CN116127111A - Image search method, device, electronic device and computer-readable storage medium - Google Patents

Abstract

The disclosure provides a picture searching method, a picture searching device, electronic equipment and a computer readable storage medium, and relates to the technical field of artificial intelligence, in particular to the field of image processing and intelligent searching. The specific implementation scheme is as follows: acquiring inquiry information; determining at least one picture style based on the query information; generating at least one input picture based on the at least one picture style; at least one target picture is searched based on the at least one input picture. According to the technical scheme, the expression effect of the search requirement of the user is improved, the expression diversity is enriched, so that the picture search result meeting the user requirement is obtained, and the search efficiency of the user is improved.

Description

Image search method, device, electronic device and computer-readable storage medium

technical field

The present disclosure relates to the technical field of artificial intelligence, in particular to the field of image processing and the field of intelligent search.

Background technique

Image search technology has been widely used on the Internet. The basic process is: the user expresses the search intention by inputting query information, and the search engine searches out relevant images based on the query information input by the user; on the page provided by the search engine, the user passes Browse and obtain pictures by means of page turning and screening.

In order to improve the effect of image search, search engines have evolved from supporting text search to images based on image content (image search), and multimodal search based on the combination of text and images as query information. However, the existing image search results still cannot well satisfy user needs.

Contents of the invention

The present disclosure provides a picture search method, device, electronic equipment and computer-readable storage medium.

According to an aspect of the present disclosure, a method for image search is provided, including:

Obtain query information;

Determine at least one picture style based on the query information;

generating at least one input image based on at least one image style;

Based on at least one input picture, at least one target picture is searched.

According to another aspect of the present disclosure, an image search device is provided, including:

A query acquisition module, configured to acquire query information;

A style determining module, configured to determine at least one picture style based on the query information;

An input determination module, configured to generate at least one input image based on at least one image style;

An image search module, configured to search for at least one target image based on at least one input image.

According to another aspect of the present disclosure, an electronic device is provided, including:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can execute any method in the embodiments of the present disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method according to any one of the embodiments of the present disclosure.

According to another aspect of the present disclosure, there is provided a computer program product, including a computer program, which, when executed by a processor, implements the method according to any one of the embodiments of the present disclosure.

According to the technical solutions of the embodiments of the present disclosure, after the query information is acquired, at least one picture style is determined based on the query information, and at least one input picture is generated based on the picture style, and the at least one input picture is used to search for a target picture. Since the image style determined based on the query information is used to generate the input image, the expression effect of the user's search needs is improved and the expression diversity is enriched, thereby helping to obtain image search results that meet the user's needs and improving the user's search efficiency.

It should be understood that what is described in this section is not intended to identify key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will be readily understood through the following description.

Description of drawings

The accompanying drawings are used to better understand the present solution, and do not constitute a limitation to the present disclosure. in:

FIG. 1 is a schematic diagram of an exemplary application scenario of a picture search method provided by an embodiment of the present disclosure;

FIG. 2 is a schematic flowchart of a picture search method provided by an embodiment of the present disclosure;

Fig. 3 is a schematic diagram of an optimization target in an image search method provided by an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of an application example of a picture search method according to an embodiment of the present disclosure;

Fig. 5 is a schematic block diagram of an image search device provided by an embodiment of the present disclosure;

Fig. 6 is a schematic block diagram of an image search device provided by another embodiment of the present disclosure;

FIG. 7 is a block diagram of an electronic device for implementing a picture search method according to an embodiment of the present disclosure.

Detailed ways

Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and they should be regarded as exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

In order to facilitate the understanding of the image search method in the embodiment of the present disclosure, the application scenarios of the method are firstly introduced below. Fig. 1 shows a schematic diagram of an exemplary application scenario of a picture search method provided by an embodiment of the present disclosure. The image search method provided by the embodiment of the present disclosure may be applied to the image search device 12 . The image search apparatus 12 may be deployed in an electronic device, such as a stand-alone or multi-computer terminal, a server or other processing devices. As shown in FIG. 1 , the image search apparatus may interact with user equipment 11 to receive a search request from the user equipment, and the request carries query information (query). The image search device 12 searches for a target image corresponding to the query information based on the query information.

According to the picture search method of the embodiment of the present disclosure, in the process of searching for a target picture, an input picture may be generated based on query information, so as to use the input picture to search for a target picture. As an example and not a limitation, as shown in Figure 1, the picture search device 12 can interact with an AIGC (AI Generated Content, artificial intelligence generated content) server, and generate an input picture through the AIGC server, thereby improving the expression effect of the user's search needs and Rich expression diversity is conducive to searching for target pictures that meet the user's search needs.

Fig. 2 shows a schematic flowchart of a picture search method provided by an embodiment of the present disclosure. As shown in Figure 2, the method may include:

Step S210, obtaining query information;

Step S220, based on the query information, determine at least one picture style;

Step S230, generating at least one input picture based on at least one picture style;

Step S240, based on at least one input picture, search for at least one target picture.

In the embodiments of the present disclosure, query information (query) is information used to characterize a user's search requirement or search intention. Exemplarily, the query information may include text information, such as keywords, query words, and the like. The query information may also include picture information, or both text information and picture information.

Exemplarily, in this embodiment of the present disclosure, at least one picture style determined based on the query information may include animation, sketch, art, abstraction, Chinese style, sense of science and technology, and the like. The picture style is used to generate at least one input picture. Exemplarily, at least one input picture may be generated based on at least one picture style and semantic information in the query information, so that the input picture may be related to the semantic information in the query information.

In the embodiment of the present disclosure, the target picture may be a picture corresponding to the query information searched in the picture library. Exemplarily, the characteristic information of the input picture may be matched with the characteristic information of each picture in the picture library, and the matched picture may be used as the target picture.

In the above image search method, by generating the input image, the expression diversity of the user's search needs can be enriched. At the same time, since the image style of the input image is determined based on the query information, it is beneficial to make the generated image consistent with the user's The search requirements are matched, and the expression effect of the user's search requirements is improved. Based on this, the above image search method is conducive to obtaining image search results that meet user needs and improving user search efficiency.

In an exemplary embodiment, step S220, determining at least one picture style based on the query information may include: searching for at least one reference picture based on the text information in the query information; based on each reference picture in the at least one reference picture The style tags are aggregated to obtain at least one image style.

In this embodiment, the query information includes text information. For query information in the form of text, at least one picture style corresponding to the query information can be obtained by searching for pictures related to the text information in the picture database as reference pictures, and then by aggregating the style tags of the reference pictures.

Exemplarily, the at least one reference picture may be grouped according to the style tag, so as to obtain at least one group of reference pictures respectively corresponding to the at least one style tag. According to the number of pictures in each group of reference pictures, the picture style corresponding to the query information is determined. For example, the style tags corresponding to a group of reference pictures with the largest number of pictures are used as the picture style corresponding to the query information.

By adopting the above embodiments, the query information in the form of text can be converted into a picture style related to the picture, which is beneficial to generate an input picture by using the query information in the form of text, and enriches the expression diversity of the user's search needs.

In an exemplary embodiment, step S220, determining at least one picture style based on the query information includes: performing style recognition based on picture information in the query information to obtain at least one picture style.

In this embodiment, the query information includes picture information. For the query information in the form of a picture, the style of the picture can be obtained by performing style recognition on the picture information in the query information, that is, referring to the visual understanding of the query information.

According to this embodiment, the style recognition based on the picture information in the query information can make the obtained picture style have a high degree of matching with the user's search requirements, thereby improving the accuracy of the picture style and correspondingly improving the input picture's effect on the user. The expression effect of search requirements.

In some examples, the query information may include text information and picture information, and the style recognition implementation may be combined with the aforementioned implementation of obtaining picture style based on reference picture aggregation. For example, when the query information includes text information and picture information, you can search for reference pictures based on the text information, and get a part of the picture style based on the style tags of the reference pictures; compare and identify the style based on the picture information, and get another part of the picture style .

Optionally, at least one picture style determined based on the query information may also be displayed to the user, so that the picture style used to generate the input picture is determined from the aggregated picture styles according to the selection instruction input by the user.

Optionally, in addition to the picture style determined based on the query information or the picture style determined according to the user's selection instruction, one or more preset picture styles can also be combined to obtain the picture style used to generate the input picture.

In an exemplary embodiment, step S230, generating at least one input picture based on at least one picture style includes: for each picture style in the at least one picture style, using AIGC (Artificial Intelligence Generated Content) to generate An input image that is related to the semantic information in the query information and has the image style.

Exemplarily, an AIGC server may be used to generate an input picture. AIGC can automatically generate more possible pictures by computer algorithms at the second level, and AIGC can be configured in various styles and forms (fusion mode, splicing mode, etc.), so that it can generate diverse input related to style and semantics picture.

According to this embodiment, the AIGC method is used to expand the diversity of input pictures, and at the same time, the AIGC method can generate input pictures with a high matching degree based on the semantic information in the query information and the predetermined style, so it can also improve the search efficiency of input pictures for users. Accuracy of expression of requirements.

Optionally, after at least one input picture is generated based on the style of the picture, the input picture may also be displayed to the user, so that according to a selection instruction input by the user, the input picture used for searching the target picture is determined among the generated input pictures.

In an exemplary embodiment, step S240, searching for at least one target picture based on at least one input picture includes: determining a feature vector of each input picture in at least one input picture; based on the feature vector of each input picture and The feature vector of each picture in the picture library, search for at least one target picture in the picture library.

With this embodiment, the picture is represented by a feature vector, so that the feature vector can be used to perform refined expression and calculation of information of each picture. Based on this, the efficiency of searching for target pictures in the picture library can be improved, and it is beneficial to control the cost of computing resources.

In an exemplary embodiment, based on the feature vector of each input picture and the feature vector of each picture in the picture library, searching for at least one target picture in the picture library includes: based on the feature vector of each input picture and The distance between the feature vectors of each picture in the picture library, search for at least one target picture that meets the optimization goal in the picture library; wherein, the optimization goal includes: each target picture in each input picture and at least one target picture The sum of the distances between them is the smallest.

In order to illustrate the optimization goal in this embodiment more clearly, FIG. 3 shows a schematic diagram of the optimization goal. As shown in FIG. 3 , assuming that a plurality of input pictures are generated based on at least one picture style, and after selection by the user, input picture 1 and input picture 2 are determined to be used for searching target pictures. The eigenvectors of each picture in the library are the eigenvectors 1 and 2 corresponding to the input pictures 1 and 2 shown in FIG. 3 , and the eigenvectors A to E corresponding to the pictures A to E in the picture library. If the number of target pictures is set to 3 in advance, then the searched 3 target pictures need to satisfy the optimization objectives of the distances between the feature vectors of these 3 target pictures and feature vector 1 and the feature vectors of these 3 target pictures The sum of the distances from the feature vector 2 is the smallest, that is, the sum of the distances between the input image and the target image is minimized. In other words, among the sum of distances corresponding to any combination of three pictures in the picture library, the sum of distances corresponding to the above three target pictures is the smallest. Taking picture A, picture D and picture E as examples, the sum of the distances is the distance between feature vector 1 and feature vector A, the distance between feature vector 1 and feature vector D, the distance between feature vector 1 and feature vector E, the distance between feature vector 2 and The sum of the distance of eigenvector A, the distance of eigenvector 2 from eigenvector D, and the distance of eigenvector 2 from eigenvector E.

In the traditional way of inputting pictures based on vectors, only one input feature vector is searched for the closest vector among the feature vectors of each picture in the picture library at a time. However, according to the above-mentioned embodiment, in a vector search process, the sum is calculated based on the respective distances corresponding to at least one input picture, and the optimization goal is to minimize the distance between the entire input picture and the target picture as a whole, so that the vectors that meet the quantity requirements can be obtained through one search. Target images, improve search speed and reduce computing resource overhead.

In an exemplary embodiment, the image search method may further include: displaying at least one style tag corresponding to at least one input image; in response to receiving a selection instruction for the first style tag in the at least one style tag, according to each The distance between the feature vectors of each target picture and the feature vector of the input picture corresponding to the first style label is used to sort and display at least one target picture.

For example, in the picture search result page, at least one target picture is firstly displayed according to the default sorting, and a style filter box is provided at the same time, and the style labels corresponding to each input picture used to search for the target picture are displayed in the style filter box, and the user's style selection is received. Label selection command. If the user selects the first style label, the input picture corresponding to the first style label is determined, and the target pictures are sorted and displayed according to the distance between the feature vectors of each target picture and the feature vector of the input picture.

In this embodiment, the target pictures are sorted and displayed according to the style tags selected by the user, so that the display effect of the target pictures can be optimized, which is beneficial for the user to quickly obtain pictures that meet the needs and improve the search efficiency.

In order to understand the above image search method more clearly, a specific application example is provided below. Fig. 4 is a schematic diagram of an application example of the image search method of the embodiment of the present disclosure. As shown in Figure 4, the method includes the following steps:

Step S401, the user inputs query information (query), and the query information is text information.

Step S402, AIGC generates input pictures, including input picture 1 to input picture N, where N is an integer greater than 2. Specifically, the input image can be generated through the following interactive process:

When you first initiate a search, traditional image search results are presented by default.

On the search result page, provide an entrance for AIGC to generate pictures, prompting users to generate pictures through AI, and initiate the search again.

If the user clicks on the entry, a floating layer will pop up. The floating layer displays the styles aggregated by the search engine based on the search results of the current query, and there are also some default built-in styles. Each style has a text description and matches an image representing the style.

The user selects one or more styles, and clicks the button to initiate AIGC image generation.

The AIGC server generates a good picture and displays the generated picture in the floating layer.

Step S403, the user selects one or more input pictures.

Specifically, the user selects one or more pictures in the floating layer as input pictures through check boxes, and clicks to initiate a related search. Then, turn off the overlay.

Step S404, based on the input picture selected by the user, perform synchronous parallel search to obtain the target picture set.

Specifically, if the user selects N pictures as input pictures, the retrieval features are extracted in parallel for the N pictures to obtain corresponding feature vectors. Then use the N feature vectors as a combination to initiate a vector retrieval. The vector retrieval engine completes the retrieval of the vector combination at the same time (to ensure that each feature vector can search for a certain result, and the total amount is controlled, and the retrieval speed will not be particularly slow, and will not bring large computing resource overhead), And return the K results with the smallest vector distance.

For example, define the following information:

N: the total number of vectors input by query;

K: the total number of vectors output by the query;

Q[i]: i ranges from 1 to N, representing the input i-th vector;

R[j]:j ranges from 1 to K, representing the vector representation of the jth result of the query output;

Dist[Q[i],R[j]]: represents the distance between two vectors Q[i] and R[j], which can generally be the Euclidean distance or cosine distance of two vectors;

best_match[i]: best_match[i] is an array subscript, ranging from 1 to K, representing the query result matched by the i-th input vector, so that the optimization goal is met. For two x, y, and x! =y, best_match[i] cannot be equal to best_match[j].

Then the optimization objective can be expressed as: min{sum(Dist[Q[i],R[best_match[j]])}i from 1 to N, j from 1 to N, that is, the sum of all distances is the smallest. That is, find K results from the image library such that the sum of their closest distances to the query input vector is the smallest.

After the above steps, K results (target pictures) can be output. On the image search result page, a group of style tags (tags) to be screened may be provided. There are N style tags in total, and each style tag corresponds to a picture generated by AIGC selected by the user. All the results are displayed by default on the image search result page. When the user clicks on a style tag, the sorting of the K images on the search result page will be adjusted in ascending order according to the distance from the input image corresponding to the style tag (that is, the distance from near to far).

It can be seen that the above application example uses pictures generated by AIGC, which improves the efficiency of users' expression of search needs, reduces the number of interaction steps, and only requires users to click and select a few pictures to achieve diversified input expressions. In addition, by adjusting the optimization goal of the search, it is more suitable for scenarios where the search requirements are difficult to express clearly and uniquely, and more diverse search results can be provided.

According to an embodiment of the present disclosure, the present disclosure also provides an image search device. Fig. 5 shows a schematic block diagram of an image search device provided by an embodiment of the present disclosure. As shown in Figure 5, the device may include:

Query acquisition module 510, configured to acquire query information;

A style determination module 520, configured to determine at least one picture style based on the query information;

An input determination module 530, configured to generate at least one input picture based on at least one picture style;

The image search module 540 is configured to search for at least one target image based on at least one input image.

Fig. 6 shows a schematic block diagram of an image search device provided by another embodiment of the present disclosure. As shown in Figure 6, the picture search device includes a query acquisition module 610, a style determination module 620, an input determination module 630 and a picture search module 640, wherein the style determination module 620 may include:

A reference picture search unit 621, configured to search for at least one reference picture based on the text information in the query information;

The style aggregation unit 622 is configured to perform aggregation based on the style tags of each reference picture in the at least one reference picture to obtain at least one picture style.

Exemplarily, as shown in Figure 6, the style determination module may also include:

The style recognition unit 623 is configured to perform style recognition based on the picture information in the query information to obtain at least one picture style.

Optionally, the input determination module 630 in the image search device is specifically used for:

For each picture style in the at least one picture style, an input picture related to the semantic information in the query information and having the picture style is generated by using artificial intelligence to generate content.

Optionally, as shown in Figure 6, the image search module 640 in the image search device may include:

A vector determining unit 641, configured to determine a feature vector of each input picture in at least one input picture;

The target picture searching unit 642 is configured to search for at least one target picture in the picture library based on the feature vector of each input picture and the feature vector of each picture in the picture library.

Optionally, the target picture search unit 642 is specifically configured to:

Based on the distance between the feature vector of each input picture and the feature vector of each picture in the picture library, at least one target picture that meets the optimization goal is searched in the picture library;

Wherein, the optimization objective includes: the sum of the distances between each input picture and each target picture in at least one target picture is the smallest.

Optionally, as shown in Figure 6, the image search device may also include:

A label display module 650, configured to display at least one style label corresponding to at least one input picture;

A sorting and displaying module 660, configured to respond to receiving a selection instruction for a first style label in at least one style label, according to the relationship between the feature vector of each target picture and the feature vector of the input picture corresponding to the first style label distance, at least one target image is sorted and displayed.

For descriptions of specific functions and examples of modules and sub-modules of the apparatus in the embodiments of the present disclosure, reference may be made to relevant descriptions of corresponding steps in the foregoing method embodiments, and details are not repeated here.

In the technical solution of the present disclosure, the acquisition, storage and application of the user's personal information involved are in compliance with relevant laws and regulations, and do not violate public order and good customs.

According to the embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

FIG. 7 shows a schematic block diagram of an example electronic device 700 that may be used to implement embodiments of the present disclosure. Electronic device is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are by way of example only, and are not intended to limit implementations of the disclosure described and/or claimed herein.

As shown in FIG. 7, the device 700 includes a computing unit 701 that can execute according to a computer program stored in a read-only memory (ROM) 702 or loaded from a storage unit 708 into a random-access memory (RAM) 703. Various appropriate actions and treatments. In the RAM 703, various programs and data necessary for the operation of the device 700 can also be stored. The computing unit 701, ROM 702, and RAM 703 are connected to each other through a bus 704. An input/output (I/O) interface 705 is also connected to the bus 704 .

Multiple components in the device 700 are connected to the I/O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, an optical disk, etc. ; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit 709 allows the device 700 to exchange information/data with other devices over a computer network such as the Internet and/or various telecommunication networks.

The computing unit 701 may be various general-purpose and/or special-purpose processing components having processing and computing capabilities. Some examples of computing units 701 include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any suitable processor, controller, microcontroller, etc. The calculation unit 701 executes various methods and processes described above, such as image search methods. For example, in some embodiments, the image search method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 708 . In some embodiments, part or all of the computer program may be loaded and/or installed on the device 700 via the ROM 702 and/or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the image search method described above can be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to execute the image search method in any other suitable manner (for example, by means of firmware).

Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips Implemented in a system of systems (SOC), load programmable logic device (CPLD), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor Can be special-purpose or general-purpose programmable processor, can receive data and instruction from storage system, at least one input device, and at least one output device, and transmit data and instruction to this storage system, this at least one input device, and this at least one output device an output device.

Program codes for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special purpose computer, or other programmable data processing devices, so that the program codes, when executed by the processor or controller, make the functions/functions specified in the flow diagrams and/or block diagrams Action is implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present disclosure, a machine-readable medium may 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 may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, 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 would include one or more wire-based electrical connections, portable computer discs, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.

To provide for interaction with the user, the systems and techniques described herein can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user. ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and can be in any form (including Acoustic input, speech input, or tactile input) to receive input from the user.

The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., as a a user computer having a graphical user interface or web browser through which a user can interact with embodiments of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system. The components of the system can be interconnected by any form or medium of digital data communication, eg, a communication network. Examples of communication networks include: Local Area Network (LAN), Wide Area Network (WAN) and the Internet.

A computer system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, a server of a distributed system, or a server combined with a blockchain.

It should be understood that steps may be reordered, added or deleted using the various forms of flow shown above. For example, each step described in the present disclosure may be executed in parallel, sequentially, or in a different order, as long as the desired result of the technical solution disclosed in the present disclosure can be achieved, no limitation is imposed herein.

The specific implementation manners described above do not limit the protection scope of the present disclosure. It should be apparent to those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made depending on design requirements and other factors. Any modifications, equivalent replacements and improvements made within the principles of the present disclosure shall be included within the protection scope of the present disclosure.

Claims (17)

1. A method for image search, comprising: Obtain query information; determining at least one picture style based on the query information; generating at least one input picture based on the at least one picture style; Based on the at least one input picture, at least one target picture is searched. 2. The method according to claim 1, wherein said determining at least one picture style based on said query information comprises: searching for at least one reference picture based on text information in the query information; Aggregating based on the style tags of each picture in the at least one reference picture to obtain at least one picture style. 3. The method according to claim 1, wherein said determining at least one picture style based on said query information comprises: Performing style recognition based on the picture information in the query information to obtain at least one picture style. 4. The method according to any one of claims 1-3, wherein said generating at least one input picture based on said at least one picture style comprises: For each picture style in the at least one picture style, an input picture related to the semantic information in the query information and having the picture style is generated by using artificial intelligence to generate content. 5. The method according to any one of claims 1-4, wherein said searching for at least one target picture based on said at least one input picture comprises: determining a feature vector for each of the at least one input picture; Searching for the at least one target picture in the picture library based on the feature vector of each input picture and the feature vector of each picture in the picture library. 6. The method according to claim 5, wherein said at least one target picture is searched in the picture library based on the feature vector of each input picture and the feature vector of each picture in the picture library ,include: Based on the distance between the feature vector of each input picture and the feature vector of each picture in the picture library, searching for at least one target picture that meets the optimization goal in the picture library; Wherein, the optimization goal includes: the sum of the distances between each input picture and each target picture in the at least one target picture is the smallest. 7. The method of any one of claims 1-6, further comprising: displaying at least one style label corresponding to the at least one input picture; In response to receiving a selection instruction for the first style label in the at least one style label, according to the distance between the feature vector of each target picture and the feature vector of the input picture corresponding to the first style label, for The at least one target picture is sorted and displayed. 8. A picture search device, comprising: A query acquisition module, configured to acquire query information; A style determining module, configured to determine at least one picture style based on the query information; An input determination module, configured to generate at least one input picture based on the at least one picture style; A picture search module, configured to search for at least one target picture based on the at least one input picture. 9. The apparatus of claim 8, wherein the style determination module comprises: a reference picture search unit, configured to search for at least one reference picture based on the text information in the query information; A style aggregation unit, configured to perform aggregation based on the style tags of each picture in the at least one reference picture, to obtain at least one picture style. 10. The apparatus of claim 8, wherein the style determination module comprises: A style recognition unit, configured to perform style recognition based on the picture information in the query information to obtain at least one picture style. 11. The device according to any one of claims 8-10, wherein the input determination module is configured to: For each picture style in the at least one picture style, an input picture related to the semantic information in the query information and having the picture style is generated by using artificial intelligence to generate content. 12. The device according to any one of claims 8-11, wherein the picture search module comprises: a vector determination unit configured to determine a feature vector of each of the at least one input picture; A target picture searching unit, configured to search for the at least one target picture in the picture library based on the feature vector of each input picture and the feature vector of each picture in the picture library. 13. The device according to claim 12, wherein the target picture search unit is used for: Based on the distance between the feature vector of each input picture and the feature vector of each picture in the picture library, searching for at least one target picture that meets the optimization goal in the picture library; Wherein, the optimization goal includes: the sum of the distances between each input picture and each target picture in the at least one target picture is the smallest. 14. The apparatus according to any one of claims 8-13, further comprising: a tag display module, configured to display at least one style tag corresponding to the at least one input image; A sorting display module, configured to, in response to receiving a selection instruction for the first style label in the at least one style label, according to the feature vector of each target picture and the feature vector of the input picture corresponding to the first style label and sorting and displaying the at least one target picture. 15. An electronic device comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions are executed by the at least one processor, so that the at least one processor can perform any one of claims 1-7. Methods. 16. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method according to any one of claims 1-7. 17. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-7.

Images