience data parsing thread can more accurately explain the description of the region in the user experience data to be parsed with the help of the interest evaluation description content, so that the VR interaction preference parsed by the user experience data parsing thread can be more perfectly associated with the region of the user experience data to be parsed, and the accuracy of the VR interaction preference is improved. In addition, the interest evaluation description content of the user experience data to be parsed is obtained by the artificial intelligence evaluation thread of the user experience data parsing thread. The user experience data parsing thread can obtain reliable interest evaluation description content, further improving the correlation between the VR interaction preference parsed subsequently and the user experience data to be parsed, thereby ensuring the pertinence and precision of the VR simulation scene control strategy.

[0037] The VR simulation scene control method based on the interest evaluation description of the present disclosure can include the following steps.

[0038] step 21: obtaining user experience data to be parsed.

[0039] step 22: obtaining interest evaluation description content of the user experience data to be parsed by an artificial intelligence evaluation thread.

[0040] In an alternative embodiment, the interest evaluation description content is an interest evaluation description cluster that covers the interest evaluation description of the experience events that differ in the user experience data to be parsed, that is, each experience event in the user experience data to be parsed has a bound interest evaluation description.

[0041] In an alternative embodiment, the artificial intelligence evaluation thread includes an interest evaluation description feature mining layer, an interest evaluation description translation unit, and a CNN thread. The interest evaluation description feature mining layer can perform key semantic mining on the user experience data to be parsed, the interest evaluation description translation unit can translate the description content and output the key description, and the CNN thread can optimize the output of the translation unit.

[0042] The present disclosure is based on a VR simulation scene control method based on interest evaluation description. The interest evaluation description content of the user experience data to be parsed is obtained by means of an artificial intelligence evaluation thread. The specific implementation steps can include the contents described in steps 221-223.

[0043] Step 221: Feature mining is performed on the user experience data to be parsed by means of an interest evaluation description feature mining layer to obtain a first regional distribution key description.

[0044] The user experience data to be parsed can be feature-mined by means of the interest evaluation description feature mining layer of the artificial intelligence evaluation thread to screen the key semantics in the user experience data to be parsed. The key semantics obtained by the interest evaluation description feature mining layer performing feature mining on the user experience data to be parsed is, for example, the distribution key semantics of the regions in the user experience data to be parsed. Finally, the interest evaluation description feature mining layer can output a first regional distribution key description, i.e., a description of the distribution of the regions in the user experience data to be parsed.

[0045] On the basis of the interest evaluation description feature mining layer having several distributions, the user experience data to be parsed can be feature-mined (i.e., key semantic mining) by means of the interest evaluation description feature mining layer. The key description obtained by each feature mining layer is a first regional distribution key description. For example, when the feature mining layer includes 8 feature mining methods for feature mining, the first feature mining method will perform feature mining on the user experience data to be parsed, and then output a first regional distribution key description. The second feature mining method takes the first regional distribution key description output by the previous feature mining method as a reference, performs feature mining again, and then outputs a bound first regional distribution key description. In addition, the completeness of the keywords in each of the first regional distribution key descriptions output by the interest evaluation description feature mining layer can be perceived to differ. At this time, the bound first regional distribution key description of the last layer is loaded into the interest evaluation description translation unit. By performing multiple feature mining on the user experience data to be parsed, more accurate key semantics can be gradually screened out, making the distribution of the regions of the user experience data to be parsed more reliable.

[0046] step 222: translating the first regional distribution key description by the interest evaluation description translation unit to obtain a translated key description.

[0047] After the interest evaluation description feature mining layer mines the features of the user experience data to be analyzed and outputs the first regional distribution key description, the first regional distribution key description can be translated by the interest evaluation description translation unit to obtain a translated key description. When the interest evaluation description translation unit translates the first regional distribution key description, the key semantics filtered by the interest evaluation description feature mining layer can be translated, and a translated key description of the prior perception category and the prior perception distinguishing degree can be reconstructed. For example, the number of key semantics in the translated key description can be 128, and the distinguishing degree is half of the user experience data to be analyzed.

[0048] For a possible embodiment, when the interest evaluation description translation unit has multiple distributions, the interest evaluation description translation unit also performs multiple translations on the first regional distribution key description in parallel. A first translation unit translates the first regional distribution key description and then outputs a bound prior translated key description. A second translation unit translates the translated key description output by the first translation unit and then outputs a bound prior translated key description. Further, the last output prior translated key description is the translated key description.

[0049] step 223: clustering the first regional distribution key description and the translated key description by the CNN thread to obtain the interest evaluation description content of the user experience data to be analyzed.

[0050] After translation by the interest evaluation description feature mining layer, in order to further optimize the key semantics output by the interest evaluation description feature mining layer and obtain more reliable regional distribution data of the user experience data to be analyzed, the first regional distribution key description and the translated key description can be clustered by the CNN thread to obtain the interest evaluation description content of the user experience data to be analyzed. For example, the key semantics in the first regional distribution key description and the key semantics in the translated key description can be clustered together to obtain the interest evaluation description content of the user experience data to be analyzed. For example, the number of key semantics in the first regional distribution key description and the number of key semantics in the translated key description are both 128, and the interest evaluation description content obtained after clustering can be 256.

[0051] For a possible embodiment, the interest evaluation description content is an interest evaluation description cluster that covers the interest evaluation description of the experience event that exists in the difference of the user experience data to be analyzed, that is, each experience event in the user experience data to be analyzed has a bound interest evaluation description.

[0052] For a possible embodiment, the first regional distribution key description of each binding can be integrated into a second regional distribution key description by means of a CNN thread based on the distribution of the interest evaluation description feature mining layer, and the second regional distribution key description and the translation key description can be integrated into a third regional distribution key description.

[0053] For some possible embodiments, the first regional distribution key description output by the part of the feature mining layer of the interest evaluation description feature mining layer can also be integrated.

[0054] In an alternative embodiment, the first regional distribution key description output by each feature mining layer can be processed by means of an optimization local thread first, so that the description categories and discrimination degrees of each second regional distribution key description are consistent.

[0055] After obtaining the third regional distribution key description, the optimization local thread can further translate based on the key semantics of the third regional distribution key description to obtain the interest evaluation description content of the user experience data to be parsed, such as interest evaluation description clusters.

[0056] After obtaining the interest evaluation description content of the user experience data to be parsed, the user experience data to be parsed can be parsed by means of the obtained interest evaluation description content to obtain the VR interaction preference of the user experience data to be parsed. When the perception experience data is needed, the above-mentioned step of "parsing the user experience data to be parsed by means of the user experience data parsing thread based on the interest evaluation description content to obtain the VR interaction preference of the user experience data to be parsed" can include the following steps.

[0057] step23: processing the user experience data to be parsed by means of the user experience data parsing thread to obtain the regional perception requirement data of the user experience data to be parsed.

[0058] The user experience data parsing thread is, for example, a global feature extraction artificial intelligence evaluation thread. The user experience data parsing thread can implement the key semantic mining step on the user experience data to be parsed to obtain the regional perception requirement data of the user experience data to be parsed. For example, the regional perception requirement data can be understood as the perception result of the region in the user experience data to be parsed. For example, the regional perception requirement data is an interest evaluation description best description set covering the best description content in the interest evaluation description of the experience event that exists in the user experience data to be parsed. The interest evaluation description best description set can be used for feature mining regional perception requirements.

[0059] For one possible embodiment, the user experience data parsing thread comprises a multiplex feature mining layer, a perceptual degree translation unit, and a somatosensory simulation degree translation unit. The processing of the user experience data to be parsed by the user experience data parsing thread to obtain the content described by the regional perceptual requirement data of the user experience data to be parsed can specifically include the following steps.

[0060] step231: The multiplex feature mining layer performs key semantic mining on the user experience data to be parsed to obtain user experience data key descriptions, and clusters the user experience data key descriptions and the first regional distribution key descriptions output by the interest evaluation description feature mining layer of the artificial intelligence evaluation thread to obtain first clustered key descriptions.

[0061] The multiplex feature mining layer refers to the key semantics extracted by the feature mining layer, which are used to obtain both perceptual experience data and somatosensory simulation degree user experience data. The first clustered key descriptions obtained after clustering can include the distribution key semantics of the region in the user experience data to be parsed and the remaining key semantics.

[0062] In an alternative embodiment, the multiplex feature mining layer comprises one or more feature mining layers arranged in a certain order, and each feature mining layer comprises an interest evaluation description best description unit.

[0063] The VR simulation scene control method based on the interest evaluation description of the present disclosure. The user experience data key descriptions and the regional distribution key descriptions output by the interest evaluation description feature mining layer of the artificial intelligence evaluation thread are clustered to obtain first clustered key descriptions, and the specific implementation steps can include the content described in steps 2311-2313.

[0064] step2311: Load the user experience data key descriptions to the first feature mining layer.

[0065] First, the remaining feature mining layers of the user experience data parsing thread can perform key semantic mining on the user experience data to be parsed to obtain user experience data key descriptions. Then, the user experience data key descriptions are loaded to the first feature mining layer. The user experience data key descriptions are further processed by the feature mining layer.

[0066] step2312: Each feature mining layer clusters the key descriptions and the regional distribution key descriptions output by the previous feature mining layer by means of the interest evaluation description best description unit to obtain the second clustered key descriptions bound by the feature mining layer; wherein the types of keywords in the regional distribution key descriptions bound by each feature mining layer are different.

[0067] After the feature mining layer obtains the key description of the user experience data, the interest evaluation description can be used to describe the best description unit to cluster the key description and the regional distribution key description output by the previous feature mining layer to obtain the second clustered key description of the feature mining layer. Among them, the key words in the regional distribution key description of each feature mining layer are different in integrity. The difference in the integrity of the key words can be understood as the difference in the distinguishability of the regional distribution key description and the direction of the key semantics. At least one of the distinguishability and the type of key semantics can be different.

[0068] For the first feature mining layer, the obtained user experience data key description is obtained by key semantic mining from the remaining feature extraction set. For the second feature mining layer, the obtained user experience data key description is the second clustered key description output by the previous feature mining layer.

[0069] When the interest evaluation description feature mining layer of the artificial intelligence evaluation thread has only one, the interest evaluation description feature mining layer only outputs one first regional distribution key description. At this time, all feature mining layers can cluster the unique first regional distribution key description with the key description output by the previous feature mining layer. When the interest evaluation description feature mining layer of the artificial intelligence evaluation thread has multiple, the first regional distribution key description output by the multiple interest evaluation description feature mining layers can be clustered with the key description output by the feature mining layer, respectively. For example, the first regional distribution key description obtained by the first interest evaluation description feature mining layer is loaded into the first feature mining layer, and the first regional distribution key description obtained by the second interest evaluation description feature mining layer is loaded into the second feature mining layer, so that the second feature mining layer can cluster the first regional distribution key description obtained by the second interest evaluation description feature mining layer with the key description output by the previous feature mining layer.

[0070] In an alternative embodiment, the interest evaluation description optimal description unit clusters the key description output by the previous feature mining layer and the region distribution key description to obtain the second clustered key description of the feature mining layer binding, which can include the following: the interest evaluation description optimal description unit debugs the region distribution key description to the region distribution key description of the previous perception range, such as debugging the discrimination and the dimension of the key semantics of the region distribution key description. The interest evaluation description optimal description unit integrates and features the debugged region distribution key description and the key description output by the previous feature mining layer to obtain the second clustered key description of the feature mining layer binding. For example, the interest evaluation description optimal description unit of the second feature mining layer can integrate and feature the second clustered key description output by the first feature mining layer and the region distribution key description transmitted thereto to obtain the second clustered key description of the second feature mining layer binding.

[0071] In this way, the interest evaluation description optimal description unit also integrates and features the region distribution key description and the key description output by the previous feature mining layer to realize clustering of the region distribution key description and the key description output by the previous feature mining layer.

[0072] For a possible embodiment, for each feature mining layer, the key description output by the previous feature mining layer can be compressed before the interest evaluation description optimal description unit clusters the key description output by the previous feature mining layer and the region distribution key description to obtain the second clustered key description of the feature mining layer binding. For example, the second feature mining layer compresses the second clustered key description output by the first feature mining layer. After compression, the second clustered key description can be simplified to make the second clustered key description more accurate.

[0073] Step 2313: obtaining the first clustered key description based on the second clustered key description of the last feature mining layer.

[0074] For a possible embodiment, on the basis that the last feature mining layer of the multiplex feature mining layer thread is not the last feature mining layer, that is, there are multiple feature mining manners after the last feature mining layer for further feature mining on the second clustered key description output by the last feature mining layer to further process the clustered key semantics. The graph output by the last feature mining layer after processing is the first clustered key description. For example, the second clustered key description output by the last feature mining layer can be compressed to further simplify the second clustered key description, and then feature mining is performed again by means of the feature mining manner to screen the key data. At this time, the output key description is the first clustered key description.

[0075] For a possible implementation embodiment, the second clustering key description can also be directly taken as the first clustering key description.

[0076] In this way, by means of the region distribution key description and the user experience data to be parsed, which are output by the artificial intelligence evaluation thread through the best description unit of the interest evaluation description, the user experience data key description obtained by the key semantic mining of the user experience data to be parsed by the user experience data parsing thread is clustered, so that the user experience data parsing thread can subsequently implement the effect of passing the key semantics obtained by the artificial intelligence evaluation thread to the user experience data parsing thread by means of the region distribution data about the region in the user experience data to be parsed in the region distribution key description, thereby improving the analysis accuracy of the VR interaction preference.

[0077] After obtaining the first clustering key description, the user experience data to be parsed can be further analyzed by means of the first clustering key description to obtain the VR interaction preference.

[0078] Step 232: translating the first clustering key description by means of the perception degree translation unit to obtain the region perception requirement data of the user experience data to be parsed.

[0079] Because the first clustering key description covers the distribution key semantics of the region and the remaining key semantics of the user experience data to be parsed, the first clustering key description can be translated by means of the perception degree translation unit to obtain the region perception requirement data of the user experience data to be parsed, such as obtaining the interest evaluation description best description set about the best description content in the interest evaluation description of each experience event in the user experience data to be parsed.

[0080] For a possible implementation embodiment, the first clustering key description and the second clustering key description of one or more interest evaluation description best description units can be translated by means of the perception degree translation unit to obtain the region perception requirement data of the user experience data to be parsed. The perception degree translation unit can simultaneously obtain the first clustering key description output by the last one of the multiplex feature mining layers of the user experience data parsing thread, and the second clustering key description of one or more interest evaluation description best description units, and translate the two key descriptions to obtain the region perception requirement data of the user experience data to be parsed. When the last one of the multiplex feature mining layers is the feature mining layer, the first clustering key description output by the last one of the multiplex feature mining layers and the second clustering key description output by the interest evaluation description best description unit of the remaining feature mining layers can be translated.

[0081] In an alternative embodiment, when the number of feature mining layers of the multiplexed feature mining layer is multiple, the perceptual degree translation unit can simultaneously obtain the second cluster key descriptions output by the several feature mining layers for translation. For example, if the perceptual degree translation unit obtains six second cluster key descriptions output by six feature mining layers, then in the perceptual degree translation unit, six feature extraction units can obtain the second cluster key descriptions output by the six feature mining layers respectively for translation. For example, the first feature extraction unit of the perceptual degree translation unit can obtain the first cluster key description output by the multiplexed feature mining layer and the second cluster key description output by the first interest evaluation description best description unit for translation, and output the key description. The second feature extraction unit of the perceptual degree translation unit can obtain the key description output by the previous feature extraction unit and the second cluster key description output by the second interest evaluation description best description unit for translation.

[0082] In an alternative embodiment, after the feature extraction units of the perceptual degree translation unit translate based on the first cluster key description and the second cluster key description, the multiple feature extraction units can be used for translation to debug the perceptual degree set finally output by the perceptual degree translation unit.

[0083] In this way, by obtaining the regional perceptual requirement data of the user experience data to be analyzed, such as the interest evaluation description best description set of the best description content in each interest evaluation description, the corresponding perceptual requirement can be established, and the accuracy of the VR interaction preference analysis obtained by the user experience data analysis thread under the premise of perceptual difficulty is improved.

[0084] Step 24: obtaining the perceptual degree experience data of the user experience data to be analyzed based on the regional perceptual requirement data and the interest evaluation description content.

[0085] After obtaining the regional perceptual requirement data of the region in the user experience data to be analyzed, the user experience data to be analyzed can be analyzed based on the regional perceptual requirement data and the interest evaluation description content output by the artificial intelligence evaluation thread to obtain the perceptual degree experience data of the user experience data to be analyzed. For example, the interest evaluation description best description set and the interest evaluation description cluster can be used to obtain the perceptual degree experience data of the user experience data to be analyzed.

[0086] For a possible embodiment, the interest evaluation description best description set and the interest evaluation description cluster can be processed by binary operation to obtain the perceptual degree experience data of the user experience data to be analyzed.

[0087] In the present embodiment, the user experience data parsing thread further comprises a haptic fidelity translation unit. Because the haptic fidelity user experience data can be obtained by translating the first clustered key description extracted by the multiplexed feature mining layer. Thus, after the key semantic mining of the user experience data to be parsed by the multiplexed feature mining layer, i.e. after step 231, the following steps can be implemented.

[0088] Step 1: translating the first clustered key description by means of the haptic fidelity translation unit to obtain the haptic fidelity user experience data of the user experience data to be parsed.

[0089] According to the above, the first clustered key description output by the last layer of the multiplexed feature mining layer covers the regional distribution key semantic of the region in the user experience data to be parsed. Thus, the first clustered key description can be translated by means of the haptic fidelity translation unit to obtain the haptic fidelity user experience data of the user experience data to be parsed.

[0090] For a possible embodiment, the first clustered key description and the second clustered key description of the one or more interest evaluation description optimal description unit can be translated by means of the haptic fidelity translation unit to obtain the haptic fidelity user experience data of the user experience data to be parsed. The haptic fidelity translation unit can simultaneously obtain the first clustered key description output by the last layer of the multiplexed feature mining layer of the user experience data parsing thread and the second clustered key description of the one or more interest evaluation description optimal description unit, and translate the two key descriptions to obtain the haptic fidelity user experience data of the user experience data to be parsed. When the last layer of the multiplexed feature mining layer is the feature mining layer, the first clustered key description output by the final feature mining layer and the second clustered key description output by the interest evaluation description optimal description unit of the remaining feature mining layer can be translated.

[0091] In an alternative embodiment, when the number of feature mining layers of the multiplexed feature mining layer is many, the somatosensory simulation degree translation unit can simultaneously obtain the second clustering key description output by the plurality of feature mining layers to perform translation. For example, if the somatosensory simulation degree translation unit obtains six second clustering key descriptions output by six feature mining layers arranged in a certain order, the somatosensory simulation degree translation unit can perceive that the six feature mining layers arranged in a certain order respectively obtain the six second clustering key descriptions output by the six feature mining layers to perform translation. For example, the first feature extraction unit of the somatosensory simulation degree translation unit can obtain the first clustering key description output by the multiplexed feature mining layer and the second clustering key description output by the first interest evaluation description best description unit to perform translation and output the key description. The second feature extraction unit of the somatosensory simulation degree translation unit can perform translation by means of the key description output by the previous feature extraction unit and the second clustering key description output by the second interest evaluation description best description unit.

[0092] In an alternative embodiment, after the feature extraction unit of the somatosensory simulation degree translation unit performs translation by means of the first clustering key description and the second clustering key description, the somatosensory simulation degree translation unit can perform translation by means of a plurality of feature extraction units to debug the somatosensory simulation degree set finally output by the somatosensory simulation degree translation unit.

[0093] In this way, by analyzing the user experience data to be analyzed by means of the first clustering key description covering the regional distribution key semantics of the region in the user experience data to be analyzed, the regional distribution key semantics is used to improve the somatosensory simulation degree in the user experience data to be analyzed, thereby improving the analysis accuracy of the VR interaction preference.

[0094] On the basis of the above, please refer to Figure 2 , provides a VR simulation scene control device 200 based on deep learning, applied to a VR simulation scene control system based on deep learning, the device comprises:

[0095] The description content evaluation module 210 is configured to obtain user experience data to be analyzed; and obtain interest evaluation description content of the user experience data to be analyzed by means of an artificial intelligence evaluation thread.

[0096] The interaction preference analysis module 220 is configured to analyze the user experience data to be analyzed based on the interest evaluation description content by means of a user experience data analysis thread to obtain a VR interaction preference of the user experience data to be analyzed.

[0097] The control strategy determination module 230 is configured to determine a VR simulation scene control strategy for the user experience data to be analyzed based on the VR interaction preference.

[0098] Based on the above, combined with the following Figure 3 , a deep learning-based VR simulation scene control system 300 is shown, which includes a processor 310 and a memory 320 that communicate with each other, the processor 310 is used to read the computer program from the memory 320 and execute to realize the above-mentioned method.

[0099] Based on the above, a computer-readable storage medium is also provided, and the computer program stored thereon realizes the above-mentioned method when running.

[0100] In summary, based on the above scheme, by obtaining the interest evaluation description content of the user experience data to be parsed, the user experience data parsing thread can more accurately explain the description of the region in the user experience data to be parsed with the help of the interest evaluation description content, so that the VR interaction preference parsed by the user experience data parsing thread can be more perfectly associated with the region of the user experience data to be parsed, and the parsing accuracy of the VR interaction preference is improved. In addition, the interest evaluation description content of the user experience data to be parsed is obtained by means of the artificial intelligence evaluation thread of the user experience data parsing thread, and the reliable interest evaluation description content can be obtained by using the user experience data parsing thread, which further improves the correlation degree between the subsequent parsed VR interaction preference and the user experience data to be parsed, thereby ensuring the pertinence and precision of the VR simulation scene control strategy.

[0101] It should be understood that the system and its modules shown above can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software or a combination of software and hardware. The hardware part can be implemented by special logic; the software part can be stored in the memory and executed by a suitable instruction execution system, such as a microprocessor or a specially designed hardware. Those skilled in the art can understand that the above-mentioned method and system can be implemented by using computer executable instructions and / or included in processor control code, such as provided on carrier media, such as magnetic disk, CD or DVD-ROM, programmable memory, such as read-only memory (firmware), or data carrier, such as optical or electronic signal carrier. The system and its modules of the present application can not only have hardware circuit implementation, such as very large scale integrated circuit or gate array, semiconductor, such as logic chip, transistor, or programmable hardware device, such as field programmable gate array, programmable logic device, etc. It can also be implemented by software executed by various types of processors, and also by a combination of the above hardware circuit and software (for example, firmware).

[0102] It should be noted that the beneficial effects produced by different embodiments can be different, and in different embodiments, the beneficial effects that can be produced can be any one or a combination of the above, or any other beneficial effects that can be obtained.

[0103] The foregoing detailed description has set forth various embodiments of the application via the use of specific terminology. As such, it is to be understood that the term "one embodiment" or "an embodiment" or "some embodiments" or "one alternative" or "an alternative" or "some alternatives" is a description of certain examples of the application. It is not intended that the application be limited to any single embodiment or combination of embodiments described herein. It is to be understood that various modifications, improvements, and / or alterations can be made to the application without departing from the spirit and scope of the application. Such modifications, improvements, and / or alterations are intended to be within the scope of the application.

[0104] Also, the use of "for example," "e.g.," "for instance," "such as," and "like" are merely meant to be non-limiting terms that indicate that other examples of the described item are possible. It is also to be understood that the use of certain words of description in the specification is not intended to limit the scope of the application to the specific embodiments described, but is intended to cover all alternatives and modifications as would be included within the spirit and scope of the application.

[0105] Moreover, those skilled in the art will appreciate that the various aspects of the application can be described in terms of a number of different kinds of systems or articles of manufacture, including any new and useful processes, machine, manufacture and compositions of matter, or any new and useful improvements thereof. Accordingly, the various aspects of the application can be embodied in whole or in part in hardware, software (including firmware, resident software, micro-code, etc.), or a combination thereof. The various aspects of the application can be embodied in a computer-readable medium containing computer program code, which can be executed by a computer. The computer-readable medium can be, for example, a floppy disk, a hard disk, a CD-ROM, a DVD, a RAM, a ROM, a PROM, a semiconductor memory, a magnetic tape, a punch card, or any other medium that can be read by an electronic device.

[0106] Computer storage media can include volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program code, or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, solid state drives (SSDs) that use flash memory, phase-change (PC) RAM, any other memory technology, or any other medium that can be used to store and / or transfer computer readable instructions and / or data. Computer storage media can also include, but is not limited to, any medium that facilitates transfer of a computer program from one place to another. Also, a computer storage medium can be, or be included in, a computer-readable medium.

[0107] Computer program code for carrying out operations of the various aspects described herein can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, and conventional procedural programming languages, such as the "C" programming language, Visual Basic, Fortran 2003, Perl, COBOL 2002, PHP, ABAP, dynamic programming languages, such as Python, Ruby and Groovy, or other programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any form of network, such as a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet) or within a cloud computing environment or as a service, such as Software as a Service (SaaS).

[0108] Furthermore, the order of presentation of the processing elements and sequences described is not intended to be an indication of their relative importance or a necessity in implementing processes and methods according to the application. Although several embodiments of the application have been disclosed in the foregoing disclosure, it will be understood by those of ordinary skill in the art that many modifications, both to materials and procedures, can be accomplished without departing from the scope of the application. It is hereby intended to embrace all such modifications as fall within the scope of the claims. For example, although the system components described above can be implemented by hardware devices, they can also be implemented by software solutions only, such as installing the described system on an existing server or mobile device.

[0109] Similarly, it is to be noticed that the term "comprising", used in the description, is not intended to exclude other elements or steps. Rather, it is used to indicate that the elements and / or steps that follow the comprising are included in the application. The application thus encompasses both a process and a product having the features of the application. It is also to be understood that the mention of one or more reference signs in the description and / or claims should not be construed as favouring a particular example or embodiment. The use of any reference signs is intended to be indicative of a particular feature or step with which the reference sign is associated. Although the above description has been made with reference to particular examples, it is to be understood that numerous modifications can be made without departing from the scope of the application. For example, although the system components described above can be implemented by hardware devices, they can also be implemented by software solutions only, such as installing the described system on an existing server or mobile device.

[0110] In some embodiments, numbers that describe amounts, dimensions, and so forth, are used in the description of the embodiments. It should be understood that the numerical values set forth in the detailed description are approximations that can vary depending upon the desired properties sought to be obtained by the embodiments. Unless otherwise indicated, the numerical values set forth in the detailed description are approximations that can vary depending upon the desired properties sought to be obtained by the embodiments. At the very least, it should be understood that all numerical values are approximations and are intended to be modified in all instances by the expressive context found in the specification as well as by the disclosure as a whole. As used herein, the indefinite articles "a" and "an" are intended to mean zero or one, unless the context clearly indicates otherwise. It is further noted that the claims can be drafted to exclude any optional element. As such, any statements regarding "a" and "an" can be understood to be reciting "one or more" rather than "at least one."

[0111] Each patent, patent application, publication, document, article, book, instruction manual, and / or other material cited in this application is hereby incorporated by reference in its entirety for all purposes to the same extent as if each individual publication, document, article, book, instruction manual, and / or other material were specifically and individually indicated to be incorporated by reference in its entirety for all purposes. Except in the Examples, or where otherwise explicitly indicated, all numerical quantities in this description are meant to be interpreted in an "about" sense. When numerical lower limits and numerical upper limits are listed herein, ranges from any lower limit to any upper limit are contemplated. Where any numerical value is presented herein, it is understood that any numerical value described is approximate unless otherwise indicated. It should be understood that any numerical value, any range of values, and any other properties and / or characteristics described herein are not limited to the precise numerical values recited.

[0112] Finally, it should be understood that the embodiments described herein are merely illustrative of the principles of the embodiments. Other embodiments can be devised which fall within the scope of the present embodiments. Accordingly, the embodiments described and illustrated herein are not to be considered as limiting the scope of the present embodiments. As such, the scope of the present embodiments should be considered limited only by the claims that follow, and equivalents thereof.

[0113] The foregoing is considered as illustrative only of the principles of the embodiments. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the application to the exact construction and practice described. Accordingly, all suitable modifications and equivalents can be resorted to falling within the scope of the application.

Claims

1. A deep learning-based VR simulation scene control method, characterized in that, The method is applied to a control system and at least comprises: obtaining user experience data to be parsed; obtaining interest evaluation description content of the user experience data to be parsed by means of an artificial intelligence evaluation thread; combining the interest evaluation description content, parsing the user experience data to be parsed by means of a user experience data parsing thread to obtain a VR interaction preference of the user experience data to be parsed; determining a VR full-sense space control strategy for the user experience data to be parsed based on the VR interaction preference; the VR interaction preference comprises perception experience data; the combination of the interest evaluation description content and the parsing of the user experience data to be parsed by means of the user experience data parsing thread to obtain the VR interaction preference of the user experience data to be parsed comprises: processing the user experience data to be parsed by means of the user experience data parsing thread to obtain regional perception requirement data of the user experience data to be parsed; combining the regional perception requirement data and the interest evaluation description content to obtain the perception experience data of the user experience data to be parsed; the regional perception requirement data is an interest evaluation description optimal description set comprising optimal description content of interest evaluation description of experience events in which the user experience data to be parsed has differences, and the interest evaluation description content is an interest evaluation description cluster comprising interest evaluation description of experience events in which the user experience data to be parsed has differences; the combination of the regional perception requirement data and the interest evaluation description content to obtain the perception experience data of the user experience data to be parsed comprises binary operation processing of the interest evaluation description optimal description set and the interest evaluation description cluster to obtain the perception experience data of the user experience data to be parsed.

2. The method of claim 1, wherein, the user experience data parsing thread comprises a multiplex feature mining layer and a perception translation unit; the processing of the user experience data to be parsed by means of the user experience data parsing thread to obtain the regional perception requirement data of the user experience data to be parsed comprises: mining key semantics of the user experience data to be parsed by means of the multiplex feature mining layer to obtain user experience data key description, and clustering the user experience data key description and first regional distribution key description output by an interest evaluation description feature mining layer of the artificial intelligence evaluation thread to obtain first clustering key description; translating the first clustering key description by means of the perception translation unit to obtain the regional perception requirement data of the user experience data to be parsed.

3. The method of claim 2, wherein, the multiplex feature mining layer comprises one or more feature mining layers arranged in a certain order, and each feature mining layer comprises an interest evaluation description optimal description unit; the clustering of the user experience data key description and the first regional distribution key description output by the interest evaluation description feature mining layer of the artificial intelligence evaluation thread to obtain the first clustering key description comprises: loading the user experience data key description to a first feature mining layer; for each feature mining layer: clustering the key description output by the previous feature mining layer and the first area distribution key description by the interest evaluation description best description unit to obtain the second clustered key description of the feature mining layer; wherein the key words in the area distribution key description of each feature mining layer are different in integrity; based on the second clustered key description of the last feature mining layer to obtain the first clustered key description; wherein, before the clustering of the key description output by the previous feature mining layer and the area distribution key description by the interest evaluation description best description unit to obtain the second clustered key description of the feature mining layer, the method further comprises: compressing the key description output by the previous feature mining layer; and / or the clustering of the key description output by the previous feature mining layer and the area distribution key description by the interest evaluation description best description unit to obtain the second clustered key description of the feature mining layer comprises: adjusting the area distribution key description to the area distribution key description within the previous perception range by the interest evaluation description best description unit, integrating the adjusted area distribution key description with the key description output by the previous feature mining layer, and extracting features to obtain the second clustered key description of the feature mining layer; wherein, the translation of the first clustered key description by the perception degree translation unit to obtain the area perception requirement data of the user experience data to be analyzed comprises: the translation of the first clustered key description and one or more second clustered key descriptions of the interest evaluation description best description unit by the perception degree translation unit to obtain the area perception requirement data of the user experience data to be analyzed.

4. The method of claim 3, wherein, The user experience data analysis thread further comprises a somatosensory simulation degree translation unit; and the analysis of the user experience data to be analyzed by the user experience data analysis thread to obtain the VR interaction preference of the user experience data to be analyzed based on the interest evaluation description content further comprises: the translation of the first clustered key description by the somatosensory simulation degree translation unit to obtain the somatosensory simulation degree user experience data of the user experience data to be analyzed.

5. The method of claim 4, wherein, The translation of the first clustered key description by the somatosensory simulation degree translation unit to obtain the somatosensory simulation degree user experience data of the user experience data to be analyzed comprises: the translation of the first clustered key description and one or more second clustered key descriptions of the interest evaluation description best description unit by the somatosensory simulation degree translation unit to obtain the somatosensory simulation degree user experience data of the user experience data to be analyzed.

6. The method of claim 5, wherein, The artificial intelligence evaluation thread comprises an interest evaluation description feature mining layer, an interest evaluation description translation unit and a CNN thread; the interest evaluation description content of the user experience data to be analyzed is obtained by means of the artificial intelligence evaluation thread, which comprises: The feature mining layer is used to mine the features of the user experience data to be analyzed to obtain first regional distribution key descriptions; The interest evaluation description translation unit is used to translate the first regional distribution key descriptions to obtain translated key descriptions; The CNN thread is used to cluster the first regional distribution key descriptions and the translated key descriptions to obtain the interest evaluation description content of the user experience data to be analyzed.

7. The method of claim 6, wherein, The artificial intelligence evaluation thread and the user experience data analysis thread are configured respectively; Before the interest evaluation description content of the user experience data to be analyzed is obtained by means of the artificial intelligence evaluation thread, the method further comprises: obtaining the artificial intelligence evaluation thread by means of a first example cluster, wherein the user experience data in the first example cluster is labeled with interest evaluation description content; obtaining example interest evaluation description content of user experience data in a second example cluster by means of the configured artificial intelligence evaluation thread, and configuring the user experience data analysis thread by means of the second example cluster and the example interest evaluation description content; The second example cluster comprises a first local example cluster and a second local example cluster, and the user experience data analysis thread is configured by means of the second example cluster and the example interest evaluation description content, which comprises: The user experience data analysis thread is configured by means of the first local example cluster and the example interest evaluation description content bound to the first local example cluster, so as to debug the variables of the multiplex feature mining layer and the perceptual degree translation unit in the user experience data analysis thread; The user experience data analysis thread is configured by means of the second local example cluster and the example interest evaluation description content bound to the second local example cluster, so as to debug the variables of the multiplex feature mining layer and the somatosensory simulation degree translation unit in the user experience data analysis thread. 8.A deep learning based VR simulation scene control system, characterized in that, The processor and the memory are in communication with each other, the processor is used to call a computer program from the memory, and the method of any one of claims 1-7 is realized by running the computer program.

Images