Object management method and device, equipment and storage medium

Abstract

Embodiments of the invention disclose an object management method and apparatus, a device and a storage medium. The method comprises the steps of determining a to-be-processed object matched with an obtained object identifier; determining an enhancement processing mode of the to-be-processed object and object description information of the to-be-processed object; determining enhancement mode coding information of the enhancement processing mode according to the enhancement processing mode; according to the to-be-processed object, the object description information and the enhancement mode coding information, an object enhancement evaluation result of the to-be-processed object is predicted, and the object enhancement evaluation result is used for representing the enhancement processing effect of the to-be-processed object based on the enhancement processing mode; and if the object enhancement evaluation result meets a preset condition, performing enhancement processing on the to-be-processed object based on an enhancement processing mode. By means of the method, effective screening whether the objects are subjected to enhancement processing or not can be achieved, it is guaranteed that enhancement processing is only carried out on the objects generating the forward feedback effect, and therefore computing power resources are saved, the enhancement processing cost is reduced, and the forward feedback effect is improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to an object management method, apparatus, device, and storage medium. Background Technology

[0002] Currently, the quality of entertainment content such as videos, images, and live streams is affected to varying degrees and types during production and transmission due to factors such as equipment, lighting, and compression. Therefore, servers distributing entertainment content need to perform repair or enhancement processing before sending it to clients to improve the browsing experience on the client side.

[0003] However, this does not mean that all entertainment-related objects are suitable for restoration or enhancement. Some processed entertainment-related objects may not provide a better visual experience for viewers. For example, enhancing images and text may actually cause them to lose their original authenticity, thus affecting the number of views for that type of entertainment-related object.

[0004] Furthermore, processing entertainment-related objects requires high computing power, and the number of entertainment-related objects that can be processed within a single processing cycle is limited. If the processed entertainment-related objects do not provide a good experience for the viewer, it not only consumes computing resources but also misses the opportunity to process other entertainment-related objects that require processing, thus diminishing the positive feedback effect that entertainment-related objects can bring. Summary of the Invention

[0005] This disclosure provides an object management method, apparatus, device, and storage medium that can effectively filter whether or not object enhancement processing is performed, thus greatly saving computing resources for enhancement processing.

[0006] In a first aspect, embodiments of this disclosure provide an object management method, the method comprising:

[0007] Identify the objects to be processed that match the acquired object identifiers;

[0008] Determine the enhancement processing method for the object to be processed and the object description information of the object to be processed;

[0009] Based on the enhancement processing method, determine the enhancement method encoding information of the enhancement processing method;

[0010] Based on the object to be processed, the object description information, and the enhancement method encoding information, the object enhancement evaluation result of the object to be processed is predicted. The object enhancement evaluation result is used to characterize the effect of enhancing the object to be processed based on the enhancement processing method.

[0011] If the object enhancement evaluation result meets the preset conditions, then the object to be processed is enhanced based on the enhancement processing method.

[0012] Secondly, embodiments of this disclosure also provide an object management device, the device comprising:

[0013] The first determination module is used to determine the object to be processed that matches the acquired object identifier;

[0014] The second determining module is used to determine the enhanced processing method of the object to be processed and the object description information of the object to be processed.

[0015] The third determining module is used to determine the enhancement method encoding information of the enhancement processing method according to the enhancement processing method;

[0016] The result evaluation module is used to predict the object enhancement evaluation result of the object to be processed based on the object to be processed, the object description information, and the enhancement method encoding information. The object enhancement evaluation result is used to characterize the effect of enhancing the object to be processed based on the enhancement processing method.

[0017] The object management module is used to perform enhancement processing on the object to be processed based on the enhancement processing method if the object enhancement evaluation result meets the preset conditions.

[0018] Thirdly, embodiments of this disclosure also provide a computer device, the computer device comprising:

[0019] One or more processors;

[0020] Storage device for storing one or more programs.

[0021] When the one or more programs are executed by the one or more processors, the one or more processors implement the object management method provided in any embodiment of this disclosure.

[0022] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the object management method provided in any embodiment of this disclosure.

[0023] Fifthly, embodiments of this disclosure also provide a computer program product, including a computer program that, when executed by a processor, implements the object management method provided in any embodiment of this disclosure.

[0024] The technical solution of this disclosure embodiment specifically discloses an object management method, apparatus, device, and storage medium. The method first determines a pending object that matches an acquired object identifier; determines an enhancement processing method for the pending object and object description information of the pending object; determines enhancement method encoding information of the enhancement processing method based on the enhancement processing method; predicts an object enhancement evaluation result for the pending object based on the pending object, the object description information, and the enhancement method encoding information, the object enhancement evaluation result being used to characterize the effect of enhancing the pending object based on the enhancement processing method; if the object enhancement evaluation result meets preset conditions, then the pending object is enhanced based on the enhancement processing method. This embodiment's technical solution is equivalent to pre-determining whether the pending object is suitable for enhancement processing before performing enhancement processing. The logic of this decision management mainly involves predicting whether the object processed using the enhancement processing method will have a positive feedback effect on the pending object, and whether positive feedback can be generated, which can be determined by comparing the determined enhancement evaluation result with preset conditions. This technical solution assumes that when the enhancement evaluation result meets preset conditions, the enhanced object is superior to the original object. Therefore, in this case, enhancement processing of the object is permitted. This decision management method effectively filters whether to enhance an object, ensuring that only objects generating positive feedback are enhanced, thereby saving computing resources and reducing enhancement processing costs. Simultaneously, it improves enhancement processing efficiency and enhances the positive feedback effect while maintaining the same number of enhanced objects. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the exemplary embodiments of this disclosure, the accompanying drawings used in describing the embodiments are briefly introduced below. Obviously, the accompanying drawings described are only a portion of the embodiments to be described in this disclosure, and not all of them. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort.

[0026] Figure 1a A flowchart illustrating an object management method provided in an embodiment of this disclosure;

[0027] Figure 1b Schematic diagrams of the first and second sub-analysis models of the object enhancement evaluation model involved in the object management method provided in this embodiment are given;

[0028] Figure 1cA schematic diagram of the specific structure of a feature processing module in the first sub-analysis model involved in the object association method provided in this embodiment is given;

[0029] Figure 1d A flowchart illustrating an implementation of the object management method provided in this embodiment in a specific scenario is given.

[0030] Figure 2 This is a schematic diagram of the structure of an object management device provided in an embodiment of the present disclosure;

[0031] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Detailed Implementation

[0032] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0033] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0034] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0035] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules, or units, and are not used to limit the order of functions performed by these devices, modules, or units or their interdependencies. It should also be noted that the modifications of "a" and "a plurality of" mentioned in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0036] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0037] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0038] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0039] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0040] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0041] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0042] Figure 1a This is a flowchart illustrating an object management method provided in an embodiment of the present disclosure. This embodiment is applicable to object management situations. The method can be executed by an object management device, which can be implemented by software and / or hardware and can be configured in a terminal and / or server to implement the object management method in this embodiment of the present disclosure.

[0043] It should be noted that one application scenario of this embodiment can be described as follows: When an entertainment application client requests the content of a certain entertainment object from an entertainment application server, according to the existing object resource distribution logic, the entertainment application server needs to first determine the enhancement processing method to be applied to the entertainment object, perform enhancement processing on the entertainment object based on the enhancement processing method, and then distribute the enhanced content of the entertainment object to the entertainment application client. In the existing method, if the distributed enhanced content does not bring positive feedback, it will instead waste the enhancement processing computing resources, and the consumption of enhancement processing computing resources will also reduce the amount of other objects with enhancement processing needs that can be processed.

[0044] Based on this, this embodiment provides an object management method that can effectively filter whether or not object enhancement processing is needed, greatly saving computational resources for enhancement processing. Specifically, such as... Figure 1a As shown, the object management method provided in this embodiment may include:

[0045] S101. Determine the object to be processed that matches the obtained object identifier.

[0046] In this embodiment, the executing entity can be viewed as a server for a certain type of application, or a terminal responding to business requests related to a certain type of application. The object to be processed can be considered as the request object involved in the business request generated by the client or user of a certain type of application. Taking an entertainment application as an example, the object to be processed can be an entertainment object requested in the business request initiated by the entertainment application client, such as a requested image or a requested video. In this embodiment, the requested object can be determined by analyzing the identifier of the received business request object and the requested object can be regarded as the object to be processed; alternatively, an object with the identifier of the object to be sent to the client or user can be regarded as the object to be processed.

[0047] It is understandable that the object to be processed can be obtained directly from the object authoring end, such as when the executing entity forwards the received business request to the object authoring end, which then provides it; or it can be an object that is pre-cached in the original object database, where the original object database can be used to cache the original objects, which can be pre-imported from external sources or uploaded by the object authoring end.

[0048] S102. Determine the enhanced processing method for the object to be processed and the object description information of the object to be processed.

[0049] In this embodiment, the enhancement processing method can be considered as an enhancement algorithm used to enhance one or more objects according to certain enhancement requirements. Taking an image or video as an example, the enhancement processing method can be an image quality enhancement algorithm, a noise removal algorithm, and a deblurring algorithm, etc.

[0050] In this embodiment, the required enhancement processing can be determined by analyzing the attribute information carried by the object to be processed. For example, taking an image as the object to be processed, the attribute information carried by the image may include image resolution, image pixel information, and the image's own content. Analyzing this information allows us to determine whether the image resolution is below a set threshold, whether the image quality composed of the image pixel information is poor, and by analyzing the image's own content, we can identify the region of interest in the image and determine whether the content of that region of interest is too blurry. Finally, based on the determination results, we decide whether image enhancement is needed and what enhancement method should be used.

[0051] It's important to understand that in the decision-making process for enhancing one or more objects, there are cases where no enhancement is needed. In such cases, the object can be directly recorded as the target object in the object database after decision filtering, or the object can be directly sent to the client. This embodiment primarily manages objects whose decision result indicates the need for enhancement and whose corresponding enhancement processing method has been determined.

[0052] In this embodiment, this step can further form object description information of the object to be processed. The object description information can be understood as information describing the object-related attribute information of the object to be processed. The object-related attribute information may include the basic attribute information of the image to be processed and related creative information.

[0053] For example, taking an image as the object to be processed, the image's basic attribute information, such as its length and width, and its image category, and related creative information, such as the capture angle of the image capture device used, and whether secondary creation was performed, such as adding special effects or filters, can be summarized into a descriptive text to constitute the object description information of the object to be processed.

[0054] S103. Determine the enhancement method encoding information of the enhancement method according to the enhancement processing method.

[0055] It should be noted that, in this embodiment, all enhancement processing methods to be applied to the object to be processed are pre-encoded, and the encoded content can be used as enhancement processing method encoding information. In this embodiment, by encoding different enhancement processing methods, the obtained enhancement processing method encoding information can be used to distinguish enhancement processing methods or to call enhancement processing methods from other storage spaces. It also facilitates the use of a unified format to characterize enhancement processing methods.

[0056] Generally, enhanced processing methods are not applied to only a single business scenario but may be used in multiple business scenarios. The content related to enhanced processing methods is not suitable for storage in the storage space corresponding to a specific business scenario; instead, it needs to be stored in a public data center or data platform. This embodiment can use enhanced processing method encoding information as a unique identifier for different enhanced processing methods, for distinguishing and invoking them. For example, the enhanced processing method encoding information can be information formed using one-hot encoding.

[0057] S104. Based on the object to be processed, the object description information, and the enhancement method encoding information, predict the object enhancement evaluation result of the object to be processed. The object enhancement evaluation result is used to characterize the effect of enhancing the object to be processed based on the enhancement processing method.

[0058] In this embodiment, this step can be used to predict the degree of positive feedback between the processed object and the original object after the object is enhanced according to the enhancement processing method. The degree of positive feedback can be determined by the object enhancement evaluation results.

[0059] In this embodiment, the enhancement evaluation probability of a specific object can be used to characterize this. For example, a higher enhancement evaluation probability indicates that the enhancement method cannot provide better positive feedback, and the object to be processed can be considered unsuitable for this enhancement method. Conversely, a lower enhancement evaluation probability indicates that the enhancement method can be used to enhance the object, because the enhanced object will provide better positive feedback when sent to the client.

[0060] In this embodiment, the object augmentation evaluation result can be determined by constructing and pre-training an object augmentation evaluation model. The object to be processed, the object description information, and the augmentation method encoding information can all be used as input information for the object augmentation evaluation model. The object augmentation evaluation model performs object feature analysis on the object to be processed and extracts object-related metadata from the object description information and the augmentation method encoding information. Finally, based on the obtained object feature vector and metadata effect vector, a data value between 0 and 1 is obtained through the fully connected and normalization processing logic in the positive feedback detection model. This data value can be regarded as the object augmentation evaluation probability as the final object augmentation evaluation result.

[0061] It should be noted that, in this embodiment, the objects managed are preferably visual objects such as images or videos. The quality of such visual objects is mainly reflected in the viewer's experience, most directly in factors such as image quality, clarity, and distortion. These factors are all related to the pixel information that constitutes the object. Based on this, the construction idea of ​​the object enhancement evaluation model can be described as follows: a portion of the network structure built into the object enhancement evaluation model can focus more on the pixel feature information of the object when analyzing the input object. For example, this portion of the network structure can be used to downsample the object features, so as to ensure that the pixel feature information has a better weight in the final decision.

[0062] Meanwhile, in addition to constructing a downsampling network structure to focus on the pixel feature information of the object itself, the object augmentation evaluation model also needs to construct a network structure to analyze the augmentation method encoding information and the object description information, etc., so as to capture the object's metadata information from the object description information and the augmentation method encoding information. The captured metadata information, such as the classification information of visual objects, whether they belong to computer graphics type, and their resolution, can all participate in the decision of whether to enhance the object's positive feedback, so as to improve the accuracy of the positive feedback decision.

[0063] Based on the network structure constructed above, we can obtain the object's own feature information (such as the pixel feature information of the visual object itself) and the metadata information related to the object (mainly from the analysis of the enhancement method encoding information and object description information), which play a key role in the object's positive feedback decision. Finally, the object enhancement evaluation model can also construct a network layer to concatenate the object's own feature information and metadata information, and a fully connected network layer to perform fusion calculation and dimensionality reduction processing on the concatenated feature information, so as to ensure that the fully connected network layer is ultimately a one-dimensional value. The object enhancement evaluation model can also include a normalization network layer to normalize the one-dimensional array, thereby obtaining the object enhancement evaluation probability between 0 and 1 output by the object enhancement evaluation model, as the final object enhancement evaluation result.

[0064] Following the above description, the network structure used to obtain the object's own feature information in the object enhancement evaluation model can be considered as a dual-loop network structure that continuously extracts and downsamples features. Furthermore, for the multiple network modules within the inner loop, long-distance skip connections are added between each pair of network modules. This dual-loop network structure, with its specific configuration, maintains a higher weight for the object's own feature information.

[0065] S105. If the object enhancement evaluation result meets the preset conditions, then the object to be processed is enhanced based on the enhancement processing method.

[0066] S104. If the object enhancement evaluation result is less than the set probability threshold, the object to be processed is allowed to be enhanced through the enhancement processing method.

[0067] In this embodiment, the above steps can predict the object enhancement evaluation result after applying enhancement processing to the object to be processed, resulting in positive feedback. This object enhancement evaluation result can be characterized as an object enhancement evaluation probability. For example, a preset condition can be set such that the object enhancement evaluation probability is less than a preset probability threshold. In this preset condition setting, if the preset condition is met, it can be considered that the probability of positive feedback after applying enhancement processing to the object to be processed is very high, thus allowing this step to continue using enhancement processing to enhance the object to be processed.

[0068] In this embodiment, the preset probability threshold can be an empirical value, determined by historical feedback information from a certain scale of objects. Specifically, for a certain scale of objects, enhancement processing can be performed on each object. Then, the enhanced and unprocessed objects are sent to the client respectively. After a certain period of time, a certain business indicator is statistically analyzed on the enhanced and unprocessed objects. The statistical results are used as historical feedback information. By analyzing the historical feedback information, the critical probability value at which positive feedback can begin to be generated can be determined, and this critical probability value can be used as the preset probability threshold.

[0069] This embodiment provides an object management method that pre-determines whether an object is suitable for enhancement processing before performing enhancement processing on it. The logic of this decision management mainly involves predicting whether the object after enhancement processing will produce a positive feedback effect compared to the original object. Whether positive feedback can be generated is determined by comparing the determined enhancement evaluation result with preset conditions. This technical solution assumes that when the enhancement evaluation result meets the preset conditions, the enhanced object is superior to the original object. Therefore, in this case, enhancement processing is allowed. This decision management method effectively filters whether to perform object enhancement processing, ensuring that only objects that produce positive feedback are enhanced, thereby saving computing resources and reducing enhancement processing costs. Simultaneously, it can improve enhancement processing efficiency and increase the positive feedback effect without changing the number of objects being enhanced.

[0070] As a first optional embodiment of this example, the process of determining the object to be processed that matches the acquired object identifier can be optimized to the following steps:

[0071] a1) Parse the object identifier in the received object retrieval request.

[0072] It should be noted that, in one scenario of determining the object to be processed in this optional embodiment, the executing entity can be used as the server of an application, and the object to be processed can be determined based on the object retrieval request sent by the client of the application.

[0073] Specifically, this step can be used to analyze the received object retrieval request and obtain the object identifier contained in the request. This object identifier can be considered as the unique tag information of the requested object.

[0074] b1) If the requested object corresponding to the object identifier is not found in the object database, the target original object found in the original object database relative to the object identifier will be determined as the object to be processed.

[0075] In this embodiment, the object database can be understood as a collection of historically requested objects. Specifically, all objects determined to be sent to the requesting end based on the received object requests can be recorded in this object database. This allows the same object to be retrieved directly from the object database when a subsequent request for the same object occurs, avoiding the need to perform enhancement processing on the original object content again.

[0076] It is known that if the requested object is located in the object database through the object identifier of the requested object, the requested object can be directly retrieved from the object database and fed back to the requesting end as the request response content. In this case, it is not necessary to perform a decision on whether to perform enhancement processing on the requested object through the object management method provided in this embodiment.

[0077] As described above, in this embodiment, if the object identifier of the requested object does not exist in the object database, it can be considered that this step is necessary to obtain the requested object and treat it as an object to be processed. Specifically, the requested object can be searched for in the original object database by its object identifier. In this embodiment, the requested object corresponding to the object identifier in the original object database is recorded as the target original object, and this target original object can be identified as an object to be processed.

[0078] The original object database can be understood as a collection of all original objects that have not undergone secondary processing. The objects stored in it can be original objects uploaded by the object creation client or original objects imported from external sources. The original object database is more like a resource pool associated with this execution entity, used to aggregate object information from various sources.

[0079] This embodiment provides a specific implementation for determining the objects to be processed. This method can better filter out objects that require enhanced processing, providing basic data support for subsequent decisions on whether to perform enhanced processing.

[0080] In this first optional embodiment, the enhancement processing method for determining the object to be processed and the object description information of the object to be processed can be further specified as the following steps:

[0081] a2) Make a decision on the object to be processed using the given augmented decision model, and obtain the augmented processing method output by the augmented decision model.

[0082] In this embodiment, a pre-defined enhancement decision model can be used to determine what kind of enhancement is needed for the object to be processed, thereby outputting the decided enhancement processing method. The enhancement decision model can use the specific enhancements to be performed for different enhancement processing methods as known information, and it can also use the processing concerns involved in different enhancement processing methods (such as whether there is noise, whether there is blur, whether the clarity is lower than a set value, etc.) as known information. Therefore, for an object to be processed, by analyzing the various attribute information carried by the object, it can be determined which processing concern the object satisfies, and finally, the enhancement processing method corresponding to the satisfied processing concern is selected as the enhancement processing method for the object.

[0083] b2) Summarize the object-related attribute information of the object to be processed to form the object description information.

[0084] It is known that after identifying the object to be processed, it carries basic attribute information such as object name, object category, and object size. It can also obtain information related to object creation or other dimensions, such as the tools used to obtain the object and whether the object underwent secondary processing. This step can summarize the various attribute information carried by the object to be processed, forming a text-based object description.

[0085] This embodiment provides a specific implementation of the enhancement processing method and object description information determination. This method enables the determination of object description information for objects requiring enhancement processing and allows for better selection of enhancement processing methods. It provides fundamental data support for subsequent decisions on whether to perform enhancement processing.

[0086] As a second optional embodiment of this embodiment, the determination of the object description information of the object to be processed and the enhancement method encoding information of the enhancement processing method can be specified as follows: obtain the binary code corresponding to the enhancement processing method from a pre-set enhancement method encoding association information set and use the binary code as the enhancement method encoding information.

[0087] It is understood that in this embodiment, different enhancement processing methods can be encoded and recorded in the enhancement processing method encoding association information set. After determining the enhancement processing method corresponding to the object to be processed, this step can obtain the encoding information of the enhancement processing method by searching the strategy encoding association information. In this embodiment, the encoding information is recorded as enhancement processing method encoding information. Among them, the enhancement processing method can be binary encoded in the form of one-hot encoding.

[0088] The above-described technical solution in this embodiment provides a specific implementation of the enhancement method encoding information. It also provides basic data support for subsequent decisions on whether to perform enhancement processing.

[0089] As a third optional embodiment of this example, predicting the object enhancement evaluation result of the object to be processed after being processed by the enhancement method, based on the object to be processed, the object description information, and the enhancement method encoding information, can be specified as the following steps:

[0090] a3) Input the object to be processed, the object description information, and the enhancement method encoding information as input data into the object enhancement evaluation model.

[0091] Specifically, in this embodiment, a positive feedback prediction after the object to be processed has undergone enhancement is mainly achieved through an object enhancement evaluation model. This step can use the determined object to be processed, object description information, and enhancement method encoding information as input data for the object enhancement evaluation model.

[0092] b3) Determine the object feature vector of the object to be processed through the first sub-analysis model in the object enhancement evaluation model.

[0093] In this embodiment, the object enhancement evaluation model includes a first sub-analysis model, which can be used to perform feature analysis on the object to be processed. This step can obtain the object feature vector output by the first sub-analysis model.

[0094] Based on the above description of the constructed object enhancement evaluation model, it can be seen that predicting whether the object to be processed will have positive feedback after enhancement requires obtaining the feature information of interest in positive feedback prediction. This type of feature information can be obtained by constructing a suitable network structure. In this embodiment, the first sub-analysis model can be considered to possess a network structure for extracting the feature information of interest, and the feature information of interest output by this first sub-analysis model can be denoted as the object feature vector.

[0095] The first sub-analysis model can be considered to contain a dual-loop network structure. This dual-loop network structure can continuously extract and downsample the features of the object to be processed, thereby filtering out the features of interest. Considering that continuous feature extraction and downsampling operations will cause the weight of the features of interest to decrease, long-distance skip connections are added to the feature processing module, which is the inner loop in the dual-loop network structure. With the support of these skip connections, the features of interest can be maintained at a high weight during the feature extraction process.

[0096] It should be noted that this first sub-analysis model, while containing a dual recurrent network structure, also includes convolutional layers, which can also be used for feature extraction of the object to be processed. It is primarily used to transform non-data objects (such as images or videos) from their original state to a feature vector state. Furthermore, when the object to be processed is a video, multiple randomly extracted video frames can be simultaneously input into the first sub-analysis model. The first sub-analysis model can output feature vectors for each of the input video frames, and these multiple sets of output feature vectors can all be considered as object feature vectors of the object to be processed (the video).

[0097] c3) By analyzing the object description information and enhancement method encoding information through the second sub-analysis model in the object enhancement evaluation model, the metadata feature vector of the object to be processed is obtained.

[0098] In this embodiment, the object enhancement evaluation model also includes a second sub-analysis model. The second sub-analysis model can be used to perform feature analysis on the determined object description information and enhancement method encoding information. This step can obtain the feature vector output by the second sub-analysis model, which can be regarded as the metadata feature vector of the object to be processed.

[0099] Similarly, based on the above description of the constructed object augmentation evaluation model, it can be seen that predicting whether the object to be processed will have positive feedback after augmentation requires not only obtaining the relevant feature information of the object itself in the positive feedback prediction, but also introducing other dimensional feature information to assist in the positive feedback prediction judgment. This includes feature information that serves as metadata features of the object to be processed, output through analysis of the augmentation method encoding information and object description information. Obtaining this type of metadata feature information can also be achieved by constructing a suitable network structure.

[0100] The second sub-analysis model in this embodiment can be considered as a model with a network structure for extracting metadata feature information, specifically represented as metadata feature vectors. This second sub-analysis model can be considered as a combined network structure containing multiple sets of fully connected layers and normalized network layers. This combined network structure can filter out suitable feature information as metadata of the object to be processed from the enhanced encoding information and object description information.

[0101] It can be seen that when the object to be processed is an image, the object description information mainly includes a summary of the relevant attribute information of the image; when the object to be processed is a video, the object description information mainly includes a summary of the relevant attribute information of the video.

[0102] d3) By concatenating the object feature vector and the metadata feature vector in the evaluation prediction sub-model of the object enhancement evaluation model, a feature concatenation vector is obtained, and the object enhancement evaluation result is output by dimensionality reduction processing of the feature concatenation vector.

[0103] In this embodiment, the first sub-analysis model and the second sub-analysis model involved in the above steps can obtain the corresponding output object feature vector and metadata feature vector. This step can be used to continue to analyze and calculate the object feature vector and metadata feature vector to obtain the final object enhancement evaluation result.

[0104] In this embodiment, the object feature vector and metadata feature vector are first concatenated using this evaluation and prediction sub-model; then, the concatenated vector undergoes fully connected processing and dimensionality reduction to finally output a one-dimensional value. The metadata feature vector can be considered a two-dimensional feature vector, while the object feature vector exhibits different dimensions depending on the object being processed. For example, when the object is an image, the object feature vector relative to the image is typically a four-dimensional feature vector, with one dimension representing the batch size, one dimension representing the number of channels, and two dimensions representing the image or / or screen size. Therefore, vector dimension alignment is required before concatenation, such as tiling the object feature vector.

[0105] Furthermore, assuming the object to be processed is a video, the object feature vector can be obtained from each frame extracted from the video. This evaluation prediction sub-model needs to concatenate the object feature vectors to process the concatenated object feature vectors into temporal feature vectors, and then align the temporal feature vectors with the metadata feature vectors to the same dimension before concatenating them.

[0106] The above-described technical solution in this embodiment provides a specific implementation for determining the object augmentation evaluation result value. By pre-creating and training the object augmentation evaluation model, the object augmentation evaluation result of the object to be processed can be accurately determined. This ensures that effective screening of whether or not to perform object augmentation can be achieved through this decision management method, thereby saving computing resources and reducing the cost of augmentation processing.

[0107] Based on the above optional embodiments, to facilitate a better understanding of the network model architecture of the object augmentation evaluation model, this embodiment provides a detailed description of the network structure of the first sub-analysis model and the second sub-analysis model. Specifically, Figure 1b The structural diagrams of the first and second sub-analysis models of the object enhancement evaluation model involved in the object management method provided in this embodiment are given.

[0108] like Figure 1b As shown, the first sub-analysis model 11 includes: a first two-dimensional convolutional layer 110, at least one feature processing module 111, and a second two-dimensional convolutional layer 112; the second sub-analysis model 12 includes: at least one descriptive information analysis module 120, and each descriptive information analysis module 120 includes: a fully connected layer, a normalized network layer, and an activation function layer.

[0109] Furthermore, such as Figure 1b As shown, the first two-dimensional convolutional layer 110 is connected to the first feature processing module 111, and the second two-dimensional convolutional layer 112 is connected to the last feature processing module 111. The output information of the object to be processed by the first two-dimensional convolutional layer 110 is used as the original input information of the first feature processing module 111. It can also be seen that at least one feature processing module 111 is connected sequentially, and the output information of the previous feature processing module 111 is used as the original input information of the next adjacent feature processing module 111. The output information of the last feature processing module 111 is used as the object feature vector of the object to be processed after being processed by the second two-dimensional convolutional layer 112.

[0110] from Figure 1b It can also be seen that in the second sub-analysis model 12, both object description information and enhancement method encoding information are used as input information, processed by at least one description information analysis module 120, and the processed output information is used as the metadata feature vector of the object to be processed.

[0111] Based on the above optimizations, and continuing from the above description, Figure 1c A schematic diagram of the specific structure of a feature processing module in the first sub-analysis model involved in the object association method provided in this embodiment is given.

[0112] like Figure 1c As shown, a feature processing module 111 includes at least one sequentially connected sub-feature processing module 1110, and an average pooling layer 1111 connected to the last said sub-feature processing module; the input information of the first sub-feature processing module 1110 in the feature processing module 111 is the original input information of the feature processing module.

[0113] Simultaneously, the fused information formed by fusing the output information of the previous sub-feature processing module 1110 with the original input information serves as the input information of the next adjacent sub-feature processing module 1110; the output information of the last sub-feature processing module 1110 is processed by the average pooling layer 1111 to output average pooling information; the fused information formed by fusing the average pooling information with the secondary processing information serves as the output information of the feature processing module, wherein the secondary processing information is the information obtained after performing a two-dimensional convolution process and an average pooling process on the original input information of the feature processing module.

[0114] It should be noted that, depending on the object type of the object to be processed, the specific implementation of the evaluation and prediction sub-model in obtaining the object enhancement evaluation result based on the object feature vector and metadata feature vector also varies.

[0115] As a third optional embodiment of this example, based on the above embodiments, when the object to be processed is an image to be processed, the feature concatenation vector obtained by concatenating the object feature vector and the metadata feature vector through the evaluation prediction sub-model in the object enhancement evaluation model, and the object enhancement evaluation result output by dimensionality reduction processing of the feature concatenation vector, can be specified as follows:

[0116] a4) Tile the object feature vector of the object to be processed to obtain the object tiled feature vector.

[0117] In this embodiment, when the object to be processed is an image, the object feature vector output by the first sub-analysis model can be represented as an image feature vector, which is typically a multi-dimensional feature vector. Before concatenating the object feature vector with the metadata feature vector, this step can first perform a tiling process on the object feature vector to align the feature dimensions of the two feature vectors. This tiling process can be understood as a dimensionality reduction process for the object feature vector; through this step, an object tiling feature vector with the same dimensions as the metadata feature vector can be obtained.

[0118] b4) Concatenate the object's flat feature vector with the metadata feature vector to form a first feature concatenation vector.

[0119] In this embodiment, the object tiling feature vector and the metadata feature vector can be concatenated through this step to obtain the first feature concatenation vector.

[0120] c4) Perform full-connection processing on the first feature splicing vector, and perform one-dimensional dimensionality reduction and normalization processing on the fully connected feature vector obtained by the processing, and determine the value obtained after processing as the object enhancement evaluation result of the image.

[0121] In this embodiment, the evaluation prediction sub-model can be considered to include a fully connected network layer. The first feature splicing vector formed by splicing can be used as the input information of the fully connected network layer. The fully connected network layer performs dimensionality reduction processing on the first feature splicing vector and then outputs a one-dimensional fully connected feature vector. Finally, this step can normalize the one-dimensional fully connected feature vector to obtain a value between 0 and 1, which can be used as the object augmentation evaluation probability and as the final object augmentation evaluation result of this embodiment.

[0122] It is known that, in specific implementations, evaluation and prediction sub-models with different network structures can be constructed according to different types of managed objects. For example, when the object to be processed is an image, the network structure in the evaluation and prediction sub-model can include a tiling processing layer; it can also include a feature stitching processing layer, a fully connected network layer, and a normalization processing layer.

[0123] The above-described technical solution in this embodiment provides a specific implementation for determining the object enhancement evaluation result when the object to be processed is an image. This method can accurately determine the object enhancement evaluation result for the image to be processed, which is used to effectively predict the positive feedback of the enhanced video. Finally, the object enhancement evaluation result is used to decide whether to perform enhancement processing on the image to be processed.

[0124] As a fourth optional embodiment of this embodiment, based on the above embodiments, when the object to be processed is a video to be processed, at least one video frame extracted from the video to be processed can be optimized and input into the first sub-analysis model respectively; the image feature vectors output by the first sub-analysis model relative to the at least one video frame respectively constitute the object feature vector.

[0125] It is understandable that the object to be processed can also be a video of type video. When the object to be processed is a video, one or more frames can be randomly selected from the video as input information for the first sub-analysis model. The first sub-analysis model can perform feature analysis on each frame of the video, thereby obtaining the image feature vector for each frame. The image feature vectors constitute the object feature vector of the video to be processed.

[0126] Based on the above optimizations, this optional embodiment can further optimize the object augmentation evaluation result by concatenating the object feature vector and the metadata feature vector through the evaluation prediction sub-model in the object augmentation evaluation model, and by performing dimensionality reduction processing on the feature concatenated vector.

[0127] a5) The image feature vectors that constitute the object feature vector are spliced ​​together to form the image splicing feature vector, and the image splicing feature vector is processed in the temporal domain through a three-dimensional convolutional layer to obtain the processed temporal feature vector.

[0128] In this embodiment, the image feature vectors corresponding to each frame of the video input to the first sub-analysis model can be obtained, and the image feature vectors can be stitched together according to the time dimension. Then, through this step, the image feature vectors can be further processed by a three-dimensional convolutional layer for temporal feature extraction, thereby obtaining a temporal feature vector.

[0129] b5) The time-series feature vector is flattened, and the flattened feature vector is concatenated with the metadata feature vector to form a second feature concatenation vector.

[0130] It can be seen that this temporal feature vector adds a time dimension compared to the aforementioned image feature vector. The metadata feature information output by the second sub-analysis model is a two-dimensional feature vector. Therefore, before concatenating the temporal feature vector with the metadata feature vector, the temporal feature vector also needs to be flattened. Then, this step concatenates the flattened feature vector with the metadata feature vector to form the second feature concatenation vector.

[0131] c5) Perform full-connection processing on the second feature splicing vector, and perform one-dimensional dimensionality reduction and normalization processing on the fully connected feature vector obtained by the processing, and determine the value obtained after processing as the object enhancement evaluation result of the video.

[0132] In this embodiment, similar to the above description of the evaluation prediction sub-model, this optional embodiment can also be considered to include a fully connected network layer in the evaluation prediction sub-model, and the second feature splicing vector formed by splicing can be used as the input information of the fully connected network layer. The fully connected network layer performs dimensionality reduction processing on the second feature splicing vector and then outputs a one-dimensional fully connected feature vector. Finally, this step can normalize the one-dimensional fully connected feature vector to obtain a value between 0 and 1, which can be used as the object augmentation evaluation probability and as the final object augmentation evaluation result of this embodiment.

[0133] By comparison, it can be found that, unlike the network structure used in the above-mentioned evaluation of object enhancement results for the processed image, the network structure used in this optional embodiment for the evaluation of object enhancement results for the processed video may include a newly added image feature stitching layer and a three-dimensional convolutional processing module. Following this, it may include a tiling processing layer for tiling effect vectors, a feature stitching processing layer, a fully connected network layer, and a normalization processing layer. The three-dimensional convolutional processing module can be considered to include multiple sequentially connected three-dimensional convolutional processing layers, as well as an average pooling layer. The information output by the average pooling layer can be denoted as a temporal feature vector.

[0134] The above-described technical solution in this embodiment provides a specific implementation for determining the object enhancement evaluation result when the object to be processed is a video. This method can accurately determine the object enhancement evaluation result for the video to be processed, which is used to effectively predict the positive feedback of the enhanced video. Finally, the decision on whether to perform enhancement processing on the video to be processed is made based on the object enhancement evaluation result.

[0135] It should be noted that the object augmentation evaluation model used in this embodiment to determine the object augmentation evaluation results can be pre-trained. Specifically, it can collect feedback from augmented objects online, and construct sample objects based on identified objects that have generated positive feedback. The probability of positive feedback can be determined based on the ratio of the number of positive feedback instances to the total number of objects distributed. Therefore, the determined sample objects and their positive feedback probabilities can form a sample data set, which, combined with a given loss function, can be used to train the initial object augmentation evaluation model.

[0136] As a fifth optional embodiment of this example, the method can be further optimized based on the above embodiments by including:

[0137] a6) The enhanced object after processing the object to be processed by the enhancement processing method is taken as the target object and recorded in the object database.

[0138] In this embodiment, in one scenario example described above, the object to be processed can be determined through an object request uploaded by the application client. Therefore, this optional embodiment can add an executable step. Specifically, this step can use the enhanced object, processed by the enhancement method, as the target object and send it to the application client that needs the object. Simultaneously, to ensure that subsequent object requests can quickly obtain the requested object, this step can also record the enhanced object in the object database.

[0139] b6) If the object enhancement evaluation result does not meet the preset conditions, the object to be processed is directly used as the target object and recorded in the object database.

[0140] In this embodiment, as another comparison branch for determining the preset conditions of object enhancement evaluation results, when the object enhancement evaluation probability is greater than or equal to the set probability threshold, it can be considered that the enhancement processing method is not suitable for enhancing the object to be processed in this case. Therefore, in this case, the object to be processed can be directly used as the target object to be sent through this newly added step, and sent to the application client that needs the object. Similarly, in order to ensure that subsequent object requests can quickly obtain the requested object, this step can also record the unenhanced object to be processed as the target object in the object database.

[0141] It is understood that during the execution of the above steps in this embodiment, there is also an execution branch, namely, that the enhanced decision model determines that the object to be processed does not need to be enhanced. In the case of this execution branch, the object to be processed can also be directly used as the target object for direct distribution to the application client and recorded in the object database.

[0142] The above technical solution of this embodiment provides a new execution logic for the object management method provided in this embodiment. This new step can ensure that the application client can quickly obtain the requested object.

[0143] To better understand the execution logic of the method provided in this embodiment in practical applications, this embodiment can illustrate the application of the object association method in an example scenario. Specifically, the example scenario could be an entertainment application software, which includes an application client and an application server. For example, Figure 1d A flowchart illustrating an implementation of the object management method provided in this embodiment in a specific scenario is given.

[0144] like Figure 1d As shown, the execution of the method provided in this embodiment can be triggered through the interaction between the application client 13 and the application server 14. Specifically, the execution process can be described as follows:

[0145] S1. The application client sends an image retrieval request to the application server for any image.

[0146] S2. The application server analyzes the received image acquisition request and obtains the image identifier.

[0147] S3. The application server determines whether the target image corresponding to the image identifier exists in the image database. If it exists, proceed to S4; otherwise, proceed to S5.

[0148] S4. The application server returns the target image to the application client.

[0149] S5. The application server searches for the original image in the original image database using the image identifier, uses the original image as the image to be processed, and inputs it into the enhanced decision model.

[0150] S6. When the application server determines through the enhanced decision model that no enhancement processing is needed for the image to be processed, it sends the image to be processed as the target image to the application client and stores the target object in the image database.

[0151] S7. After the application server determines the enhancement processing method of the image to be processed through the enhancement decision model, it determines the image description information of the image to be processed and the enhancement method encoding information of the enhancement processing method.

[0152] S8. The application server inputs the image to be processed, image description information, and enhancement method encoding information into the object enhancement evaluation model, and obtains the object enhancement evaluation result output by the object enhancement evaluation model.

[0153] S9. When the application server determines that the enhancement evaluation probability of the object is less than the set probability threshold, it enhances the image to be processed through the enhancement processing method, and records the enhanced image in the image database and sends it to the application client as the target object.

[0154] S10. When the application server determines that the enhancement evaluation probability of the object is greater than or equal to the set probability threshold, it directly sends the image to be processed as the target image to the application client and stores the target object in the image database.

[0155] Figure 2 This is a schematic diagram of an object management device provided in an embodiment of the present disclosure. This embodiment is applicable to object management situations. The device can be implemented by software and / or hardware and can be configured in a terminal and / or server to implement the object management method in this embodiment of the present disclosure. Specifically, the device may include: a first determining module 21, a second determining module 22, a third determining module 23, a result evaluation module 24, and an object management module 25.

[0156] The first determining module 21 is used to determine the object to be processed that matches the acquired object identifier;

[0157] The second determining module 22 is used to determine the enhanced processing method of the object to be processed and the object description information of the object to be processed.

[0158] The third determining module 23 is used to determine the enhancement method encoding information of the enhancement processing method according to the enhancement processing method;

[0159] The result evaluation module 24 is used to predict the object enhancement evaluation result of the object to be processed based on the object to be processed, the object description information and the enhancement method encoding information. The object enhancement evaluation result is used to characterize the effect of enhancing the object to be processed based on the enhancement processing method.

[0160] The object management module 25 is used to perform enhancement processing on the object to be processed based on the enhancement processing method if the object enhancement evaluation result meets the preset conditions.

[0161] The object management device provided in this embodiment is equivalent to making a decision-making management on whether an object is suitable for enhancement processing before performing enhancement processing on it. The logic of this decision-making management mainly involves predicting whether the object processed by the enhancement method will produce a positive feedback effect compared to the original object, and whether positive feedback can be generated. This is determined by comparing the determined enhancement evaluation result with preset conditions. This technical solution assumes that when the enhancement evaluation result meets the preset conditions, the object after enhancement processing is superior to the original object. Therefore, in this case, enhancement processing can be allowed. This decision-making management method can effectively screen whether to perform object enhancement processing, ensuring that only objects that produce positive feedback are enhanced, thereby saving computing resources and reducing enhancement processing costs. Simultaneously, it can improve enhancement processing efficiency and increase the positive feedback effect without changing the number of objects being enhanced.

[0162] Furthermore, the first determining module 21 can specifically be used for:

[0163] Parse the received object to obtain the object identifier in the request;

[0164] If the requested object corresponding to the object identifier is not found in the object database, the target original object found in the original object database relative to the object identifier will be determined as the object to be processed.

[0165] Furthermore, the second determining module 22 can specifically be used for:

[0166] The enhanced processing method for the object to be processed is determined by using a trained enhanced decision model;

[0167] The object-related attribute information of the object to be processed is summarized to form the object description information.

[0168] Furthermore, the third determining module 23 can specifically be used for:

[0169] Obtain the binary code corresponding to the enhancement processing method from a pre-set set of enhancement method encoding association information, and use the binary code as the enhancement method encoding information.

[0170] Furthermore, the results evaluation module 24 may specifically include:

[0171] The input unit is used to input the object to be processed, the object description information, and the enhancement method encoding information as input data into the object enhancement evaluation model;

[0172] The first analysis unit is used to determine the object feature vector of the object to be processed through the first sub-analysis model in the object enhancement evaluation model;

[0173] The second analysis unit is used to analyze the object description information and enhancement method encoding information through the second sub-analysis model in the object enhancement evaluation model to obtain the metadata feature vector of the object to be processed.

[0174] The evaluation result prediction unit is used to obtain a feature concatenation vector by concatenating the object feature vector and the metadata feature vector through the evaluation prediction sub-model in the object enhancement evaluation model, and output the object enhancement evaluation result by dimensionality reduction processing of the feature concatenation vector.

[0175] Furthermore, the first sub-analysis model includes: a first two-dimensional convolutional layer, at least one feature processing module, and a second two-dimensional convolutional layer;

[0176] The second sub-analysis model includes: at least one descriptive information analysis module, each descriptive information analysis module including: a fully connected layer, a normalized network layer, and an activation function layer;

[0177] The first two-dimensional convolutional layer is connected to the first feature processing module, and the second two-dimensional convolutional layer is connected to the last feature processing module.

[0178] The output information of the object to be processed by the first two-dimensional convolutional layer is used as the original input information of the first feature processing module;

[0179] The at least one feature processing module is connected sequentially, and the output information of the previous feature processing module is used as the original input information of the next adjacent feature processing module.

[0180] The output information of the last feature processing module, after being processed by the second two-dimensional convolutional layer, is used as the object feature vector of the object to be processed.

[0181] The output information of the object description information and the enhancement method encoding information after being processed by the at least one description information analysis module is used as the metadata feature vector of the object to be processed.

[0182] Furthermore, each feature processing module includes at least one sequentially connected sub-feature processing module, and an average pooling layer connected to the last said sub-feature processing module;

[0183] The input information of the first sub-feature processing module in each feature processing module is the original input information of the feature processing module;

[0184] The fused information formed by fusing the output information of the previous sub-feature processing module with the original input information is used as the input information of the next adjacent sub-feature module;

[0185] The output information of the last sub-feature processing module is processed by the average pooling layer to output average pooling information.

[0186] The fused information formed by fusing the average pooling information and the secondary processing information is used as the output information of the feature processing module. The secondary processing information is the information obtained by performing a two-dimensional convolution process and an average pooling process on the original input information of the feature processing module.

[0187] Furthermore, the evaluation result prediction unit can be specifically used for:

[0188] When the object to be processed is an image, the object feature vector of the object to be processed is tiled to obtain the object tiled feature vector.

[0189] The object's flat feature vector is concatenated with the metadata feature vector to form a first feature concatenation vector;

[0190] The first feature concatenation vector is subjected to fully connected processing, and the fully connected feature vector obtained after processing is subjected to one-dimensional dimensionality reduction and normalization processing. The value obtained after processing is determined as the object enhancement evaluation result of the image.

[0191] Furthermore, when the object to be processed is a video to be processed, at least one video frame extracted from the video to be processed is respectively input into the first sub-analysis model; the image feature vectors output by the first sub-analysis model relative to the at least one video frame respectively constitute the object feature vector;

[0192] Correspondingly, the evaluation result prediction unit can also be used for:

[0193] The image feature vectors that constitute the object feature vector are spliced ​​together to form the image splicing feature vector. The image splicing feature vector is then processed in the temporal domain through a three-dimensional convolutional layer to obtain the processed temporal feature vector.

[0194] The time-series feature vector is flattened, and the flattened feature vector is concatenated with the metadata feature vector to form a second feature concatenation vector.

[0195] The second feature concatenation vector is subjected to fully connected processing, and the fully connected feature vector obtained after processing is subjected to one-dimensional dimensionality reduction and normalization processing. The value obtained after processing is determined as the object enhancement evaluation result of the video.

[0196] Furthermore, the device also includes:

[0197] The recording module is used to record the enhanced object after processing the object to be processed through the enhancement processing method as the target object in the object database; if the object enhancement evaluation result does not meet the preset conditions, the object to be processed is directly used as the target object and recorded in the object database; the preset conditions are that the object enhancement evaluation probability, which characterizes the object enhancement evaluation result, is less than the set probability threshold.

[0198] The above-described apparatus can execute the methods provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the methods.

[0199] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.

[0200] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Reference is made below. Figure 3 It illustrates a computer device suitable for implementing embodiments of the present disclosure (e.g., Figure 3The diagram below shows the structure of the terminal device or server 30. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 3 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0201] like Figure 3 As shown, the computer device 30 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 31, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 32 or a program loaded from a storage device 38 into a random access memory (RAM) 33. The RAM 33 also stores various programs and data required for the operation of the computer device 30. The processing unit 31, the ROM 32, and the RAM 33 are interconnected via a bus 35. An edit / output (I / O) interface 34 is also connected to the bus 35.

[0202] Typically, the following devices can be connected to I / O interface 34: input devices 36 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 37 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 38 including, for example, magnetic tapes, hard disks, etc.; and communication devices 39. Communication device 39 allows computer device 30 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 A computer device 30 with various devices is shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have instead.

[0203] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 39, or installed from a storage device 38, or installed from a ROM 32. When the computer program is executed by the processing device 31, it performs the functions defined in the methods of embodiments of this disclosure.

[0204] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0205] The computer device provided in this embodiment and the object management method provided in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0206] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the object management method provided in the above embodiments.

[0207] It should be noted that the computer-readable medium described above in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0208] In this disclosure, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0209] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0210] The aforementioned computer-readable medium may be included in the aforementioned computer device; or it may exist independently and not assembled into the computer device.

[0211] The aforementioned computer-readable medium carries one or more programs that, when executed by the computer device, cause the computer device to:

[0212] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0213] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0214] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".

[0215] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0216] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0217] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0218] Furthermore, although the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0219] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. An object management method, characterized in that, include: Identify the objects to be processed that match the acquired object identifiers; Determine the enhancement processing method for the object to be processed and the object description information of the object to be processed; Based on the enhancement processing method, determine the enhancement method encoding information of the enhancement processing method; Based on the object to be processed, the object description information, and the enhancement method encoding information, the object enhancement evaluation result of the object to be processed is predicted. The object enhancement evaluation result is used to characterize the effect of enhancing the object to be processed based on the enhancement processing method. If the object enhancement evaluation result meets the preset conditions, then the object to be processed is enhanced based on the enhancement processing method.

2. The method according to claim 1, characterized in that, The process of determining the object to be processed that matches the acquired object identifier includes: Parse the received object to obtain the object identifier in the request; If the requested object corresponding to the object identifier is not found in the object database, the target original object found in the original object database relative to the object identifier will be determined as the object to be processed.

3. The method according to claim 1, characterized in that, The determination of the enhanced processing method for the object to be processed and the object description information of the object to be processed includes: The enhanced processing method for the object to be processed is determined by using a trained enhanced decision model; The object-related attribute information of the object to be processed is summarized to form the object description information.

4. The method according to claim 1, characterized in that, The step of determining the enhancement method encoding information of the enhancement processing method according to the enhancement processing method includes: Obtain the binary code corresponding to the enhancement processing method from a pre-set set of enhancement method encoding association information, and use the binary code as the enhancement method encoding information.

5. The method according to claim 1, characterized in that, The step of predicting the object enhancement evaluation result of the object after processing by the enhancement method based on the object to be processed, the object description information, and the enhancement method encoding information includes: The object to be processed, the object description information, and the enhancement method encoding information are used as input data and input into the object enhancement evaluation model. The object feature vector of the object to be processed is determined by the first sub-analysis model in the object enhancement evaluation model. The second sub-analysis model in the object enhancement evaluation model is used to analyze the object description information and enhancement method encoding information to obtain the metadata feature vector of the object to be processed. The object enhancement evaluation result is output by concatenating the object feature vector and the metadata feature vector through the evaluation prediction sub-model in the object enhancement evaluation model, and by performing dimensionality reduction processing on the feature concatenated vector.

6. The method according to claim 5, characterized in that, The first sub-analysis model includes: a first two-dimensional convolutional layer, at least one feature processing module, and a second two-dimensional convolutional layer; The second sub-analysis model includes: at least one descriptive information analysis module, each descriptive information analysis module including: a fully connected layer, a normalized network layer, and an activation function layer; The first two-dimensional convolutional layer is connected to the first feature processing module, and the second two-dimensional convolutional layer is connected to the last feature processing module. The output information of the object to be processed by the first two-dimensional convolutional layer is used as the original input information of the first feature processing module; The at least one feature processing module is connected sequentially, and the output information of the previous feature processing module is used as the original input information of the next adjacent feature processing module. The output information of the last feature processing module, after being processed by the second two-dimensional convolutional layer, is used as the object feature vector of the object to be processed. The output information of the object description information and the enhancement method encoding information after being processed by the at least one description information analysis module is used as the metadata feature vector of the object to be processed.

7. The method according to claim 6, characterized in that, Each feature processing module includes at least one sequentially connected sub-feature processing module, and an average pooling layer connected to the last said sub-feature processing module; The input information of the first sub-feature processing module in each feature processing module is the original input information of the feature processing module; The fused information formed by fusing the output information of the previous sub-feature processing module with the original input information is used as the input information of the next adjacent sub-feature module; The output information of the last sub-feature processing module is processed by the average pooling layer to output average pooling information. The fused information formed by fusing the average pooling information and the secondary processing information is used as the output information of the feature processing module. The secondary processing information is the information obtained by performing a two-dimensional convolution process and an average pooling process on the original input information of the feature processing module.

8. The method according to claim 5, characterized in that, When the object to be processed is an image, the step of concatenating the object feature vector and the metadata feature vector through the evaluation prediction sub-model in the object enhancement evaluation model to obtain a concatenated feature vector, and outputting the object enhancement evaluation result through dimensionality reduction processing of the concatenated feature vector, includes: The object feature vector of the object to be processed is tiled to obtain the object tiled feature vector; The object's flat feature vector is concatenated with the metadata feature vector to form a first feature concatenation vector; The first feature concatenation vector is subjected to fully connected processing, and the fully connected feature vector obtained after processing is subjected to one-dimensional dimensionality reduction and normalization processing. The value obtained after processing is determined as the object enhancement evaluation result of the image.

9. The method according to claim 5, characterized in that, When the object to be processed is a video to be processed, at least one video frame extracted from the video to be processed is respectively input into the first sub-analysis model; the image feature vectors output by the first sub-analysis model relative to the at least one video frame respectively constitute the object feature vector; Accordingly, the step of concatenating the object feature vector and the metadata feature vector through the evaluation prediction sub-model in the object augmentation evaluation model to obtain a concatenated feature vector, and outputting the object augmentation evaluation result through dimensionality reduction processing of the concatenated feature vector, includes: The image feature vectors that constitute the object feature vector are spliced ​​together to form the image splicing feature vector. The image splicing feature vector is then processed in the temporal domain through a three-dimensional convolutional layer to obtain the processed temporal feature vector. The time-series feature vector is flattened, and the flattened feature vector is concatenated with the metadata feature vector to form a second feature concatenation vector. The second feature concatenation vector is subjected to fully connected processing, and the fully connected feature vector obtained after processing is subjected to one-dimensional dimensionality reduction and normalization processing. The value obtained after processing is determined as the object enhancement evaluation result of the video.

10. The method according to any one of claims 1-9, characterized in that, Also includes: The enhanced object, after being processed by the aforementioned enhancement method, is taken as the target object and recorded in the object database; If the object enhancement evaluation result does not meet the preset conditions, the object to be processed is directly used as the target object and recorded in the object database; The preset condition is that the probability of object enhancement evaluation of the characterizing object enhancement evaluation result is less than the set probability threshold.

11. An object management device, characterized in that, include: The first determination module is used to determine the object to be processed that matches the acquired object identifier; The second determining module is used to determine the enhanced processing method of the object to be processed and the object description information of the object to be processed. The third determining module is used to determine the enhancement method encoding information of the enhancement processing method according to the enhancement processing method; The result evaluation module is used to predict the object enhancement evaluation result of the object to be processed based on the object to be processed, the object description information, and the enhancement method encoding information. The object enhancement evaluation result is used to characterize the effect of enhancing the object to be processed based on the enhancement processing method. The object management module is used to perform enhancement processing on the object to be processed based on the enhancement processing method if the object enhancement evaluation result meets the preset conditions.

12. A computer device, characterized in that, The computer device includes: One or more processors; a storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the object management method as described in any one of claims 1-10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the object management method as described in any one of claims 1-10.

14. A computer program product comprising a computer program that, when executed by a processor, implements the object management method according to any one of claims 1-10.

Images