Model data processing method and system based on multi-identification network

Abstract

The invention discloses a model data processing method and system based on a multi-identification network. The method comprises the following steps: acquiring model data uploaded by a target user and corresponding slice data; determining three-dimensional cavity data corresponding to the model data based on a three-dimensional identification neural network; determining slice cavity data corresponding to the slice data based on a two-dimensional recognition neural network; and according to the three-dimensional cavity data and the slice cavity data, analyzing a problem cavity position and a cavity danger degree corresponding to the model data. Therefore, two-dimensional and three-dimensional cavity detection can be fully combined to improve the cavity detection precision, the printing stability of a user model and the intelligent degree of printing service are improved, and better use experience is provided for a user.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a model data processing method and system based on a multi-recognition network. Background Art

[0002] With the development of FDM technology, more and more users or organizations have begun to pay attention to technical issues such as the management and transmission of 3D printing models. Among them, how to effectively identify and remind users of problems such as cavities in the model slice data on the platform is an important user demand. However, most of the existing model data processing technologies are still based only on two-dimensional slice recognition neural networks to identify cavity problems in the model data uploaded by users. There is no further combination of three-dimensional and two-dimensional recognition to improve the recognition accuracy of cavities and analyze possible problem cavities in the model. Therefore, the user's model stability cannot be improved, and the intelligence level of 3D printing services is also lacking. It can be seen that the existing technology has defects that need to be solved urgently. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a model data processing method and system based on a multi-recognition network, which can fully combine two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence of the printing service, and give users a better user experience.

[0004] In order to solve the above technical problems, the first aspect of the present invention discloses a model data processing method based on a multi-recognition network, the method comprising: Obtain the model data and corresponding slice data uploaded by the target user; Determining three-dimensional cavity data corresponding to the model data based on a three-dimensional recognition neural network; Determining slice cavity data corresponding to the slice data based on a two-dimensional recognition neural network; The problematic cavity position and cavity danger level corresponding to the model data are analyzed according to the three-dimensional cavity data and the sliced ​​cavity data.

[0005] As an optional embodiment, in the first aspect of the present invention, the three-dimensional recognition neural network includes a three-dimensional model vectorization network, a three-dimensional feature extraction network and a cavity recognition classification network connected in sequence, and the three-dimensional recognition neural network is trained by a training data set including multiple training three-dimensional models and corresponding cavity position annotations.

[0006] As an alternative implementation, in the first aspect of the present invention, the 3D model vectorization network is used to classify different 3D parts of the model data to obtain multiple 3D data parts of different types, and each of the 3D data parts is input into a vectorization network module corresponding to the type to obtain corresponding vectorized data; the 3D feature extraction network is used to extract data features in the vectorized data and input the data features into the cavity recognition and classification network; the cavity recognition and classification network is used to identify the cavity type and corresponding prediction probability corresponding to the data features.

[0007] As an alternative implementation, in the first aspect of the present invention, determining the 3D cavity data corresponding to the model data based on the 3D recognition neural network includes: Inputting the model data into the 3D recognition neural network to obtain the cavity type and prediction probability corresponding to different data parts of the model data; For any two adjacent data parts, adjusting the prediction probability based on the cavity type to obtain the adjusted probability corresponding to each data part; Screening out all the data parts with the adjusted probability greater than the probability threshold to obtain multiple cavity parts; Determining the multiple cavity parts as the 3D cavity data corresponding to the model data.

[0008] As an alternative implementation, in the first aspect of the present invention, for any two adjacent data parts, adjusting the prediction probability based on the cavity type to obtain the adjusted probability corresponding to each data part includes: For any two adjacent data parts, determining the cavity types corresponding to the two data parts; the cavity types include at least one of full cavity, partial cavity, multi-bubble cavity, transverse cavity, longitudinal cavity, and strip cavity; According to the preset historical adjacent cavity type record, calculating the proportion of the occurrence times corresponding to the cavity types of the two data parts; the proportion of the occurrence times is the ratio of the number of records in which the cavity types corresponding to the two data parts exist simultaneously in the historical adjacent cavity type record to the total number of records; Calculating a probability weight proportional to the proportion of the occurrence times; Calculating the product of the prediction probability corresponding to the two data parts and the probability weight to obtain the adjusted probability corresponding to the two data parts.

[0009] As an alternative implementation, in the first aspect of the present invention, determining the slice cavity data corresponding to the slice data based on the two-dimensional recognition neural network includes: Input each slice part in the slice data into the trained two-dimensional recognition neural network to obtain the cavity region and cavity type corresponding to each slice part; the two-dimensional recognition neural network is trained by a training data set including a plurality of training slice data and corresponding cavity region annotations; For any plurality of continuously adjacent slice parts, adjust the cavity region corresponding to each slice part according to the continuity rule of the cavity region to obtain an adjusted cavity region; Determine the adjusted cavity region corresponding to each slice part as the slice cavity data corresponding to the slice data.

[0010] As an alternative implementation, in the first aspect of the present invention, for any plurality of continuously adjacent slice parts, adjusting the cavity region corresponding to each slice part according to the continuity rule of the cavity region to obtain an adjusted cavity region includes: For all slice parts, based on a preset rule of continuous change of the cavity region, filter out at least one slice set that conforms to the rule; the slice set includes a plurality of continuously adjacent slice parts whose cavity regions conform to the rule; For any two adjacent slice parts in each slice set, calculate the regional difference degree between the cavity regions of the two slice parts; Calculate the regional difference degrees between the two slice parts and another adjacent slice part respectively corresponding to the two slice parts to obtain the adjacent difference degrees corresponding to the two slice parts respectively; Calculate the average value of the adjacent difference degrees corresponding to the two slice parts respectively to obtain a reference difference degree; Calculate the difference between the regional difference degree and the reference difference degree to obtain a correction value; Input the correction value and the cavity regions corresponding to the two slice parts into the trained correction neural network to obtain the adjusted cavity regions corresponding to the two slice parts respectively; the correction neural network is trained by a training data set including a plurality of training cavity regions and corresponding correction value annotations and corrected region annotations.

[0011] As an alternative implementation, in the first aspect of the present invention, analyzing the problem cavity position and cavity danger degree corresponding to the model data according to the three-dimensional cavity data and the slice cavity data includes: Calculate the intersection part of the three-dimensional cavity data and the sliced cavity data to obtain multiple intersection cavity parts; For any multiple adjacent intersection cavity parts, input the positions and corresponding cavity types of the multiple intersection cavity parts into the trained hazard determination classification model to obtain the predicted hazard level corresponding to each intersection cavity part; the hazard determination classification model is trained through a training data set including multiple training cavity positions and corresponding cavity type annotations and hazard level annotations; Calculate the average value of all the predicted hazard levels corresponding to each intersection cavity part to obtain the position hazard level corresponding to each intersection cavity part; Screen out all the intersection cavity parts whose position hazard level is greater than the hazard level threshold to obtain multiple problem cavity positions.

[0012] A second aspect of the embodiments of the present invention discloses a model data processing system based on a multi-recognition network, the system includes: An acquisition module, configured to acquire model data uploaded by a target user and corresponding sliced data; A first determination module, configured to determine three-dimensional cavity data corresponding to the model data based on a three-dimensional recognition neural network; A second determination module, configured to determine sliced cavity data corresponding to the sliced data based on a two-dimensional recognition neural network; An analysis module, configured to analyze the problem cavity positions and cavity hazard degrees corresponding to the model data according to the three-dimensional cavity data and the sliced cavity data.

[0013] As an optional implementation manner, in the second aspect of the present invention, the three-dimensional recognition neural network includes a three-dimensional model vectorization network, a three-dimensional feature extraction network, and a cavity recognition classification network connected in sequence, and the three-dimensional recognition neural network is trained through a training data set including multiple training three-dimensional models and corresponding cavity position annotations.

[0014] As an optional implementation manner, in the second aspect of the present invention, the three-dimensional model vectorization network is configured to perform type division on different three-dimensional parts of the model data to obtain multiple three-dimensional data parts of different types, and input each three-dimensional data part into a vectorization network module corresponding to the type to obtain corresponding vectorized data; the three-dimensional feature extraction network is configured to extract data features in the vectorized data and input the data features into the cavity recognition classification network; the cavity recognition classification network is configured to identify the cavity type and corresponding prediction probability corresponding to the data features.

[0015] As an alternative implementation, in the second aspect of the present invention, the specific manner in which the first determination module determines the three-dimensional cavity data corresponding to the model data based on a three-dimensional recognition neural network includes: Input the model data into the three-dimensional recognition neural network to obtain the cavity type and prediction probability corresponding to different data parts of the model data; For any two adjacent data parts, adjust the prediction probability based on the cavity type to obtain the adjusted probability corresponding to each data part; Select all the data parts with the adjusted probability greater than the probability threshold to obtain a plurality of cavity parts; Determine the plurality of cavity parts as the three-dimensional cavity data corresponding to the model data.

[0016] As an alternative implementation, in the second aspect of the present invention, the specific manner in which the first determination module adjusts the prediction probability based on the cavity type for any two adjacent data parts to obtain the adjusted probability corresponding to each data part includes: For any two adjacent data parts, determine the cavity type corresponding to the two data parts; the cavity type includes at least one of a full cavity, a partial cavity, a multi-bubble cavity, a transverse cavity, a longitudinal cavity, and a strip-shaped cavity; According to the preset historical adjacent cavity type record, calculate the occurrence frequency ratio corresponding to the cavity type of the two data parts; the occurrence frequency ratio is the ratio of the number of records in which the cavity types corresponding to the two data parts exist simultaneously in the historical adjacent cavity type record to the total number of records; Calculate the probability weight proportional to the occurrence frequency ratio; Calculate the product of the prediction probability corresponding to the two data parts and the probability weight to obtain the adjusted probability corresponding to the two data parts.

[0017] As an alternative implementation, in the second aspect of the present invention, the specific manner in which the second determination module determines the slice cavity data corresponding to the slice data based on a two-dimensional recognition neural network includes: Input each slice part in the slice data into the trained two-dimensional recognition neural network to obtain the cavity region and cavity type corresponding to each slice part; the two-dimensional recognition neural network is trained by a training data set including a plurality of training slice data and corresponding cavity region annotations; For any number of continuously adjacent said slice portions, adjust the cavity region corresponding to each said slice portion according to the continuity rule of the said cavity region to obtain an adjusted cavity region; Determine the adjusted cavity region corresponding to each said slice portion as the slice cavity data corresponding to the slice data.

[0018] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the second determination module adjusts the cavity region corresponding to each said slice portion according to the continuity rule of the said cavity region for any number of continuously adjacent said slice portions to obtain an adjusted cavity region includes: For all said slice portions, based on a preset rule of continuous change of the cavity region, filter out at least one slice set that conforms to the rule; the slice set includes a plurality of continuously adjacent said slice portions whose cavity regions conform to the rule; For any two adjacent said slice portions in each said slice set, calculate the regional difference degree between the cavity regions of the two said slice portions; Calculate the regional difference degree between each of the two said slice portions and another adjacent said slice portion respectively to obtain the adjacent difference degree corresponding to each of the two said slice portions; Calculate the average value of the adjacent difference degrees corresponding to each of the two said slice portions to obtain a reference difference degree; Calculate the difference between the regional difference degree and the reference difference degree to obtain a correction value; Input the correction value and the cavity regions corresponding to the two said slice portions into a trained correction neural network to obtain the adjusted cavity regions corresponding to each of the two said slice portions; the correction neural network is trained by a training data set including a plurality of training cavity regions and corresponding correction value annotations and corrected region annotations.

[0019] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the analysis module analyzes the problem cavity position and cavity danger degree corresponding to the model data according to the three-dimensional cavity data and the slice cavity data includes: Calculate the intersection part of the three-dimensional cavity data and the slice cavity data to obtain a plurality of intersection cavity parts; For any number of adjacent said intersection cavity parts, input the positions and corresponding cavity types of the plurality of said intersection cavity parts into a trained danger determination classification model to obtain the predicted danger degree corresponding to each said intersection cavity part; the danger determination classification model is trained by a training data set including a plurality of training cavity positions and corresponding cavity type annotations and danger degree annotations. Calculate the average value of all the predicted risks corresponding to each of the intersection cavity parts to obtain the position risk corresponding to each of the intersection cavity parts; Screen out all the intersection cavity parts whose position risk is greater than the risk threshold to obtain multiple problem cavity positions.

[0020] The third aspect of the present invention discloses another model data processing system based on a multi-recognition network, and the system includes: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory and executes some or all of the steps in the model data processing method based on a multi-recognition network disclosed in the first aspect of the present invention.

[0021] The fourth aspect of the present invention discloses a computer storage medium, and the computer storage medium stores computer instructions, which are used to execute some or all of the steps in the model data processing method based on a multi-recognition network disclosed in the first aspect of the present invention when being called.

[0022] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention can determine the cavity data in the model data and the slice data based on the three-dimensional recognition neural network and the two-dimensional recognition neural network respectively, so as to comprehensively analyze the problem cavity positions and the cavity risk levels corresponding to the model data, thereby being able to fully combine the two-dimensional and three-dimensional cavity detections to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligent level of the printing service, and give the user a better use experience. Description of the Drawings

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings according to these drawings without creative efforts.

[0024] Figure 1 It is a schematic flowchart of a model data processing method based on a multi-recognition network disclosed in an embodiment of the present invention.

[0025] Figure 2 It is a schematic structural diagram of a model data processing system based on a multi-recognition network disclosed in an embodiment of the present invention.

[0026] Figure 3It is a schematic structural diagram of another model data processing system based on a multi-recognition network disclosed in an embodiment of the present invention. Detailed implementation manners

[0027] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] The terms "first", "second", etc. in the specification and claims of the present invention and the above drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or equipment.

[0029] Referring to "embodiment" herein means that a specific feature, structure or characteristic described in connection with the embodiment may be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0030] The present invention discloses a model data processing method and system based on a multi-recognition network, which can respectively determine the cavity data in the model data and the slice data based on a three-dimensional recognition neural network and a two-dimensional recognition neural network, so as to comprehensively analyze the problem cavity position and cavity danger degree corresponding to the model data, thereby being able to fully combine the two-dimensional and three-dimensional cavity detections to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience. The following will be described in detail respectively.

[0031] Embodiment 1 Please refer to Figure 1 , Figure 1 It is a schematic flowchart of a model data processing method based on a multi-recognition network disclosed in an embodiment of the present invention. Among them, Figure 1The described model data processing method based on a multi-recognition network can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). As Figure 1 shown, the model data processing method based on the multi-recognition network may include the following operations: 101. Obtain the model data uploaded by the target user and the corresponding slice data.

[0032] 102. Based on the three-dimensional recognition neural network, determine the three-dimensional cavity data corresponding to the model data. 103. Based on the two-dimensional recognition neural network, determine the slice cavity data corresponding to the slice data. 104. According to the three-dimensional cavity data and the slice cavity data, analyze the problem cavity position and cavity danger level corresponding to the model data.

[0033] It can be seen that the above-mentioned invention embodiments can respectively determine the cavity data in the model data and the slice data based on the three-dimensional recognition neural network and the two-dimensional recognition neural network, so as to comprehensively analyze the problem cavity position and cavity danger level corresponding to the model data, thereby being able to fully combine the two-dimensional and three-dimensional cavity detections to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience.

[0034] As an optional embodiment, in the above steps, the three-dimensional recognition neural network includes a three-dimensional model vectorization network, a three-dimensional feature extraction network, and a cavity recognition and classification network connected in sequence. The three-dimensional recognition neural network is trained by a training data set including multiple training three-dimensional models and corresponding cavity position annotations.

[0035] It can be seen that through the above optional embodiment, the network architecture of the three-dimensional recognition neural network is defined to accurately identify the three-dimensional cavities of the model data, assist in fully combining the two-dimensional and three-dimensional cavity detections to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience.

[0036] As an optional embodiment, in the above steps, the three-dimensional model vectorization network is used to classify different three-dimensional parts of the model data to obtain multiple three-dimensional data parts of different types, and input each three-dimensional data part into the vectorization network module corresponding to the type to obtain the corresponding vectorized data; the three-dimensional feature extraction network is used to extract the data features in the vectorized data and input the data features into the cavity recognition and classification network; the cavity recognition and classification network is used to identify the cavity type corresponding to the data features and the corresponding prediction probability.

[0037] It can be seen that through the above optional embodiments, the functions of the 3D model vectorization network, the 3D feature extraction network, and the cavity recognition and classification network are defined to accurately identify the 3D cavity types and prediction probabilities of model data, assist in realizing the cavity detection that fully combines 2D and 3D to improve the cavity detection accuracy, improve the printing stability of user models and the intelligence level of printing services, and give users a better use experience.

[0038] As an optional embodiment, in the above steps, determining the 3D cavity data corresponding to the model data based on the 3D recognition neural network includes: Inputting the model data into the 3D recognition neural network to obtain the cavity types and prediction probabilities corresponding to different data parts of the model data; For any two adjacent data parts, adjusting the prediction probability based on the cavity type to obtain the adjusted probability corresponding to each data part; Selecting all data parts with adjusted probabilities greater than the probability threshold to obtain multiple cavity parts; Determining the multiple cavity parts as the 3D cavity data corresponding to the model data.

[0039] It can be seen that through the above optional embodiments, it is possible to identify the types and probabilities of 3D cavities based on the 3D recognition neural network, and after adjusting the probabilities based on the types, screen out accurate 3D cavities, so as to accurately analyze the cavity problems of the model, assist in realizing the cavity detection that fully combines 2D and 3D to improve the cavity detection accuracy, improve the printing stability of user models and the intelligence level of printing services, and give users a better use experience.

[0040] As an optional embodiment, in the above steps, for any two adjacent data parts, adjusting the prediction probability based on the cavity type to obtain the adjusted probability corresponding to each data part includes: For any two adjacent data parts, determining the cavity types corresponding to the two data parts; optionally, the cavity types include at least one of full cavity, partial cavity, multi-bubble cavity, transverse cavity, longitudinal cavity, and long-strip cavity; According to the preset historical adjacent cavity type record, calculating the proportion of the occurrence times of the cavity types corresponding to the two data parts; optionally, the proportion of the occurrence times is the ratio of the number of records in which the cavity types corresponding to the two data parts exist simultaneously in the historical adjacent cavity type record to the total number of records; Calculating the probability weight proportional to the proportion of the occurrence times; Calculating the product of the prediction probabilities corresponding to the two data parts and the probability weight to obtain the adjusted probability corresponding to the two data parts.

[0041] It can be seen that through the above optional embodiments, the cavity probability of the data part can be adjusted based on the cavity type of the data part in the adjacent position in the historical adjacent cavity type record, so as to screen out accurate three-dimensional cavities in the subsequent process, facilitate the accurate analysis of the cavity problem of the model, assist in realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience.

[0042] As an optional embodiment, in the above steps, determining the slice cavity data corresponding to the slice data based on the two-dimensional recognition neural network includes: Input each slice part in the slice data into the trained two-dimensional recognition neural network to obtain the cavity region and cavity type corresponding to each slice part; optionally, the two-dimensional recognition neural network is trained through a training data set including multiple training slice data and corresponding cavity region annotations; For any number of continuously adjacent slice parts, adjust the cavity region corresponding to each slice part according to the continuity rule of the cavity region to obtain the adjusted cavity region; Determine the adjusted cavity region corresponding to each slice part as the slice cavity data corresponding to the slice data.

[0043] It can be seen that through the above optional embodiments, the cavity region and type of each slice can be recognized based on the trained two-dimensional recognition neural network, and more accurate two-dimensional cavity data can be obtained after region adjustment based on the continuity rule, so as to facilitate the accurate analysis of the cavity problem of the model, assist in realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience.

[0044] As an optional embodiment, in the above steps, for any number of continuously adjacent slice parts, adjusting the cavity region corresponding to each slice part according to the continuity rule of the cavity region to obtain the adjusted cavity region includes: For all slice parts, based on the preset rule of continuous change of the cavity region, screen out at least one slice set that conforms to the rule; optionally, the slice set includes multiple continuously adjacent slice parts whose cavity regions conform to the rule; For any two adjacent slice parts in each slice set, calculate the region difference degree between the cavity regions of the two slice parts; Calculate the region difference degree between each of the two slice parts and another adjacent slice part to obtain the adjacent difference degrees corresponding to the two slice parts; Calculate the average value of the adjacent difference degrees corresponding to the two slice parts to obtain the reference difference degree; Calculate the difference between the regional difference degree and the reference difference degree to obtain a correction value; Input the correction value and the cavity regions corresponding to the two slice parts into the trained correction neural network to obtain the adjusted cavity regions corresponding to the two slice parts respectively; the correction neural network is trained through a training data set including multiple training cavity regions and corresponding correction value annotations and corrected region annotations.

[0045] It can be seen that through the above optional embodiments, more accurate cavity regions can be obtained based on the correction of the difference degree between the cavity regions of the slice parts, so as to facilitate the accurate analysis of the cavity problems of the model, assist in realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better use experience.

[0046] As an optional embodiment, in the above steps, analyzing the problem cavity positions and cavity danger levels corresponding to the model data according to the three-dimensional cavity data and the sliced cavity data includes: Calculate the intersection part of the three-dimensional cavity data and the sliced cavity data to obtain multiple intersection cavity parts; For any multiple adjacent intersection cavity parts, input the positions and corresponding cavity types of the multiple intersection cavity parts into the trained risk determination classification model to obtain the predicted risk degree corresponding to each intersection cavity part; optionally, the risk determination classification model is trained through a training data set including multiple training cavity positions and corresponding cavity type annotations and risk degree annotations; Calculate the average value of all the predicted risk degrees corresponding to each intersection cavity part to obtain the position risk degree corresponding to each intersection cavity part; Screen out all the intersection cavity parts whose position risk degree is greater than the risk degree threshold to obtain multiple problem cavity positions.

[0047] It can be seen that through the above optional embodiments, risk determination can be performed according to the positions and cavity types of the intersection cavity parts of the three-dimensional cavity data and the sliced cavity data, so as to screen out accurate cavities and analyze accurate problem degrees and risk degrees, thereby realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better use experience.

[0048] Embodiment 2 Please refer to Figure 2 , Figure 2 which is a schematic structural diagram of a model data processing system based on a multi-recognition network disclosed in an embodiment of the present invention. Among them, Figure 2The described model data processing system based on a multi-recognition network can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). As Figure 2 shown, the model data processing system based on a multi-recognition network may include: An acquisition module 201, configured to acquire model data uploaded by a target user and corresponding slice data.

[0049] A first determination module 202, configured to determine three-dimensional cavity data corresponding to the model data based on a three-dimensional recognition neural network. A second determination module 203, configured to determine slice cavity data corresponding to the slice data based on a two-dimensional recognition neural network. An analysis module 204, configured to analyze a problem cavity position and a cavity risk level corresponding to the model data according to the three-dimensional cavity data and the slice cavity data.

[0050] It can be seen that the above-mentioned invention embodiments can respectively determine cavity data in model data and slice data based on a three-dimensional recognition neural network and a two-dimensional recognition neural network, so as to comprehensively analyze the problem cavity position and cavity risk level corresponding to the model data, thereby being able to fully combine two-dimensional and three-dimensional cavity detections to improve cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better usage experience.

[0051] As an optional embodiment, the three-dimensional recognition neural network includes a three-dimensional model vectorization network, a three-dimensional feature extraction network, and a cavity recognition and classification network that are connected in sequence. The three-dimensional recognition neural network is trained by a training data set including a plurality of training three-dimensional models and corresponding cavity position annotations.

[0052] It can be seen that through the above optional embodiment, the network architecture of the three-dimensional recognition neural network is defined to accurately identify the three-dimensional cavity of the model data, assisting in fully combining two-dimensional and three-dimensional cavity detections to improve cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better usage experience.

[0053] As an optional embodiment, the three-dimensional model vectorization network is configured to perform type division on different three-dimensional parts of the model data to obtain a plurality of three-dimensional data parts of different types, and input each three-dimensional data part into a vectorization network module corresponding to the type to obtain corresponding vectorized data; the three-dimensional feature extraction network is configured to extract data features in the vectorized data and input the data features into the cavity recognition and classification network; the cavity recognition and classification network is configured to identify the cavity type and corresponding prediction probability corresponding to the data features.

[0054] It can be seen that through the above optional embodiments, the functions of the 3D model vectorization network, the 3D feature extraction network, and the cavity recognition and classification network are defined to accurately identify the 3D cavity types and prediction probabilities of model data, assist in realizing the full combination of 2D and 3D cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better use experience.

[0055] As an optional embodiment, the specific manner in which the first determination module determines the 3D cavity data corresponding to the model data based on the 3D recognition neural network includes: Input the model data into the 3D recognition neural network to obtain the cavity types and prediction probabilities corresponding to different data parts of the model data; For any two adjacent data parts, adjust the prediction probability based on the cavity type to obtain the adjusted probability corresponding to each data part; Select all the data parts whose adjusted probabilities are greater than the probability threshold to obtain multiple cavity parts; Determine the multiple cavity parts as the 3D cavity data corresponding to the model data.

[0056] It can be seen that through the above optional embodiments, it is possible to identify the types and probabilities of 3D cavities based on the 3D recognition neural network, and after adjusting the probabilities based on the types, screen out the accurate 3D cavities, so as to accurately analyze the cavity problems of the model, assist in realizing the full combination of 2D and 3D cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better use experience.

[0057] As an optional embodiment, the specific manner in which the first determination module adjusts the prediction probability based on the cavity type for any two adjacent data parts to obtain the adjusted probability corresponding to each data part includes: For any two adjacent data parts, determine the cavity types corresponding to the two data parts; optionally, the cavity types include at least one of full cavity, partial cavity, multi-bubble cavity, transverse cavity, longitudinal cavity, and long-strip cavity; According to the preset historical adjacent cavity type record, calculate the proportion of the occurrence times of the cavity types corresponding to the two data parts; optionally, the proportion of the occurrence times is the ratio of the number of records in which the cavity types corresponding to the two data parts exist simultaneously in the historical adjacent cavity type record to the total number of records; Calculate the probability weight proportional to the proportion of the occurrence times; Calculate the product of the prediction probabilities corresponding to the two data parts and the probability weight to obtain the adjusted probability corresponding to the two data parts.

[0058] It can be seen that through the above optional embodiments, the cavity probability of the data part can be adjusted based on the cavity type of the data part in the adjacent position in the number of occurrences of the historical adjacent cavity type record, so as to screen out accurate three-dimensional cavities in the subsequent process, facilitate the accurate analysis of the cavity problem of the model, assist in realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience.

[0059] As an optional embodiment, the specific manner in which the second determination module determines the sliced cavity data corresponding to the sliced data based on the two-dimensional recognition neural network includes: Input each sliced part in the sliced data into the trained two-dimensional recognition neural network to obtain the cavity region and cavity type corresponding to each sliced part; optionally, the two-dimensional recognition neural network is trained through a training data set including a plurality of training sliced data and corresponding cavity region annotations; For any plurality of continuously adjacent sliced parts, adjust the cavity region corresponding to each sliced part according to the continuity rule of the cavity region to obtain the adjusted cavity region; Determine the adjusted cavity region corresponding to each sliced part as the sliced cavity data corresponding to the sliced data.

[0060] It can be seen that through the above optional embodiments, the cavity region and type of each slice can be recognized based on the trained two-dimensional recognition neural network, and more accurate two-dimensional cavity data can be obtained after region adjustment based on the continuity rule, so as to facilitate the accurate analysis of the cavity problem of the model, assist in realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give the user a better use experience.

[0061] As an optional embodiment, the specific manner in which the second determination module adjusts the cavity region corresponding to each sliced part according to the continuity rule of the cavity region for any plurality of continuously adjacent sliced parts to obtain the adjusted cavity region includes: For all sliced parts, based on the preset rule of continuous change of the cavity region, screen out at least one slice set that conforms to the rule; optionally, the slice set includes a plurality of continuously adjacent sliced parts whose cavity regions conform to the rule; For any two adjacent sliced parts in each slice set, calculate the region difference degree between the cavity regions of the two sliced parts; Calculate the region difference degree between each of the two sliced parts and another adjacent sliced part to obtain the adjacent difference degrees corresponding to the two sliced parts respectively; Calculate the average of the adjacent difference degrees corresponding to the two slice parts respectively to obtain the reference difference degree; Calculate the difference between the regional difference degree and the reference difference degree to obtain the correction value; Input the correction value and the cavity regions corresponding to the two slice parts into the trained correction neural network to obtain the adjusted cavity regions corresponding to the two slice parts respectively; the correction neural network is trained through a training data set including multiple training cavity regions and corresponding correction value annotations and corrected region annotations.

[0062] It can be seen that through the above optional embodiments, more accurate cavity regions can be obtained based on the correction of the difference degrees between the cavity regions of the slice parts, so as to facilitate the accurate analysis of the cavity problems of the model, assist in realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better use experience.

[0063] As an optional embodiment, the specific manner in which the analysis module analyzes the problem cavity positions and cavity danger levels corresponding to the model data according to the three-dimensional cavity data and the slice cavity data includes: Calculate the intersection part of the three-dimensional cavity data and the slice cavity data to obtain multiple intersection cavity parts; For any multiple adjacent intersection cavity parts, input the positions and corresponding cavity types of the multiple intersection cavity parts into the trained danger determination classification model to obtain the predicted danger level corresponding to each intersection cavity part; optionally, the danger determination classification model is trained through a training data set including multiple training cavity positions and corresponding cavity type annotations and danger level annotations; Calculate the average value of all the predicted danger levels corresponding to each intersection cavity part to obtain the position danger level corresponding to each intersection cavity part; Screen out all the intersection cavity parts whose position danger levels are greater than the danger level threshold to obtain multiple problem cavity positions.

[0064] It can be seen that through the above optional embodiments, danger determination can be performed according to the positions and cavity types of the intersection cavity parts of the three-dimensional cavity data and the slice cavity data, so as to screen out accurate cavities and analyze accurate problem degrees and danger levels, thereby realizing the full combination of two-dimensional and three-dimensional cavity detection to improve the cavity detection accuracy, improve the printing stability of the user model and the intelligence level of the printing service, and give users a better use experience.

[0065] Embodiment III Please refer to Figure 3 , Figure 3 which is another model data processing system based on a multi-recognition network disclosed in the embodiments of the present invention. Figure 3The described model data processing system based on a multi-recognition network is applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). As Figure 3 shown, the model data processing system based on a multi-recognition network may include: A memory 301 storing executable program code; A processor 302 coupled to the memory 301; Wherein, the processor 302 invokes the executable program code stored in the memory 301 to execute the steps of the model data processing method described in Embodiment 1.

[0066] Embodiment 4 An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program causes a computer to execute the steps of the model data processing method described in Embodiment 1.

[0067] Embodiment 5 An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps of the model data processing method described in Embodiment 1.

[0068] The above describes specific embodiments of this specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily have to be performed in the particular order or sequential order shown to achieve the desired result. In certain implementations, multitasking and parallel processing are also possible or may be advantageous.

[0069] The systems, devices, modules, or units illustrated in the above embodiments may be specifically implemented by a computer chip or entity, or by a product with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0070] For convenience of description, when describing the above devices, they are described separately as various units according to their functions. Of course, when implementing this specification, the functions of each unit may be implemented in one or more software and / or hardware.

[0071] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, the embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0072] This specification is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0073] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0074] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0075] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0076] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0077] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0078] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0079] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0080] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiment.

[0081] Finally, it should be noted that the model data processing method and system based on multiple recognition networks disclosed in the embodiments of the present invention only disclose the preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, rather than limiting them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features. However, these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A model data processing method based on a multi-recognition network, characterized in that The method includes: Obtaining model data uploaded by a target user and corresponding slice data; Determining three-dimensional cavity data corresponding to the model data based on a three-dimensional recognition neural network; Determining slice cavity data corresponding to the slice data based on a two-dimensional recognition neural network; Analyzing the problem cavity position and cavity danger degree corresponding to the model data according to the three-dimensional cavity data and the slice cavity data.

2. The model data processing method based on a multi-recognition network according to claim 1, wherein The three-dimensional recognition neural network includes a three-dimensional model vectorization network, a three-dimensional feature extraction network, and a cavity recognition and classification network connected in sequence. The three-dimensional recognition neural network is trained by a training data set including a plurality of training three-dimensional models and corresponding cavity position annotations.

3. The model data processing method based on a multi-recognition network according to claim 2, wherein, The three-dimensional model vectorization network is used to classify different three-dimensional parts of the model data to obtain a plurality of three-dimensional data parts of different types, and input each of the three-dimensional data parts into a vectorization network module corresponding to the type to obtain corresponding vectorized data; the three-dimensional feature extraction network is used to extract data features in the vectorized data and input the data features into the cavity recognition and classification network; the cavity recognition and classification network is used to identify the cavity type and corresponding prediction probability corresponding to the data features.

4. The model data processing method based on a multi-recognition network according to claim 3, characterized in that The determining of the three-dimensional cavity data corresponding to the model data based on the three-dimensional recognition neural network includes: Inputting the model data into the three-dimensional recognition neural network to obtain the cavity type and prediction probability corresponding to different data parts of the model data; For any two adjacent data parts, adjusting the prediction probability based on the cavity type to obtain an adjusted probability corresponding to each data part; Screening out all the data parts with the adjusted probability greater than the probability threshold to obtain a plurality of cavity parts; Determining the plurality of cavity parts as the three-dimensional cavity data corresponding to the model data.

5. The model data processing method based on a multi-recognition network according to claim 4, wherein The adjusting of the prediction probability based on the cavity type for any two adjacent data parts to obtain an adjusted probability corresponding to each data part includes: For any two adjacent data parts, determining the cavity types corresponding to the two data parts; the cavity types include at least one of full cavity, partial cavity, multi-bubble cavity, transverse cavity, longitudinal cavity, and long-strip cavity; Calculating the occurrence frequency ratio corresponding to the cavity types of the two data parts according to a preset historical adjacent cavity type record; the occurrence frequency ratio is the ratio of the number of records in which the cavity types corresponding to the two data parts exist simultaneously in the historical adjacent cavity type record to the total number of records; Calculating a probability weight proportional to the occurrence frequency ratio; Calculating the product of the prediction probability corresponding to the two data parts and the probability weight to obtain the adjusted probability corresponding to the two data parts.

6. The model data processing method based on a multi-recognition network according to claim 1, characterized in that The determining of the slice cavity data corresponding to the slice data based on the two-dimensional recognition neural network includes: Input each slice part in the slice data into a trained two-dimensional recognition neural network to obtain the cavity region and cavity type corresponding to each slice part; the two-dimensional recognition neural network is trained by a training data set including a plurality of training slice data and corresponding cavity region annotations; For any plurality of continuously adjacent slice parts, adjust the cavity region corresponding to each slice part according to the continuity rule of the cavity region to obtain an adjusted cavity region; Determine the adjusted cavity region corresponding to each slice part as the slice cavity data corresponding to the slice data.

7. The model data processing method based on a multi-recognition network according to claim 6, wherein The step of adjusting the cavity region corresponding to each slice part according to the continuity rule of the cavity region for any plurality of continuously adjacent slice parts to obtain an adjusted cavity region includes: For all slice parts, based on a preset rule of continuous change of the cavity region, screen out at least one slice set that meets the rule; the slice set includes a plurality of continuously adjacent slice parts whose cavity regions meet the rule; For any two adjacent slice parts in each slice set, calculate the regional difference degree between the cavity regions of the two slice parts; Calculate the regional difference degrees between the two slice parts and another adjacent slice part respectively corresponding to the two slice parts to obtain the adjacent difference degrees respectively corresponding to the two slice parts; Calculate the average value of the adjacent difference degrees respectively corresponding to the two slice parts to obtain a reference difference degree; Calculate the difference between the regional difference degree and the reference difference degree to obtain a correction value; Input the correction value and the cavity regions corresponding to the two slice parts into a trained correction neural network to obtain the adjusted cavity regions respectively corresponding to the two slice parts; the correction neural network is trained by a training data set including a plurality of training cavity regions and corresponding correction value annotations and corrected region annotations.

8. The model data processing method based on a multi-recognition network according to claim 1, wherein The step of analyzing the problem cavity position and cavity danger degree corresponding to the model data according to the three-dimensional cavity data and the slice cavity data includes: Calculate the intersection part of the three-dimensional cavity data and the slice cavity data to obtain a plurality of intersection cavity parts; For any plurality of adjacent intersection cavity parts, input the positions and corresponding cavity types of the plurality of intersection cavity parts into a trained risk determination classification model to obtain the predicted risk degree corresponding to each intersection cavity part; the risk determination classification model is trained by a training data set including a plurality of training cavity positions and corresponding cavity type annotations and risk degree annotations; Calculate the average value of all the predicted risk degrees corresponding to each intersection cavity part to obtain the position risk degree corresponding to each intersection cavity part; Screen out all the intersection cavity parts whose position risk degrees are greater than the risk degree threshold to obtain a plurality of problem cavity positions.

9. A model data processing system based on a multi-recognition network, characterized in that, The system includes: An acquisition module for acquiring the model data and corresponding slice data uploaded by the target user; A first determination module, configured to determine three-dimensional cavity data corresponding to the model data based on a three-dimensional recognition neural network; A second determination module, configured to determine slice cavity data corresponding to the slice data based on a two-dimensional recognition neural network; An analysis module, configured to analyze a problem cavity position and a cavity risk level corresponding to the model data according to the three-dimensional cavity data and the slice cavity data.

10. A model data processing system based on a multi-recognition network, characterized in that, The system includes: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory and executes the model data processing method based on a multi-recognition network according to any one of claims 1-8.

Images