KR20220007703A - Data collection methods, devices, instruments and storage media - Google Patents

Abstract

A data collection method, apparatus, apparatus and storage medium are provided, the method comprising: acquiring image data including a preset scene on a screen; in the image data, recognizing an object in the preset scene to obtain a recognition result; and collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule.

Description

Data collection methods, devices, instruments and storage media

Cross-referencing of related applications

The present invention claims priority to the Singapore patent application filed with the Singapore Intellectual Property Office on September 16, 2021, application number 10202110226V, the entire contents of which are incorporated herein by reference.

Embodiments of the present invention relate to the field of image processing technology, and to a data collection method, apparatus, device, and storage medium, but are not limited thereto.

In the image recognition process, since the data distribution in the production environment is different from the test environment, the image recognition accuracy is reduced; In the related art, supplemental information is collected through a manual way to improve the image recognition effect, which consumes a lot of time and energy.

An embodiment of the present invention provides a data collection technical solution.

The technical solution of the embodiment of the present invention is implemented as follows.

An embodiment of the present invention provides a data collection method, the method comprising:

acquiring image data including a preset scene on a screen;

in the image data, recognizing an object in the preset scene to obtain a recognition result; and

and collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule.

In some embodiments, after the step of recognizing an object in the preset scene in the image data to obtain a recognition result, the method may further include: determining a task demand corresponding to the image data; and determining the preset collection rule based on the task demand. In this way, by analyzing the task demand to determine a collection rule for image data, it is possible to better satisfy the user demand.

In some embodiments, when the image data is a single-frame image, the determining of the preset collection rule based on the task demand may include: determining parameter information related to the recognition result; and determining a first collection rule based on the task demand and the parameter information, wherein the preset collection rule includes the first collection rule. As such, in the case of a single-frame image, by combining the parameter information of the image recognition result and the application demand of the image to determine a collection rule, image data that meets the user demand can be selectively and automatically collected.

In some embodiments, the parameter information includes at least one of reliability, object type, and data state, and based on the task demand and the parameter information, the step of determining a first collection rule is based on the task demand to determine a target parameter from at least one of the reliability, the object type, and the data state; and determining the first collection rule based on the target parameter. In this way, by determining the first collection rule according to the target parameter matching the task demand, the collected image data can satisfy the task demand.

In some embodiments, in response to at least one of the recognition result and the image data satisfying a preset collection rule, the collecting of the image data and the recognition result comprises: a value of a target parameter of the recognition result. determining; and in response to the value of the target parameter of the recognition result satisfying the first collection rule, storing the image data and the recognition result. In this way, by determining whether the data of the target parameter of the recognition result satisfies the first collection rule, it is possible to selectively and automatically collect images satisfying the task demand.

In some embodiments, based on the task demand, the determining of the preset collection rule may include: determining business information related to the object in the preset scene; and determining a second collection rule based on the business information and the task demand, wherein the preset collection rule includes the second collection rule. In this way, by determining target business information from business information according to the task demand and determining the second collection rule related to the business information of the object, video data satisfying the task demand may be collected.

In some embodiments, when the image data is video data, the acquiring image data including a preset scene on the screen may include: determining a business step included in an execution process of an object in the preset scene; and determining the video data generated by the object in the preset scene from the business initial stage to the business end stage. In this way, by dividing the execution process of the object into a plurality of stages, and determining the entire video data from the start stage to the end stage, it is easier to logically select the video data that needs to be collected from the subsequent collected video data. do.

In some embodiments, in response to at least one of the recognition result and the image data satisfying a preset collection rule, collecting the image data and the recognition result comprises: determining business information of the video data; step; and in response to the business information of the video data satisfying the second collection rule, storing the video data and the recognition result. In this way, by determining whether the business information in the video data satisfies the second collection rule, it is possible to selectively and automatically collect video stream data satisfying the task demand.

In some embodiments, when the preset scene is a game scene, the object in the preset scene is a game object, and the image data is video data of the game object during any one round game, and business information of the video data is The determining includes determining that at least one of a video duration of the video data, a type of a game object included in the video data, and warning information in the video data is the business information. In this way, in the game scene, by using the warning information or the game duration appearing in the game as business information, the determined collection rule can automatically select video data that satisfies the business information.

In some embodiments, after the step of collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule, the method comprises: determining a network to train for recognizing an object; updating a training data set of the network to be trained based on the collected image data and the recognition result to obtain production environment data; and performing training on the network to be trained using the production environment data to obtain a trained network capable of recognizing an object in the image data. Therefore, it is possible to solve the problem that the production environment and test environment data do not match, and on the basis of this collected image data, the network model is retrained to achieve a better recognition effect.

An embodiment of the present invention provides a data collection device, the device comprising:

a first acquisition module for acquiring image data including a preset scene on the screen;

a first recognition module for recognizing an object in the preset scene from the image data to obtain a recognition result; and

and a first collection module configured to collect the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule.

Correspondingly, an embodiment of the present invention provides a computer storage medium, in which computer-executable instructions are stored, and after the computer-executable instructions are executed, the steps of the method can be implemented.

An embodiment of the present invention provides a computer device, wherein the computer device includes a memory and a processor, wherein the memory stores computer-executable instructions, and the processor executes the computer-executable instructions in the memory. steps can be implemented.

An embodiment of the present invention provides a data collection method, apparatus, apparatus and storage medium, for recognizing an object in image data including a preset scene; determine whether the recognition result and image data satisfy a preset collection rule; When the recognition result and image data satisfy the preset collection rule, the image data and the recognition result are automatically saved. In this way, image data is selectively collected in a storage device through an automatic collection method for subsequent processing after collection.

1 is a schematic diagram of an implementation process of a data collection method provided in an embodiment of the present invention.
2 is a schematic diagram of another implementation process of the data collection method provided in the embodiment of the present invention.
3 is a schematic diagram of a structure of a data collection device according to an embodiment of the present invention.
4 is a schematic diagram of a composition structure of a computer device according to an embodiment of the present invention.

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the specific technical solutions of the present invention will be described in more detail in conjunction with the drawings in the embodiments of the present invention below. The following examples are intended to illustrate the present invention and not to limit the scope of the present invention.

In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but “some embodiments” may be the same subset or different subsets of all possible embodiments, and, unless inconsistent, each other can be combined.

In the following description, the term “first/second/third” only distinguishes similar objects, does not indicate a specific order for objects, and it can be understood that “first/second/third” is acceptable Wherever applicable, the specified order or precedence order is interchanged so that the embodiments of the invention described herein may be implemented in an order other than that shown or described herein.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art of this invention. The terminology used in the text is merely for describing the purpose of the embodiments of the present invention and is not intended to limit the present invention.

Before describing the embodiment of the present invention in detail, terminology and terminology related to the embodiment of the present invention will be described, and the terminology and terminology related to the embodiment of the present invention may be applied to the following description.

1)Computer Vision is the science that studies how machines can “see”, and refers to the additional processing of images by recognizing, tracking, and measuring a target with a camera and computer instead of the human eye.

2) Image recognition refers to a technology that uses a computer to process, analyze, and understand images to recognize various and different modes of targets and objects. It is divided into four stages: → feature extraction → image recognition.

An exemplary application of the data collection device provided in the embodiment of the present invention will be described below, and the device provided in the embodiment of the present invention includes a notebook computer, tablet computer, desktop computer, camera, and mobile device ( For example, it may be implemented as various types of user terminals such as personal information terminals, dedicated messaging devices, and portable game devices, or may be implemented as servers. An exemplary application in the case where the device is implemented as a terminal or a server will be described below.

The method may be applied to a computer device, and the functions implemented by the method may be implemented by calling a program code by a processor in the computer device, of course, the program code may be stored in a computer storage medium, as can be seen, The computer device includes at least a processor and a storage medium.

An embodiment of the present invention provides a data collection method, and as shown in FIG. 1 , the steps shown in FIG. 1 are combined and described.

In step S101, image data including a preset scene on the screen is acquired.

In some embodiments, the preset scene is a user-specified collection scene, such as a scene to which the game location belongs, an outdoor scene (eg, a road with pedestrians), or an indoor scene (eg, an indoor public place such as a shopping mall, hospital, etc.) can be Image data is single frame image or video data collected in the scene. For example, if the preset scene is a game place, the image data may be an image collected by a corresponding process in the game place, or may be video data collected during a game process.

In step S102, in the image data, an object in the preset scene is recognized, and a recognition result is obtained.

In some embodiments, a convolutional neural network is used to recognize an object in image data to obtain a recognition result. For example, if the preset scene is a game place, the objects are game-related objects such as game coins, players, game managers, and game tables in the game. A game object is recognized from the image data to obtain the recognition result.

In step S103, the image data and the recognition result are collected in response to at least one of the recognition result and the image data satisfying a preset collection rule.

In some embodiments, the preset collection rule may be a rule related to a recognition result, a rule related to image data, or a rule related to both the recognition result and image data. if the collection rule relates only to the recognition result, the recognition result satisfies the collection rule, that is, it is determined that at least one of the recognition result and the image data satisfies a preset collection rule; Similarly, if the collection rule relates to both the recognition result and the image data, the recognition result and the image data are both related, and when the collection rule is satisfied at the same time, the recognition result and the image data are collected. If the preset scene is the game location, the image data is a single frame image, and the preset collection rule is that the image includes a preset object (eg, a preset type of game coin), the recognition result of the image data to determine whether a game coin of a preset type is included in the recognition result; If the recognition result includes a game coin of a preset type, the image data and the recognition result are stored.

In an embodiment of the present invention, an object in image data including a preset scene is recognized; determine whether the recognition result and image data satisfy a preset collection rule; When the recognition result and image data satisfy the preset collection rule, the image data and the recognition result are automatically saved. In this way, image data is selectively collected in a storage device through an automatic collection method for subsequent processing after collection.

In some embodiments, the application demand of image data is analyzed to determine a preset collection rule, that is, further comprising the following steps before step S102 , as shown in FIG. 2 , the following description is shown in FIG. 2 . The steps are described by combining them.

In step S201, a task demand corresponding to the image data is determined.

In some embodiments, the task demand is a demand to apply the image data. For example, task demands include using the image data to train a model for recognizing images, recognizing the image data to obtain a specific type of object, finding abnormal images, and the like.

In step S202, the preset collection rule is determined based on the task demand.

In some embodiments, rules for collecting image data are determined to satisfy a task demand. For example, if the task demand is to train a model for recognizing an image using the image data, the determined preset collection rule may collect the same image data as the test environment; If the task demand is to recognize the image data to acquire a specific type of object, the determined preset collection rule may be to collect image data of the type of object included in the recognition result; If the task demand is to find an abnormal image, the determined preset collection rule may be to collect abnormal image data in the recognition process. In this way, by analyzing the task demand to determine a collection rule for image data, it is possible to better satisfy the user demand.

In an embodiment of the present invention, each single-frame image and video data are collected by using different collection rules, wherein the collecting process for the single-frame image is the same as the method 1, and the collecting process for the video data is the method Same as 2.

In method 1, in some embodiments, when the image data is a single-frame image, it is determined that the collection rule is a rule related to the recognition result, that is, the step S202 is performed by the following steps S221 and S222 (not shown). can be implemented.

In step S221, parameter information related to the recognition result is determined.

In one possible implementation manner, when the image data is a single-frame image, an object in the single-frame image is recognized to obtain a recognition result. the parameter information related to the recognition result is a parameter mentioned in the recognition result, and includes at least one of reliability, number of objects, object type, and data state; Here, the reliability is the reliability of the recognition result, the object type is the type of the object included in the recognition result, and the data state is the recognition result, such as whether an abnormality appears in the process of recognizing an object of a single frame image, is a normal recognition result whether it is aware

In step S222, a first collection rule is determined based on the task demand and the parameter information.

In one possible implementation manner, the preset collection rule includes the first collection rule. A first collection rule related to a recognition result is determined by combining the parameter information and the task demand corresponding to the single frame image. As such, in the case of a single-frame image, by combining the parameter information of the image recognition result and the application demand of the image to determine a collection rule, image data that meets the user demand can be selectively and automatically collected.

In some possible implementation manners, the first collection rule is determined based on at least one of task demand and reliability, the object type, and the data state, that is, the step S222 may be implemented by the following steps.

In a first step, a target parameter is determined from at least one of the reliability, the object type, and the data state, based on the task demand.

In one possible implementation manner, if the task demand is to recognize the image data to obtain a specific type of object, determine, in reliability, the object type, the data state, that the target parameter is the object type in the parameter information. Or, if the task demand is to find an abnormal image, the target parameter is the data state.

In a second step, the first collection rule is determined according to the target parameter.

In one possible implementation manner, according to a target parameter, a first collection rule is determined, for example, if the target parameter is an object type, according to a specific object type required for a task demand, the first collection rule is determined in the recognition result according to the specific object type. It is decided to collect a single frame image containing the object type. If the target parameter is the reliability, according to the demand for the reliability threshold among the task demands, the first collection rule is determined to collect the single-frame image in which the reliability of the recognition result satisfies the reliability threshold. In this way, by determining the first collection rule according to the target parameter matching the task demand, the collected image data can satisfy the task demand.

In some embodiments, when the image data is a single-frame image, after determining the first collection rule by combining task demands according to the parameter information of the recognition result, whether the recognition result of the single-frame image satisfies the first collection rule to determine whether to collect the single-frame image; That is, the step S103 may be implemented by the following steps S131 and S132 (not shown).

In step S131, a value of a target parameter of the recognition result is determined.

In one possible implementation manner, data of each parameter in the recognition result is first determined, for example, the reliability level of the recognition result, whether the data state is abnormal, and the specific type and number of objects included in the recognition result, etc. to decide

In step S132, in response to the value of the target parameter of the recognition result satisfying the first collection rule, the image data and the recognition result are stored.

In one possible implementation manner, if the target parameter in the recognition result satisfies the first collection rule, it describes that the frame image is an image that needs to be collected during task demand, so that the image data and the recognition result are stored in a storage system for subsequent to facilitate the implementation of the above task. In this way, it is determined whether the target parameter data of the recognition result satisfies the first collection rule, and images satisfying the task demand can be selectively and automatically collected.

In method 2, when the image data is video data, the business step for the implementation process of the object is divided, and the video in the preset task step is collected, that is, the step S101 is the following steps S111 and S112 (not shown). ) can be implemented by

In step S111, a business step included in the execution process of the object in the preset scene is determined.

In some embodiments, the business step included in the object in the preset scene is a business step included in the entire implementation process among the preset scenes of the object. If the preset scene is a game scene and the object is a game, for example, the business phase is the round game, such as the game start preparation phase, the manager's game participation phase, the player's game participation phase, the game phase, the game end and the result output phase. It includes each game phase from the beginning to the end.

In step S112, the video data generated by the object in the preset scene is determined from the business initial stage to the business end stage.

In some embodiments, the video data includes other business information generated by the object in the process, such as a video stream of the object in the process from the business initiation stage to the business end stage, and alert information in the whole process. Taking as an example that the preset scene is a game scene and the object is a game, the video data includes a video stream from the game start preparation stage to the game end and result output stage of a game of one round, and a warning in one round of the game including the number of times and types of warnings. In this way, by dividing the execution process of the object into a plurality of stages, and determining the entire video data from the start stage to the end stage, it is easier to logically select the video data that needs to be collected from the subsequent collected video data. .

In some embodiments, for the case where the image data is video, the second collection rule is determined by analyzing the object business information in the preset scene and combining the task demand, that is, the step S202 is performed by the following steps S21 and S22 may be implemented.

In step S21, business information related to the object in the preset scene is determined.

In one possible implementation manner, the business information related to the object is business information of the object that may exist in the preset scene, and the business information is information describing the business to which the object belongs. For example, the business information includes whether warning information exists in the object. If the object is a game as an example, the business information includes the number of warnings or warning types in a game in which the game is one round, and a duration of a game in one round, for example. That is, it is determined that at least one of a video duration of the video data, a type of a game object included in the video data, and warning information in the video data is the business information. In this way, in the game scene, by using the warning information or the game duration appearing in the game as business information, the determined collection rule can automatically select video data that satisfies the business information.

In step S22, a second collection rule is determined based on the business information and the task demand.

In one possible implementation manner, the preset collection rule includes the second collection rule. By combining the business information and the task demand, a collection rule that satisfies the task demand is determined. When the image data is video data, according to the task demand, target business information matching the task demand in the business information is determined, and then a second collection rule is determined based on the target business information. For example, if the task demand is to detect warning information of a preset type appearing in a video, the target business information is warning information of the preset type, and the second collection rule is video data including warning information of a preset type is to collect In this way, by determining target business information from business information according to the task demand and determining the second collection rule related to the business information of the object, video data satisfying the task demand may be collected.

In some embodiments, when the image data is video data, after determining the second collection rule by combining the task demand according to the business information of the object, it is determined whether the video data satisfies the second collection rule, It is determined whether or not to collect data, that is, the step S103 may be implemented by the following steps S141 and S142 (not shown).

In step S141, business information of the video data is determined.

In one possible implementation manner, the target business information of the video data is determined according to the target business information targeted when determining the second collection rule. For example, when the target business information for determining the second collection rule is warning information of a preset type, the target business information of the video data is a warning information type.

In step S142, in response to the business information of the video data satisfying the second collection rule, the video data and the recognition result are stored.

In one possible implementation manner, if the target business information in the business information of the video data satisfies the second collection rule, it describes that the video data is video data that needs to be collected during the task demand, so that the video data and the recognition result are stored stored in the system to facilitate subsequent implementation of the above task. For example, if the preset scene is a game scene, the video data includes a video stream collected for one round game (eg, a video stream in the entire process from the start of the game to the end) and the warning information generated in the course of a game; If the second collection rule is to collect video data including warning information of a preset type, it is determined whether the type of warning information in the video data is a preset type to determine whether to collect the video data.

In an embodiment of the present invention, by judging whether the business information in the video data satisfies the second collection rule, the video stream data satisfying the task demand may be selectively and automatically collected.

In some embodiments, after the image data is collected, in order to improve the recognition effect of the image recognition network, the image data is used as training data of a network model to be trained, and may be implemented through the following process.

First, a network to be trained for recognizing an object in the image data is determined.

In one possible implementation manner, the network to train may be any network to train for performing image recognition, such as a convolutional neural network to train.

Next, based on the collected image data and the recognition result, the training data set of the network to be trained is updated to obtain production environment data.

In one possible implementation manner, the collected image data and recognition results are used as supplemented training data of the network to be trained, and the training data set is added to obtain production environment data.

Finally, by using the production environment data, training is performed on the network to be trained to obtain a trained network capable of recognizing an object in the image data.

In one possible implementation manner, the production environment data is used as an updated training data set, and the network to be trained is trained to obtain a trained network having a relatively good image recognition effect. Therefore, it is possible to solve the problem that the production environment and test environment data do not match, and on the basis of this collected image data, the network model is retrained to achieve a better recognition effect.

Below, an exemplary application of the embodiment of the present invention will be described in a practical application scene, taking a game place as an example, and automatic retrieval of artificial intelligence (AI) system data of the game place will be described.

In general, image recognition has a problem of poor accuracy in a real production environment. The reason for this is as follows. First, the data distribution in the production environment is different from the test environment (eg, many faces wearing masks appeared after the epidemic, and such face information occupies a small part in the test data); Second, abnormal data appears in the production environment and causes system errors.

In order to solve the problem of inconsistency between production environment and test environment data, it is necessary to collect sufficient supplementary information (eg images) from the production environment by personnel in the relevant technology, and retrain on the basis of these supplementary information to further Achieving a good perception effect. As such, it takes a lot of time and effort to re-collect data after the production environment is changed.

An embodiment of the present invention provides a data collection method for selectively collecting production environment data in a specific storage device in an automatic collection method to facilitate subsequent processing after collection. In some embodiments, a data collection module is added to the edge AI device (the module incorporates relevant information such as processing results of the AI system, time, device configuration parameters, service alerts, etc.), and retrieves the data according to the determined collection rules It is determined whether or not to do so, and data that needs to be retrieved is automatically written to the designated storage device. In an embodiment of the present invention, data may be collected based on an image or video, wherein:

The image-based data collection process steps are as follows.

In the first step, the recognition result of each image is collected.

In a second step, a data collection rule determined during configuration is obtained.

In the third step, according to the data collection rule, it is determined whether to collect the image and the processing result in the storage system.

In one possible implementation manner, the data collection rules include the reduced number of game coins among the processing results, the reliability of the processing results, or whether there is an error in the results, and the like.

The video-based data collection process steps are as follows.

In the first stage, the relevant data in the game is divided into rounds according to the game stage fed back by the business layer, and the game data is retrieved for each game division round. Save different rounds of games in their respective folders.

In the second step, the redemption data includes a video stream, a recognition result, and warning information generated in a program execution process.

In the third step, the collection rules for each type of data are determined through the configuration file distributed at the cloud end. For example, collect all game videos that have a game duration of less than 30 seconds, or collect game videos in which a warning appears 3 or more times in a game.

In the fourth step, according to the collection rule, it is determined whether to automatically collect the game video in the storage system.

After collecting data through the above two methods, the collected data can be applied to the following business service directions.

First, the stored data is used by the test team to reproduce the problems presented in the production environment.

Second, the stored data is used for model training.

Third, the stored data is used as a test data set for other new services.

In an embodiment of the present invention, the side-end device obtains data collection rules, selectively automatically collects data according to the rules, groups the video for each round according to the information fed back by the service layer, and stores the game round information and; In this way, it is possible to more easily manage data by automatically collecting production environment data, selectively collecting valid information according to data collection rules, and storing game videos by group according to business layer feedback information.

An embodiment of the present invention provides a data collection device, and FIG. 3 is a structural composition schematic diagram of the embodiment data collection device of the present invention, and as shown in FIG. 3 , the data collection device 300 includes:

a first acquisition module 301 for acquiring image data including a preset scene on the screen;

a first recognition module (302) for recognizing an object in the preset scene in the image data to obtain a recognition result; and

and a first collection module (303) configured to collect the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule.

In some embodiments, the device comprises:

a first determining module for determining a task demand corresponding to the image data; and

The method further includes a second determining module configured to determine the preset collection rule based on the task demand.

In some embodiments, when the image data is a single frame image, the second determining module is

a first determining submodule for determining parameter information related to the recognition result; and

a second determining submodule to determine a first collection rule based on the task demand and the parameter information, wherein the preset collection rule includes the first collection rule.

In some embodiments, the parameter information includes at least one of reliability, object type, and data state, and the second determining submodule comprises:

a first determining unit configured to determine, in at least one of the reliability, the object type, and the data state, a target parameter based on the task demand; and

and a second determining unit to determine the first collection rule based on the target parameter.

In some embodiments, the first collection module 303 includes:

a third determining submodule for determining a value of a target parameter of the recognition result; and

and a first storage submodule configured to store the image data and the recognition result in response to the value of the target parameter of the recognition result satisfying the first collection rule.

In some embodiments, the second determining module comprises:

a fourth determining sub-module for determining business information related to the object in the preset scene; and

a fifth determining submodule to determine a second collection rule based on the business information and the task demand, wherein the preset collection rule includes the second collection rule.

In some embodiments, when the image data is video data, the first acquiring module 301 is

a sixth determining sub-module for determining a business step included in the execution process of an object in the preset scene; and

and a seventh determining sub-module for determining the video data generated by the object in the preset scene from the business initial stage to the business end stage.

In some embodiments, the first collection module 303 includes:

an eighth determining submodule for determining business information of the video data; and

and a second storage submodule configured to store the video data and the recognition result in response to the business information of the video data satisfying the second collection rule.

In some embodiments, when the preset scene is a game scene, the object in the preset scene is a game object, the image data is video data of the game object in any one round game, and the eighth determining submodule includes: ,

and a third determining unit, configured to determine that at least one of a video duration of the video data, a type of a game object included in the video data, and warning information in the video data is the business information.

In some embodiments, the device comprises:

a third determining module for determining a network to be trained for recognizing an object in the image data;

a first update module configured to obtain production environment data by updating a training data set of the network to be trained based on the collected image data and the recognition result; and

The method further includes a first training module configured to perform training on the network to be trained by using the production environment data to obtain a trained network capable of recognizing an object in the image data.

It should be noted that the above description of the device embodiment is similar to the description of the method embodiment, and has similar advantageous effects as the method embodiment. Technical details not mentioned in the device embodiment of the present invention should be understood with reference to the description of the method embodiment of the present invention.

It should be described that, in an embodiment of the present invention, the data collection method is implemented in the form of a software function module, and even when sold or used as a separate product, it may be stored in one computer-readable storage medium. Based on this understanding, a part of the technical solutions of the embodiments of the present invention essentially or contributing to the prior art may be reflected in the form of a software product, the computer software product being stored in one storage medium, and one It includes a plurality of instructions for causing a computer device (which may be a terminal, a server, etc.) or a processor to perform all or some steps of the method according to each embodiment of the present invention. The above-described storage medium includes various media capable of storing a program code, such as a U disk, a removable hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk. As such, embodiments of the present invention are not limited to any particular combination of hardware and software.

Correspondingly, an embodiment of the present invention further provides a computer program product that includes computer-executable instructions, and after the computer-executable instructions are executed, can implement the steps in the data collection method provided in the embodiment of the present invention do.

Correspondingly, an embodiment of the present invention further provides a computer storage medium storing computer-executable instructions, and implementing the steps of the data collection method provided in the embodiment when the computer-executable instructions are executed by a processor.

Correspondingly, an embodiment of the present invention provides a computer device, FIG. 4 is a schematic diagram of the composition structure of a computer device according to an embodiment of the present invention, and as shown in FIG. 4 , the electronic device 400 includes a processor ( 401 ), at least one communication bus, a communication interface 402 , at least one external communication interface and a memory 403 . Here, the communication interface 402 is configured to implement connection communication between these components. Here, the communication interface 402 may include a display screen, and the external communication interface may include a standard wired interface and a wireless interface. Here, the processor 401 is configured to execute the image processing program in the memory to implement the steps of the data collection method provided in the above embodiment.

The above is a description of the data collection device, computer device and storage medium embodiment, is similar to the description of the method embodiment, has a technical description and beneficial effect similar to the corresponding method embodiment, and with the limitation of the width, the method embodiment is Reference may be made to the described content, which will not be repeated here any longer. Technical details not mentioned in the data collection device, computer device and storage medium embodiment of the present invention should be understood with reference to the description of the method embodiment of the present invention.

It should be understood that reference to “one embodiment” or “an embodiment” in the preceding specification means that a particular feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the present invention. Accordingly, “in one embodiment” or “in an embodiment” described in each part of the first half of the specification may not refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In various embodiments of the present invention, the size of the sequence number of each process does not mean the precedence and afterward of the execution order, and the execution order of each process should be determined according to its function and inherent logic, and according to the embodiment of the present invention, It can be understood that there is no limitation on the implementation process. The order numbers of the embodiments of the present invention are for illustrative purposes only, and do not represent advantages and disadvantages of the embodiments. It should be noted that, as used herein, the terms “inclusive”, “comprising” or any other variation are intended to include non-exclusive inclusions, such that a process, method, article or apparatus comprising a set of elements is intended to include those elements as well as the explicit inclusion of those elements. It is intended to further include other elements not explicitly listed, or elements unique to such processes, methods, articles, or devices. In the absence of further restrictions, “one … An element defined by the sentence "comprises" does not exclude the presence of other identical elements in a process, method, article or apparatus comprising the element.

In the plurality of embodiments provided by the present invention, it will be understood that the disclosed devices and methods may be implemented in other ways. The device embodiment described above is merely exemplary, for example, the partition of the unit is only a logical function partition, and there may be other partitioning methods in actual implementation, for example, a plurality of units or components are different from each other. may be combined or integrated into the system of , some features may be omitted or not implemented. Further, the coupling or direct coupling or communication connection between the respective component parts shown or discussed may be through some interface, and the indirect coupling or communication connection of the device or unit may be in an electrical, mechanical or other form. have.

A unit described as a separating member may or may not be physically separated, and a member denoted as a unit may or may not be a physical unit; may be located in one place or distributed in a plurality of network units; According to the actual demand, some or all of the units may be selected to implement the purpose of the solution of the present embodiment.

In addition, in each embodiment of the present invention, each functional unit may be integrated into one processing unit, each unit may exist independently as one unit, or two or two or more units may be integrated into one unit may be; The integrated unit may be implemented in the form of hardware, or may be implemented in the form of a combination of hardware and software functional units. As those skilled in the art can understand, the implementation of all or some steps of the embodiment of the method may be completed through hardware related to program instructions, and the above-described program may be stored in a computer-readable storage medium, and , when the program is executed, performing the steps of the method embodiment; The aforementioned storage medium includes various media capable of storing a program code, such as a mobile storage device, a read only memory (ROM), a magnetic disk or an optical disk.

Alternatively, even when the integrated unit of the present invention is implemented in the form of a software function module and sold or used as a separate product, it may be stored in a computer-readable storage medium. Based on this understanding, a part of the technical solutions of the present invention essentially or contributing to the prior art may be reflected in the form of a software product, the computer software product being stored in one storage medium, and one computer device It includes a plurality of instructions for causing (which may be a personal computer, a server, a network device, etc.) to perform all or a part of the method according to each embodiment of the present invention. The above-described storage medium includes various media capable of storing a program code such as a mobile storage device, a ROM, a magnetic disk, or an optical disk. The above description is only specific embodiments of the present invention, the protection scope of the present invention is not limited thereto, and any person skilled in the art can easily come up with changes or replacements within the technical scope disclosed in the present invention, All of these should be included within the protection scope of the present invention. Accordingly, the protection scope of the present invention should be based on the protection scope of the above claims.

Claims (20)

A data collection method comprising:
acquiring image data including a preset scene on a screen;
in the image data, recognizing an object in the preset scene to obtain a recognition result; and
and collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule. According to claim 1,
In the image data, after the step of recognizing an object in the preset scene to obtain a recognition result, the data collection method includes:
determining a task demand corresponding to the image data; and
The method of claim 1, further comprising the step of determining the preset collection rule based on the task demand. 3. The method of claim 2,
When the image data is a single frame image, the step of determining the preset collection rule based on the task demand includes:
determining parameter information related to the recognition result; and
and determining a first collection rule based on the task demand and the parameter information, wherein the preset collection rule includes the first collection rule. 4. The method of claim 3,
the parameter information includes at least one of reliability, object type, and data state; The step of determining a first collection rule based on the task demand and the parameter information,
determining, based on the task demand, a target parameter in at least one of the reliability, the object type, and the data state; and
and determining the first collection rule based on the target parameter. 4. The method of claim 2 or 3,
The step of collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule,
determining a value of a target parameter of the recognition result; and
and in response to the value of the target parameter of the recognition result satisfying the first collection rule, storing the image data and the recognition result. 3. The method of claim 2,
The step of determining the preset collection rule based on the task demand,
determining business information related to the object in the preset scene; and
and determining a second collection rule based on the business information and the task demand, wherein the preset collection rule includes the second collection rule. 7. The method of claim 6,
When the image data is video data, acquiring image data including a preset scene on the screen includes:
determining a business step included in the execution process of the object in the preset scene; and
and determining video data generated by the object in the preset scene from the initial business stage to the business end stage. 8. The method of claim 6 or 7,
The step of collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule,
determining business information of the video data; and
in response to the business information of the video data satisfying the second collection rule, storing the video data and the recognition result. 9. The method of claim 8,
When the preset scene is a game scene, the object in the preset scene is a game object, the image data is video data of the game object during any one round game, and the step of determining business information of the video data includes: ,
and determining that at least one of a video duration of the video data, a type of a game object included in the video data, and warning information in the video data is the business information. 10. The method according to any one of claims 1 to 9,
After the step of collecting the image data and the recognition result in response to at least one of the recognition result and the image data satisfying a preset collection rule, the data collection method comprises:
determining a network to be trained for recognizing an object in the image data;
updating a training data set of the network to be trained based on the collected image data and the recognition result to obtain production environment data; and
The method of claim 1, further comprising: using the production environment data to train the network to be trained to obtain a trained network capable of recognizing an object in the image data. A computer device comprising:
The computer device includes a memory and a processor, wherein the memory stores computer-executable instructions, and when the processor executes the computer-executable instructions in the memory,
acquiring image data including a preset scene on the screen;
in the image data, recognizing an object in the preset scene to obtain a recognition result;
and in response to at least one of the recognition result and the image data satisfying a preset collection rule, collect the image data and the recognition result. 12. The method of claim 11,
In the image data, after recognizing an object in the preset scene to obtain a recognition result, the processor further comprises:
determine a task demand corresponding to the image data;
and determine the preset collection rule according to the task demand. 13. The method of claim 12,
When the image data is a single-frame image, when determining the preset collection rule based on the task demand, the processor,
determine parameter information related to the recognition result;
and determine a first collection rule based on the task demand and the parameter information, wherein the preset collection rule includes the first collection rule. 14. The method of claim 13,
the parameter information includes at least one of reliability, object type, and data state; When determining a first collection rule based on the task demand and the parameter information, the processor,
determine, based on the task demand, a target parameter in at least one of the reliability, the object type, and the data state;
and determine the first collection rule based on the target parameter. 14. The method of claim 12 or 13,
When the image data and the recognition result are collected in response to at least one of the recognition result and the image data satisfying a preset collection rule, the processor,
determine a value of a target parameter of the recognition result;
and in response to the value of the target parameter of the recognition result satisfying the first collection rule, store the image data and the recognition result. 13. The method of claim 12,
When determining the preset collection rule based on the task demand, the processor,
determine business information related to the object in the preset scene;
and determine a second collection rule based on the business information and the task demand, wherein the preset collection rule includes the second collection rule. 17. The method of claim 16,
When the image data is video data, when acquiring image data including a preset scene on a screen, the processor,
determining a business step included in the execution process of the object in the preset scene;
and determine the video data generated by the object in the preset scene from the business initial stage to the business end stage. 18. The method of claim 16 or 17,
When the image data and the recognition result are collected in response to at least one of the recognition result and the image data satisfying a preset collection rule, the processor,
determine business information of the video data;
and in response to the business information of the video data satisfying the second collection rule, store the video data and the recognition result. A computer storage medium comprising:
Computer-executable instructions are stored in the computer storage medium, and after the computer-executable instructions are executed,
acquiring image data including a preset scene on the screen;
in the image data, recognizing an object in the preset scene to obtain a recognition result;
and in response to at least one of the recognition result and the image data satisfying a preset collection rule, collect the image data and the recognition result. A computer program comprising:
comprising computer instructions executable by an electronic device;
When the computer instructions are executed by a processor in the electronic device,
acquiring image data including a preset scene on the screen;
in the image data, recognizing an object in the preset scene to obtain a recognition result;
and in response to at least one of the recognition result and the image data satisfying a preset collection rule, collect the image data and the recognition result.

Images