CN109493427B

CN109493427B - Intelligent model training method and device and collision detection method and device

Info

Publication number: CN109493427B
Application number: CN201811433780.3A
Authority: CN
Inventors: 梁宇轩
Original assignee: Chengdu Xishanju Shiyou Technology Co ltd; Zhuhai Kingsoft Digital Network Technology Co Ltd
Current assignee: Chengdu Xishanju Shiyou Technology Co ltd; Zhuhai Kingsoft Digital Network Technology Co Ltd
Priority date: 2018-11-28
Filing date: 2018-11-28
Publication date: 2023-04-11
Anticipated expiration: 2038-11-28
Also published as: CN109493427A

Abstract

The intelligent model training method comprises the steps of obtaining a set of object sample sets, wherein the set of object samples comprises a plurality of groups of object sample sets and object sample set labels corresponding to the object sample sets, each group of object sample sets comprises the size of object samples, the number of the object samples and the density of the object samples, and the object sample set labels comprise a grid layer establishing method; training an intelligent model through the set of object sample sets to obtain the intelligent model, wherein the intelligent model enables the object sample sets and the object sample set labels to be associated. The intelligent model can obtain the corresponding object sample set label, namely the method for establishing the grid layer, according to the object sample set, time consumption is saved, and the precision of collision detection results of subsequent detection of the object sample set is greatly improved.

Description

Intelligent model training method and device and collision detection method and device

Technical Field

The present application relates to the field of internet technologies, and in particular, to an intelligent model training method and apparatus, a collision detection method and apparatus, a computing device, and a storage medium.

Background

Currently, collision detection is widely applied to different technical fields, such as games, animation and the like. In the prior art, the detected object is placed in a proper grid layer to be subjected to collision detection, and the method for establishing the grid layer is used for realizing the collision detection and the detection accuracy of the detected object, but the time consumption is long because a proper grid layer establishing method needs to be selected from a large number of grid layer establishing methods when the grid layer is established according to the detected object, and the selected grid layer establishing method is not necessarily optimal, so that the result of the collision detection is inaccurate, and the time consumption of the collision detection is long.

Disclosure of Invention

In view of this, embodiments of the present application provide an intelligent model training method and apparatus, a collision detection method and apparatus, a computing device, and a storage medium, so as to solve technical defects in the prior art.

In a first aspect, an embodiment of the present application discloses an intelligent model training method, including:

acquiring a set of object sample sets, wherein the set of object samples comprises a plurality of groups of object sample sets and object sample set labels corresponding to each group of object sample sets, each group of object sample sets comprises the size of object samples, the number of object samples and the density of the object samples, and the object sample set labels comprise a grid layer establishing method;

training an intelligent model through the set of object sample sets to obtain the intelligent model, wherein the intelligent model enables the object sample sets and the object sample set labels to be associated.

Optionally, before acquiring the set of object sample sets, the method further includes:

and determining the object sample set label corresponding to each group of the object sample sets.

Optionally, determining the object sample set label corresponding to each group of the object sample sets includes:

establishing at least two corresponding grid layers for each group of the object sample set according to a set comprising at least two grid layer establishing methods;

determining at least two collision relations of the object samples in each group of the object sample sets according to the at least two grid layers;

determining at least two collision detection results of the object samples in each group of the object sample sets according to the at least two collision relations, wherein the collision detection results comprise time consumed for detecting the collision of the object samples in the object sample sets once;

and selecting a grid layer establishing method corresponding to the collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set.

Optionally, after selecting a mesh layer establishing method corresponding to a collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set, the method further includes:

receiving a new mesh layer establishment method and adding the new mesh layer establishment method to the set comprising at least two mesh layer establishment methods.

Optionally, determining at least two collision relationships of the object samples in each set of object sample sets according to the at least two mesh layers includes:

determining at least two grid layer coordinates of the object samples in each set of object samples in the at least two grid layers corresponding thereto;

and determining at least two collision relations of the object samples in each group of the object sample sets according to the at least two grid layer coordinates of the object samples in each group of the object sample sets.

Optionally, the grid layer coordinates include an object granularity of the object samples in each set of object sample sets, an abscissa of the object samples in each set of object sample sets on the grid layer and an ordinate of the object samples in each set of object sample sets on the grid layer,

determining at least two grid layer coordinates of the object samples in each set of object samples at their corresponding at least two grid layers comprises:

and converting the at least two grid layers into at least two granularity coordinate systems by taking the object granularity of the object samples in each group of the object sample sets as a z-axis, the abscissa of the object samples in each group of the object sample sets on the grid layer as an x-axis and the ordinate of the object samples in each group of the object sample sets on the grid layer as a y-axis, and determining the coordinates of the object samples in each group of the object sample sets on the at least two granularity coordinate systems.

Optionally, determining at least two collision relationships of the object samples in each set of object sample sets according to the at least two grid layer coordinates of the object samples in each set of object sample sets includes:

and determining at least two collision relations of the object samples in each group of the object sample sets according to the coordinates of the object samples in each group of the object sample sets in the at least two granularity coordinate systems.

In a second aspect, an embodiment of the present application further provides a collision detection method, including:

acquiring an object set to be detected comprising at least two objects to be detected, wherein the object set to be detected comprises the size of the objects to be detected, the number of the objects to be detected and the density of the objects to be detected;

determining a grid layer establishing method for the to-be-detected object set according to a pre-generated intelligent model, and establishing a corresponding grid layer for the to-be-detected object set according to the grid layer establishing method;

and determining the collision relation of the objects to be detected in the object set to be detected according to the grid layer.

Optionally, determining the collision relationship of the objects to be detected in the object set to be detected according to the grid layer includes:

determining grid layer coordinates of the objects to be detected in the object set to be detected on the corresponding grid layer;

and determining the collision relation of the objects to be detected in the object set to be detected according to the respective grid layer coordinates of the objects to be detected in the object set to be detected.

Optionally, the coordinates of the grid layer include object granularity of the objects to be detected in the object set to be detected, the horizontal coordinates of the objects to be detected in the object set to be detected on the grid layer and the vertical coordinates of the objects to be detected in the object set to be detected on the grid layer,

determining the grid layer coordinates of the objects to be detected in the object set to be detected in the corresponding grid layer comprises:

and converting the grid layer into a granularity coordinate system by taking the object granularity of the objects to be detected in the objects to be detected set as a z-axis, the abscissa of the objects to be detected in the objects to be detected set on the grid layer as an x-axis and the ordinate of the objects to be detected in the objects to be detected set on the grid layer as a y-axis, and determining the coordinates of the objects to be detected set on the granularity coordinate system.

Optionally, determining the collision relationship of the objects to be detected in the object set to be detected according to the respective grid layer coordinates of the objects to be detected in the object set to be detected includes:

and determining the collision relation of the objects to be detected in the object set to be detected according to the coordinates of the objects to be detected in the object set to be detected in the granularity coordinate system.

Optionally, the pre-generated intelligent model includes any one of the intelligent models described above.

In a third aspect, an embodiment of the present application further provides an intelligent model training apparatus, including:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is configured to acquire a set of object sample sets, the set of object samples comprises a plurality of sets of object sample sets and object sample set labels corresponding to each set of object sample sets, each set of object sample sets comprises the size of an object sample, the number of the object samples and the concentration of the object sample, and the object sample set labels comprise a grid layer establishing method;

a training module configured to train an intelligent model through the set of object sample sets, resulting in the intelligent model, the intelligent model associating the object sample sets and the object sample set labels.

Optionally, the apparatus further comprises:

a first determination module configured to determine an object exemplar set label corresponding to each set of the object exemplar sets.

Optionally, the first determining module includes:

a first establishing sub-module configured to establish at least two corresponding mesh layers for each set of the object sample sets according to a set including at least two mesh layer establishing methods;

a second determining submodule configured to determine at least two collision relationships for the object samples in each set of object sample sets from the at least two mesh layers;

a third determining sub-module, configured to determine at least two collision detection results of the object samples in the object sample sets of each group according to the at least two collision relationships, where the collision detection results include a collision detection time consumed for detecting the object samples in the object sample sets once;

the selection submodule is configured to select a grid layer establishing method corresponding to one collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set.

Optionally, the apparatus further comprises:

an adding module configured to receive a new mesh layer establishing method and add the new mesh layer establishing method to the set comprising at least two mesh layer establishing methods.

Optionally, the second determining sub-module is further configured to:

a fifth determining submodule configured to determine at least two grid layer coordinates of the object samples in the at least two grid layers corresponding to the object samples in each set of object samples;

a sixth determining submodule configured to determine at least two collision relationships of the object samples in each set of the object sample sets according to the at least two mesh layer coordinates of the object samples in each set of the object sample sets.

the fifth determination submodule further configured to:

Optionally, the sixth determining submodule is further configured to:

In a fourth aspect, an embodiment of the present application further provides a collision detection apparatus, including:

the second acquisition module is configured to acquire an object set to be detected, which comprises at least two objects to be detected, wherein the object set to be detected comprises the size of the objects to be detected, the number of the objects to be detected and the density of the objects to be detected;

the second establishing module is configured to determine a grid layer establishing method for the to-be-detected object set according to a pre-generated intelligent model, and establish a corresponding grid layer for the to-be-detected object set according to the grid layer establishing method;

and the fourth determining module is configured to determine the collision relation of the objects to be detected in the object set to be detected according to the grid layer.

Optionally, the fourth determining module includes:

the grid layer coordinate determination submodule is configured to determine grid layer coordinates of the objects to be detected in the object set to be detected on the corresponding grid layer;

the collision relation determining submodule is configured to determine the collision relation of the objects to be detected in the object set to be detected according to the grid layer coordinates of the objects to be detected in the object set to be detected.

the mesh layer coordinate determination submodule is further configured to:

and converting the grid layer into a particle size coordinate system by taking the object particle size of the objects to be detected in the objects to be detected set as the z axis, the abscissa of the objects to be detected in the objects to be detected set on the grid layer as the x axis and the ordinate of the objects to be detected set on the grid layer as the y axis, and determining the coordinates of the objects to be detected set on the particle size coordinate system.

Optionally, the collision relation determining submodule is further configured to:

In a fifth aspect, an embodiment of the present application further provides a computing device, including a memory, a processor, and computer instructions stored on the memory and executable on the processor, where the processor executes the instructions to implement the steps of the intelligent model training method or the collision detection method as described above when the instructions are executed by the processor.

In a sixth aspect, an embodiment of the present application further provides a computer-readable storage medium storing computer instructions, which when executed by a processor, implement the steps of the intelligent model training method or the collision detection method as described above.

The application provides an intelligent model training method and device, a collision detection method and device, a computing device and a storage medium, wherein the intelligent model training method comprises the steps of obtaining a set of object sample sets, wherein the set of object samples comprises a plurality of groups of object sample sets and object sample set labels corresponding to each group of object sample sets, each group of object sample sets comprises the size of the object samples, the number of the object samples and the density of the object samples, and the object sample set labels comprise a grid layer establishing method; training an intelligent model through the set of object sample sets to obtain the intelligent model, wherein the intelligent model enables the object sample sets and the object sample set labels to be associated. The intelligent model can obtain the corresponding object sample set label, namely the grid layer establishing method, according to the object sample set, time consumption is saved, the grid layer establishing method obtained according to the object sample set is better through the machine learning method, and the collision detection result precision of detecting the object sample set by adopting the grid layer establishing method in the follow-up process is greatly improved.

Drawings

FIG. 1 is a block diagram of a computing device, according to one or more embodiments of the present disclosure;

FIG. 2 is a flow diagram of a method for training an intelligent model according to one or more embodiments of the present disclosure;

FIG. 3 is a flow diagram of a method for intelligent model training provided in one or more embodiments of the present disclosure;

FIG. 4 is a schematic diagram illustrating a method for determining multiple collision detection results for an object sample set according to multiple mesh layer building methods in an intelligent model training method provided in one or more embodiments of the present disclosure;

FIG. 5 is a flow diagram of a method for training an intelligent model according to one or more embodiments of the present disclosure;

FIG. 6 is a schematic structural diagram of a granular coordinate system of an intelligent model training method according to one or more embodiments of the present disclosure;

FIG. 7 is a structural diagram illustrating a collision relationship of an object sample in a multi-layer grid layer of an intelligent model training method according to one or more embodiments of the present disclosure;

FIG. 8 is a flow diagram of a collision detection method provided in one or more embodiments of the present description;

FIG. 9 is a schematic structural diagram of an intelligent model training device according to one or more embodiments of the present disclosure;

fig. 10 is a schematic structural diagram of a collision detection device according to one or more embodiments of the present disclosure.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit and scope of this application, and thus this application is not limited to the specific implementations disclosed below.

The terminology used in the description of the one or more embodiments is for the purpose of describing the particular embodiments only and is not intended to be limiting of the description of the one or more embodiments. As used in this specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It will be understood that, although the terms first, second, etc. may be used herein in one or more embodiments to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first can also be referred to as a second and, similarly, a second can also be referred to as a first without departing from the scope of one or more embodiments of the present description. The word "if," as used herein, may be interpreted as "at \8230; \8230when" or "when 8230; \823030when" or "in response to a determination," depending on the context.

First, the noun terms to which one or more embodiments of the present invention relate are explained.

Doorkeeper: the chinese language in the embodiment of the present specification is interpreted as "watchdog", and is a grid layer determination system that determines the number of grid layers corresponding to an entering object and the grid granularity of the grid layers by monitoring the object entering the system. The Doorkeeper records the sizes of all objects passing through the system, counts the sizes of the objects, and determines the number of layers of the grid layers to be established and the grid granularity of each grid layer by utilizing the algorithm of the Doorkeeper to the network layers.

In the present application, an intelligent model training method and apparatus, a collision detection method and apparatus, a computing device, and a storage medium are provided, which are described in detail in the following embodiments one by one.

Fig. 1 is a block diagram illustrating a configuration of a computing device 100 according to an embodiment of the present specification. The components of the computing device 100 include, but are not limited to, memory 110 and processor 120. The processor 120 is coupled to the memory 110 via a bus 130 and a database 150 is used to store data.

Computing device 100 also includes access device 140, access device 140 enabling computing device 100 to communicate via one or more networks 160. Examples of such networks include the Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. Access device 140 may include one or more of any type of network interface (e.g., a Network Interface Card (NIC)) whether wired or wireless, such as an IEEE802.11 Wireless Local Area Network (WLAN) wireless interface, a worldwide interoperability for microwave access (Wi-MAX) interface, an ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.

In one embodiment of the present description, the other components of the computing device 100 described above and not shown in FIG. 1 may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device architecture shown in FIG. 1 is for purposes of example only and is not limiting as to the scope of the description. Those skilled in the art may add or replace other components as desired.

Computing device 100 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), a mobile phone (e.g., smartphone), a wearable computing device (e.g., smartwatch, smartglasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or PC. Computing device 100 may also be a mobile or stationary server.

Wherein the processor 120 may perform the steps of the method shown in fig. 2. FIG. 2 is a schematic flow chart diagram illustrating an intelligent model training method according to an embodiment of the present application, including steps 202-204.

Step 202: the method comprises the steps of obtaining a set of object sample sets, wherein the set of object samples comprises a plurality of groups of object sample sets and object sample set labels corresponding to the object sample sets, each group of object sample sets comprises the size of object samples, the number of the object samples and the concentration of the object samples, and the object sample set labels comprise a grid layer establishing method.

In one or more embodiments of the present specification, the set of object sample sets includes at least two sets of object sample sets, each set of object sample sets includes at least two object samples, and the object samples include, but are not limited to, regular or irregular figures in a game scene, such as characters or attack props.

Each set of the object sample sets includes the size of the object samples, the number of the object samples, the concentration of the object samples, i.e., the size of each object sample in each set of the object sample sets, the total number of the object samples, and the concentration of the object samples in the set of the object samples.

In one or more embodiments of the present specification, the object sample sets and each group of object sample sets are multiple history object sets that pass through a Doorkeeper, the corresponding object sample set labels establish a grid layer for each history object set through multiple existing grid layer establishing methods in the Doorkeeper, then a collision detection result is obtained according to a collision relationship of the history object sets in the grid layer, and finally an optimal object sample set label corresponding to each history object set is determined according to the collision detection result.

Step 204: training an intelligent model through the set of object sample sets to obtain the intelligent model, wherein the intelligent model enables the object sample sets and the object sample set labels to be associated.

In one or more embodiments of the present description, before acquiring the set of object sample sets, the method further includes:

Wherein, determining the object sample set label corresponding to each group of the object sample sets comprises steps 302 to 308.

Step 302: and establishing at least two corresponding grid layers for each group of the object sample set according to a set comprising at least two grid layer establishing methods.

In one or more embodiments of the present specification, each mesh layer is a mesh layer with a corresponding number of layers and a corresponding granularity, which is established according to the mesh layer establishing method. For example, the first mesh layer is a mesh layer with 4 layers and a mesh granularity a, the second mesh layer is a mesh layer with 2 layers and a mesh granularity b, and the like. The specific type of mesh layer is determined according to the existing mesh layer establishing method, which is not limited in this application.

In one or more embodiments of the present disclosure, if the set including at least two grid layer establishing methods includes 3 grid layer establishing methods, each grid layer establishing method may establish a corresponding grid layer for each group of sample sets; then at least two corresponding grid layers are established for each group of the object sample sets according to a set comprising at least two grid layer establishing methods, that is, 3 corresponding grid layers are established for each group of the object sample sets according to the 3 grid layer establishing methods.

Step 304: and determining at least two collision relations of the object samples in each group of the object sample sets according to the at least two grid layers.

In one or more embodiments of the present disclosure, 3 grid layers are illustrated for each set of the object sample sets.

And determining at least two collision relations of the object samples in each group of the object sample sets according to the at least two grid layers, namely determining 3 collision relations of the object samples in each group of the object sample sets according to the 3 grid layers.

Step 306: and determining at least two collision detection results of the object samples in each group of the object sample sets according to the at least two collision relations, wherein the collision detection results comprise the time consumed for detecting the collision of the object samples in the object sample sets once.

In one or more embodiments of the present disclosure, 3 collision relationships corresponding to each set of the object sample sets are taken as an example for description.

And determining at least two collision detection results of the object samples in each group of the object sample sets according to the at least two collision relations, namely determining 3 collision detection results of the object samples in each group of the object sample sets according to the 3 collision relations.

In one or more embodiments of the present disclosure, the collision detection result may also be accuracy of object samples collected by detected object samples, and the like, which is not limited in this application.

Step 308: and selecting a grid layer establishing method corresponding to the collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set.

In one or more embodiments of the present specification, 3 types of collision detection results are exemplified for description.

And selecting a grid layer establishing method corresponding to the collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set, namely selecting a grid layer establishing method corresponding to the collision detection result with the shortest time consumption from the 3 collision detection results as an object sample set label corresponding to the object sample set.

In one or more embodiments of the present specification, after selecting a mesh layer establishing method corresponding to a collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set, the method further includes:

The method comprises the steps of receiving a new grid layer establishing method, adding the new grid layer establishing method to a set comprising at least two grid layer establishing methods, establishing a new grid layer for each group of object sample sets by the new grid layer establishing method, determining a new collision relation of the object samples in each group of object sample sets according to the new grid layer, determining a new collision detection result of the object samples in each group of object sample sets according to the new collision relation, comparing the new collision detection result with a previous collision detection result, reselecting a grid layer establishing method corresponding to the collision detection result as an object sample set label corresponding to the object sample set, and achieving continuous optimization of a trained intelligent model.

Referring to fig. 4, taking a group of object sample sets a as an example, the determination of the object sample set label corresponding to each group of object sample sets is described in detail.

If 3 mesh layer establishing methods exist: a mesh layer establishing method 1, a mesh layer establishing method 2, and a mesh layer establishing method 3, respectively establishing 3 corresponding mesh layers for the object sample set a according to the 3 mesh layer establishing methods: the device comprises a grid layer 1, a grid layer 2 and a grid layer 3, wherein the number of the grid layers and the grid granularity of the grid layer 1, the grid layer 2 and the grid layer 3 are determined according to the size of object samples in an object sample set A, the number of the object samples and the density of the object samples;

then determining 3 collision relations of the object samples in the object sample set A according to the grid layer 1, the grid layer 2 and the grid layer 3, namely putting the object samples in the object sample set A into the grid layer 1 to obtain the collision relation 1, putting the object samples in the object sample set A into the grid layer 2 to obtain the collision relation 2, putting the object samples in the object sample set A into the grid layer 3 to obtain the collision relation 3, wherein the collision relation is who the object samples in the object sample set A collide with whom;

and determining 3 collision detection results of the object samples in the object sample set A according to the collision relation 1, the collision relation 2 and the collision relation 3, namely obtaining the collision detection result 1 of the object samples in the object sample set A according to the collision relation 1, obtaining the collision detection result 2 of the object samples in the object sample set A according to the collision relation 2, and obtaining the collision detection result 3 of the object samples in the object sample set A according to the collision relation 3, wherein the value of the collision detection result 1 is time-consuming 3s, the value of the collision detection result 2 is time-consuming 2s, and the value of the collision detection result 3 is time-consuming 5s.

And finally, selecting the grid layer establishing method 2 corresponding to the collision detection result 2 which consumes the least time from the collision detection result 1, the collision detection result 2 and the collision detection result 3 as an object sample set label corresponding to the object sample set A.

If a grid layer establishing method 4 and a grid layer establishing method 5 exist, the grid layer establishing method 4 and the grid layer establishing method 5 are added into a set of previous grid layer establishing methods, then the grid layer 4 and the grid layer 5 are established for an object sample set A according to the grid layer establishing method 4 and the grid layer establishing method 5, then a collision relation 4 and a collision relation 5 of the object samples in the object sample set A are determined according to the grid layer 4 and the grid layer 5, then a collision detection result 4 and a collision detection result 5 of the object samples in the object sample set A are determined according to the collision relation 4 and the collision relation 5, wherein the value of the collision detection result 4 is time-consuming 1s, the value of the collision detection result 5 is time-consuming 3s, then the collision detection result 4 and the collision detection result 5 are compared with the previous collision detection result, and the grid layer establishing method 4 corresponding to the collision detection result 4 with the lowest time-consuming is selected again as an object sample set label corresponding to the object sample set A.

In practical applications, the selection of the collision detection result may be selected according to practical requirements, including but not limited to selecting the collision detection result with a maximum value, a minimum value or an intermediate value, which is not limited in this application.

In one or more embodiments of the present specification, the intelligent model training method is adopted, which includes collecting a large number of object sample sets and a grid layer establishing method, detecting a collision relation of each group of object sample sets according to the collected large number of grid layer establishing methods, obtaining a collision detection result according to the collision relation, selecting an optimal grid layer establishing method for each group of object sample sets, performing machine learning training on an intelligent model according to the object sample sets and a corresponding grid layer establishing method, and continuously and automatically adjusting and optimizing labels corresponding to the object sample sets according to a new grid layer establishing method, so that the intelligent model trained by the method can quickly select the optimal grid layer establishing method for the object sample sets in subsequent applications.

Referring to fig. 5, one or more embodiments of the present specification provide a method for training an intelligent model, where determining an object sample set label corresponding to each group of the object sample sets includes steps 502 to 510.

Step 502: and establishing at least two corresponding grid layers for each group of the object sample set according to a set comprising at least two grid layer establishing methods.

Step 504: determining at least two grid layer coordinates of the object samples in each set of object sample sets at the at least two grid layers corresponding thereto.

In one or more embodiments of the present description, the grid layer coordinates include an object granularity for the object samples in each of the object sample sets, an abscissa of the object samples in each of the object sample sets on the grid layer and an ordinate of the object samples in each of the object sample sets on the grid layer,

Referring to fig. 6, taking that the object sample set B includes 2 object samples, and corresponds to one of the grid layers as an example, coordinates of the object samples in the object sample set B in a granularity coordinate system are described.

Firstly, converting the 2 layers of grid layers into a granularity coordinate system, wherein the granularity of the object is taken as a z-axis in the granularity coordinate system, the position of the object sample on a granularity plane is represented by an x-axis and a y-axis, and the coordinate of the central point of the object sample on the granularity plane is represented by the coordinate of the central point of the object sample on the granularity plane. As can be seen from the granularity coordinate system, the coordinate of the object sample a in the lower grid layer in the granularity coordinate system is [ 6.5,5.5,2 ], that is, the object sample is on a plane with granularity of 2mm, and the coordinate thereof on the plane is (6.5,5.5). The coordinate of the object sample B in the upper grid layer in the grain size coordinate system is [ 6.5,5.5,4 ], that is, the coordinate of the object sample on the plane with the grain size of 4mm is (6.5,5.5). Then, by mapping object a to the upper grid layer, the mapping of object a overlaps with object B; or by mapping the object B to the underlying mesh layer, the mapping of the object B overlaps with the object a, whereby it can be determined that the object a and the object B are bound to collide.

Step 506: determining at least two collision relationships of the object samples in each set of object sample sets according to the at least two grid layer coordinates of the object samples in each set of object sample sets.

In one or more embodiments of the present disclosure, at least two collision relationships of the object samples in each set of object sample sets are determined according to the at least two grid-layer coordinates of the object samples in each set of object sample sets, that is, at least two collision relationships of the object samples in each set of object sample sets are determined according to the coordinates of the object samples in each set of object sample sets in the at least two granular coordinate systems.

Wherein determining at least two collision relationships for the object samples in each set of object samples according to the at least two grid layer coordinates for the object samples in each set of object samples comprises:

for the object samples located on the same grid layer, determining a first collision relation of the object samples in each group of object sample sets in the x-axis and y-axis coordinates of the object samples in each group of object sample sets in the granularity coordinate system;

for the objects to be detected positioned on different grid layers, determining a second collision relation of the object samples in each group of object sample sets according to the x-axis, y-axis and z-axis coordinates of the object samples in each group of object sample sets in the granularity coordinate system;

and determining at least two collision relations of the object samples in each group of the object sample sets according to the first collision relation and the second collision relation.

In actual use, firstly, each group of the object sample concentrated object samples are placed into the matched multilayer grid layers according to the object granularity, then the object samples are subjected to collision relation detection in each grid layer to obtain a first collision detection result, then the object samples in the multiple grid layers are subjected to cross-layer collision detection, the collision relation of the object samples in the adjacent grid layers is detected to obtain a second collision detection result, and finally the final collision relation detection result of the object samples in the multilayer grid layers can be obtained from the first collision detection result and the second collision detection result. The object samples in each group of the object sample sets correspond to different established grid layers, so that the object samples in each group of the object sample sets correspond to a final collision relation detection result, and if there are multiple groups of the object samples in the object sample sets, multiple corresponding final collision relation detection results are generated.

Referring to fig. 7, the detection of the collision relationship of the object samples in the object sample set is described by taking a multi-layer grid layer corresponding to the object samples in one of the object sample sets as an example.

If the multi-layer mesh layer includes layers a, b, and c, the collision relation detection method of the object sample will be described by taking three mesh layers a, b, and c as examples.

In fig. 7, the mesh layer a1 is a map of the mesh layer a.

The detection of the collision relation of the object sample comprises two modes, one mode is detection towards the direction of a father layer according to the coordinates of a grid layer with smaller object granularity and the coordinates of a grid layer with larger object granularity at the previous layer, and the detection is called upward cross-layer detection; the other is detection in the sub-layer direction according to the coordinates of the grid layer with larger object granularity and the coordinates of the grid layer with smaller object granularity at the next layer, which is called as cross-layer detection.

The upward cross-layer detection mode specifically comprises the following steps: and mapping the object 1 in the grid layer b into a certain coordinate identical area 11 in the grid layer a, and detecting which objects in the grid layer a collide with the object 1. Firstly, the object 1 inevitably collides with the object in the area 11 with the same coordinate in the grid layer a, and then collision detection is performed on the object 1 and the objects in 8 gray grid sub-areas around the area 11 with the same coordinate in the grid layer a according to the same-layer detection method.

The downward cross-layer detection mode specifically comprises the following steps: mapping a rectangular ABCD area corresponding to one object in the grid layer b into the grid layer c, and then determining an object detection area in the grid layer c according to the rectangular ABCD area: the rectangle a 'B' C 'D', wherein the area of the rectangle a 'B' C 'D' is obtained by outwardly expanding a sub-layer partition size at the boundary of the rectangular ABCD area mapped on the mesh layer C, and the sub-layer partition size is the size of a mesh of the mesh layer C.

In the cross-layer down detection mode, it is also necessary to detect which objects in the mesh layer c the objects mapped in the rectangular ABCD area of the mesh layer c will collide with. Firstly, the object in the rectangular ABCD area will inevitably collide with the object in the rectangular a 'B' C 'D' area, and then the object in the rectangular ABCD area and the objects in the other gray grid areas except the corresponding rectangular ABCD area in the rectangular a 'B' C 'D' area are subjected to collision detection according to the existing peer detection method.

In practical application, when performing collision detection on object samples in a single-layer grid layer, a hash structure is adopted, one object sample occupies a nine-grid detection mode, but the grid layers in the embodiment of the specification are multiple, and the division granularity of each object sample is different, the spatial grid granularity of each layer is different like a quad-tree, then the object samples with different object granularity sizes are placed in the grid layers with different grid granularities, when detecting the collision relation of the object samples, the nine-grid mode is used for detecting the collision relation among the object samples of the same layer, and the similar quad-tree mode is used for detecting the collision relation among the object samples of different layers.

Therefore, the operation of removing the duplicate is completely avoided, and because the object samples with different sizes are placed into different grid layers, the overlapping detection among the object samples on the same layer is avoided, and the screening precision can be always kept at certain high precision.

In one or more embodiments of the present disclosure, each collision relation of the object samples in each set of the object sample sets is obtained by the above method.

Step 508: and determining at least two collision detection results of the object samples in each group of the object sample sets according to the at least two collision relations, wherein the collision detection results comprise the time consumed for detecting the collision of the object samples in the object sample sets once.

Step 510: and selecting a grid layer establishing method corresponding to the collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set.

In one or more embodiments of the present disclosure, a corresponding grid layer is established for each set of object sample sets according to a preset grid layer establishing method, then the grid layer is converted into a three-dimensional granularity coordinate system, and then coordinates of the object samples in each set of object sample sets in the corresponding grid layer are determined according to the granularity coordinate system, so that a collision relationship of the object samples in each set of object sample sets is determined more accurately according to the grid layer coordinates of the object samples in each set of object sample sets, an accurate collision detection result is obtained according to the collision relationship, and then an optimal object sample set label is selected for the object sample sets according to the accurate collision detection result.

Processor 120 may perform the steps in the method shown in fig. 8. Fig. 8 is a schematic flow chart diagram illustrating a collision detection method according to an embodiment of the specification, including steps 802 through 806.

Step 802: the method comprises the steps of obtaining an object set to be detected comprising at least two objects to be detected, wherein the object set to be detected comprises the size of the objects to be detected, the number of the objects to be detected and the density of the objects to be detected.

In one or more embodiments of the present specification, the number of the objects to be detected in the object set to be detected is at least two, and may also be three or more, and the object set to be detected with a suitable number of the objects to be detected is obtained according to practical applications, which is not limited in this application.

The object to be detected includes, but is not limited to, a regular or irregular figure in a game scene, such as a character or an attack prop.

Step 804: and determining a grid layer establishing method for the to-be-detected object set according to a pre-generated intelligent model, and establishing a corresponding grid layer for the to-be-detected object set according to the grid layer establishing method.

In one or more embodiments of the present specification, the pre-generated intelligent model is a trained intelligent model in the above embodiments.

Step 806: and determining the collision relation of the objects to be detected in the object set to be detected according to the grid layer.

In one or more embodiments of the present specification, determining, according to the grid layer, a collision relationship of the objects to be detected in the set of objects to be detected includes:

Wherein the grid layer coordinates comprise the object granularity of the objects to be detected in the object set to be detected, the horizontal coordinates of the objects to be detected in the object set to be detected on the grid layer and the vertical coordinates of the objects to be detected in the object set to be detected on the grid layer,

In one or more embodiments of the present specification, determining the collision relationship of the objects to be detected in the object set to be detected according to the grid layer coordinates of the respective objects to be detected in the object set to be detected includes:

In one or more embodiments of the present specification, reference may be made to the above-mentioned embodiments for a method for determining a collision relationship between objects to be detected in the set of objects to be detected specifically according to the grid layer, and details are not described here.

In one or more embodiments of the present specification, after an object set to be detected is obtained, the collision detection method determines an optimal mesh layer establishment method for the object set to be detected through an intelligent model, then places the objects to be detected in the object set to be detected into the mesh layer established by the optimal mesh layer establishment method, and then accurately detects a collision relationship between the objects to be detected in the object set to be detected, which not only improves detection accuracy of a system, but also saves time and greatly improves user experience.

Referring to fig. 9, an embodiment of the present application further provides an intelligent model training apparatus, including:

a first obtaining module 902 configured to obtain a set of object sample sets, where the set of object samples includes a plurality of sets of object sample sets and object sample set labels corresponding to each set of object sample sets, each set of object sample sets includes a size of an object sample, a number of object samples, and a density of the object sample, and the object sample set labels include a mesh layer establishing method;

a training module 904 configured to train an intelligent model through the set of object sample sets, resulting in the intelligent model, the intelligent model associating the object sample sets and the object sample set labels.

Optionally, the apparatus further comprises:

Optionally, the first determining module includes:

a first establishing submodule configured to establish at least two corresponding mesh layers for each set of the object sample sets according to a set including at least two mesh layer establishing methods;

Optionally, the apparatus further comprises:

Optionally, the second determining submodule is further configured to:

a sixth determining sub-module configured to determine at least two collision relationships for the object samples in each set of object sample sets from the at least two grid layer coordinates of the object samples in each set of object sample sets.

Optionally, the coordinates of the grid layer include object granularity of the object samples in each set of the object samples, the abscissa of the object samples in each set of the object samples in the grid layer and the ordinate of the object samples in each set of the object samples in the grid layer,

the fifth determination submodule further configured to:

Optionally, the sixth determining sub-module is further configured to:

Referring to fig. 10, an embodiment of the present application further provides a collision detection apparatus, including:

a second obtaining module 1002, configured to obtain an object set to be detected including at least two objects to be detected, where the object set to be detected includes a size of the objects to be detected, a number of the objects to be detected, and a density of the objects to be detected;

a second establishing module 1004 configured to determine a mesh layer establishing method for the set of objects to be detected according to a pre-generated intelligent model, and establish a corresponding mesh layer for the set of objects to be detected according to the mesh layer establishing method;

a fourth determining module 1006, configured to determine, according to the grid layer, a collision relationship of the objects to be detected in the set of objects to be detected.

Optionally, the fourth determining module 1006 includes:

the grid layer coordinate determination submodule, further configured to:

In one or more embodiments of the present specification, after the collision detection apparatus acquires the set of objects to be detected, an optimal mesh layer establishment method is determined for the set of objects to be detected through an intelligent model, then the objects to be detected in the set of objects to be detected are placed in the mesh layer established by the optimal mesh layer establishment method, and then the collision relationship between the objects to be detected in the set of objects to be detected is accurately detected, so that not only is the detection accuracy of the system improved, but also time consumption is saved, and user experience is greatly improved.

An embodiment of the present application further provides a computer readable storage medium storing computer instructions, which when executed by a processor, implement the steps of the intelligent model training method or the collision detection method as described above.

The above is an illustrative scheme of a computer-readable storage medium of the present embodiment. It should be noted that the technical solution of the storage medium belongs to the same concept as the technical solution of the intelligent model training method or the collision detection method, and details of the technical solution of the storage medium, which are not described in detail, can be referred to the description of the technical solution of the intelligent model training method or the collision detection method.

The computer instructions comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, or the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, read-Only Memory (ROM), random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer-readable medium may contain suitable additions or subtractions depending on the requirements of legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer-readable media may not include electrical carrier signals or telecommunication signals in accordance with legislation and patent practice.

It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present application is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

The preferred embodiments of the present application disclosed above are intended only to aid in the explanation of the application. Alternative embodiments are not exhaustive and do not limit the invention to the precise embodiments described. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the application and the practical application, to thereby enable others skilled in the art to best understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. An intelligent model training method, comprising:

acquiring a set of object sample sets, wherein the set of object sample sets comprises a plurality of sets of object sample sets and object sample set labels corresponding to each set of object sample sets, each set of object sample sets comprises the size of object samples, the number of object samples and the concentration of the object samples, and the object sample set labels comprise a grid layer establishing method which establishes grid layers for the object sample sets, determines grid layer coordinates of the object samples in the object sample sets on the corresponding grid layers, and determines collision relations of the object samples in the object sample sets according to the grid layer coordinates of the object samples in the object sample sets on the corresponding grid layers;

training an intelligent model through the set of the object sample set to obtain the intelligent model, wherein the intelligent model enables the object sample set and the object sample set labels to be associated, and the intelligent model obtains the corresponding object sample set labels according to the object sample set, namely a grid layer establishing method.

2. The method of claim 1, wherein prior to obtaining the set of object sample sets, further comprising:

3. The method of claim 2, wherein determining an object exemplar set label for each set of the object exemplar sets comprises:

establishing at least two corresponding grid layers for each group of the object sample set according to a set comprising at least two grid layer establishing methods; determining at least two collision relations of the object samples in each group of the object sample sets according to the at least two grid layers;

4. The method according to claim 3, wherein after selecting a mesh layer establishing method corresponding to the collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set, the method further comprises: receiving a new mesh layer establishment method and adding the new mesh layer establishment method to the set comprising at least two mesh layer establishment methods.

5. The method of claim 3, wherein determining at least two collision relationships for the object samples in each set of object samples from the at least two grid layers comprises:

determining at least two grid layer coordinates of the object samples in each set of object sample sets in the at least two grid layers corresponding thereto; determining at least two collision relationships of the object samples in each set of object sample sets according to the at least two grid layer coordinates of the object samples in each set of object sample sets.

6. The method of claim 5, wherein the grid layer coordinates include an object granularity for the object samples in each of the object sample sets, an abscissa of the object samples in each of the object sample sets at the grid layer and an ordinate of the object samples in each of the object sample sets at the grid layer,

7. The method of claim 6, wherein determining at least two collision relationships for the object samples in each set of the object sample sets from the at least two grid-layer coordinates of the object samples in each set of the object sample sets comprises:

8. A collision detection method, characterized by comprising:

determining a grid layer establishing method for the set of objects to be detected according to a pre-generated intelligent model, and establishing a corresponding grid layer for the set of objects to be detected according to the grid layer establishing method, wherein the intelligent model is an intelligent model pre-generated by the intelligent model training method according to any one of claims 1-7;

9. The method according to claim 8, wherein the grid layer coordinates include object granularity of the objects to be detected in the set of objects to be detected, the abscissa of the objects to be detected in the set of objects to be detected on the grid layer and the ordinate of the objects to be detected in the set of objects to be detected on the grid layer,

10. The method according to claim 9, wherein determining the collision relationship of the objects to be detected in the set of objects to be detected according to the respective grid layer coordinates of the objects to be detected in the set of objects to be detected comprises:

11. An intelligent model training device, comprising:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the set of the object sample set comprises a plurality of sets of object samples and object sample set labels corresponding to each set of the object sample set, each set of the object sample set comprises the sizes of the object samples, the number of the object samples and the concentration of the object samples, and the object sample set labels comprise a grid layer establishing method which establishes grid layers for the object sample set, determines grid layer coordinates of the object samples in the object sample set on the corresponding grid layers, and determines the collision relation of the object samples in the object sample set according to the grid layer coordinates of the object samples in the object sample set on the corresponding grid layers;

a training module configured to train an intelligent model through the set of the object sample set to obtain the intelligent model, wherein the intelligent model associates the object sample set with the object sample set labels, and the intelligent model obtains the corresponding object sample set labels, namely, a grid layer establishing method, according to the object sample set.

12. The apparatus of claim 11, further comprising:

13. The apparatus of claim 12, wherein the first determining module comprises:

a second determining submodule configured to determine at least two collision relationships of the object samples in each set of the object samples according to the at least two mesh layers;

a third determining sub-module, configured to determine at least two collision detection results of the object samples in the object sample sets of each group according to the at least two collision relationships, wherein the collision detection results include a collision detection time taken to detect the object samples in the object sample sets once;

and the selection submodule is configured to select a grid layer establishing method corresponding to one collision detection result from the at least two collision detection results as an object sample set label corresponding to the object sample set.

14. A collision detecting apparatus, characterized by comprising:

the second acquisition module is configured to acquire an object set to be detected, which comprises at least two objects to be detected, wherein the object set to be detected comprises the size of the objects to be detected, the number of the objects to be detected and the density of the objects to be detected; a second establishing module, configured to determine a mesh layer establishing method for the set of objects to be detected according to a pre-generated intelligent model, and establish a corresponding mesh layer for the set of objects to be detected according to the mesh layer establishing method, wherein the intelligent model is an intelligent model pre-generated by the intelligent model training method according to any one of claims 1 to 7;

the fourth determining module comprises a grid layer coordinate determining submodule and a collision relation determining submodule, wherein the grid layer coordinate determining submodule is configured to determine grid layer coordinates of the objects to be detected in the object set to be detected on the corresponding grid layer;

15. The apparatus according to claim 14, wherein the grid layer coordinates include object granularity of the objects to be detected in the object set to be detected, the abscissa of the object to be detected in the object set to be detected on the grid layer and the ordinate of the object to be detected in the object set to be detected on the grid layer,

the grid layer coordinate determination submodule, further configured to:

16. A computing device comprising a memory, a processor and computer instructions stored on the memory and executable on the processor, wherein the processor when executing the instructions implements the steps of the method of any one of claims 1-7 or 8-10 when executed by the processor.

17. A computer-readable storage medium storing computer instructions, which when executed by a processor, perform the steps of the method of any one of claims 1 to 7 or 8 to 10.