Picture retrieval method, device and system and medium

Abstract

The invention discloses a picture retrieval method, device and system and a medium, and the method comprises the steps: receiving retrieval keywords which comprise Chinese characters and / or phonetic alphabets; matching the retrieval keyword with picture content stored in a search server, and determining an alternative picture matched with the retrieval keyword; performing retrieval recommendation according to the alternative pictures and a picture library neural network constructed based on retrieval records, and determining corresponding recommended pictures; and displaying the alternative pictures and the recommended pictures, and determining a target retrieval picture in the alternative pictures and the recommended pictures according to a selection operation of a user. According to the method, the picture content is matched with the retrieval keyword, and intelligent picture retrieval recommendation is performed based on the retrieved alternative pictures and the pre-constructed picture library neural network, so that a user can efficiently and quickly retrieve the most relevant picture by directly inputting the keyword, and the picture retrieval efficiency and accuracy are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and in particular to an image retrieval method, device, system and medium. Background Art

[0002] Image retrieval technology has many applications in the financial field, especially in banking scenarios. For example, image retrieval technology can be used to recommend similar or related financial products based on the pictures of financial products viewed by users on the bank's website or app; or image retrieval can be used during internal training to quickly locate relevant pictures in training materials to improve training efficiency.

[0003] However, traditional image retrieval only stores images on the server. If users want to quickly find images, they need to record the address where the images are stored, or manually add notes when storing images, which affects the efficiency of image upload and storage. It is difficult to efficiently locate the images to be retrieved through the image content, which increases the workload. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the object of the present invention is to provide an image retrieval method, device, system and medium that can be applied to financial technology or other related fields, the main purpose of which is to improve the efficiency and accuracy of image retrieval.

[0005] The technical solution of the present invention is as follows:

[0006] A first aspect of the present invention provides an image retrieval method, comprising:

[0007] Receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters;

[0008] Matching the search keyword with the picture content stored in the search server to determine candidate pictures that match the search keyword;

[0009] Performing retrieval recommendation based on the candidate images and the image library neural network constructed based on the retrieval records to determine the corresponding recommended images;

[0010] The candidate images and recommended images are displayed, and a target retrieval image is determined from the candidate images and recommended images according to a user's selection operation.

[0011] In one embodiment, before receiving the search keyword, the method further includes:

[0012] Get the picture to be stored;

[0013] Performing image and text recognition on the image to be stored to obtain corresponding image content;

[0014] Uploading the image to be stored to the distributed file system, and uploading the image content and storage address to the search server;

[0015] A node for the image to be stored is created in the image library neural network, and the image library neural network is used to maintain a retrieval relationship between a retrieval user node and an image node according to a retrieval record.

[0016] In one embodiment, after uploading the image to be stored to the distributed file system and uploading the image content and storage address to the search server, the method further includes:

[0017] The search server is configured to support Chinese characters and / or pinyin letters as the retrieval data type.

[0018] In one embodiment, when the search keyword includes Chinese characters and pinyin letters, matching the search keyword with the image content stored in the search server to determine the candidate image matching the search keyword includes:

[0019] Matching the Chinese characters in the search keyword with the image content stored in the search server to obtain a first search result;

[0020] Convert the Chinese characters and pinyin letters in the search keyword into pinyin and then match them with the picture content stored in the search server to obtain a second search result;

[0021] The first search result and the second search result are combined to determine a combined candidate image.

[0022] In one embodiment, the searching and recommending based on the candidate images and the image library neural network constructed based on the search records to determine the corresponding recommended images specifically includes:

[0023] The gallery neural network is called according to the candidate images, and a number of images with the highest number of clicks are obtained from the gallery neural network according to a three-degree relationship algorithm as the recommended images.

[0024] In one embodiment, after displaying the candidate images and the recommended images and determining the target search image from the candidate images and the recommended images according to the user's selection operation, the method further comprises:

[0025] Update the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network.

[0026] In one embodiment, updating the search relationship between the current user node and the node of the target search image in the gallery neural network specifically includes:

[0027] The number of clicks on the target retrieval image by the current user in the gallery neural network is increased by 1.

[0028] A second aspect of the present invention provides an image retrieval device, comprising:

[0029] A receiving module, used for receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters;

[0030] A matching module, used to match the search keyword with the image content stored in the search server, and determine the candidate images that match the search keyword;

[0031] A recommendation module, used to perform retrieval and recommendation based on the candidate images and a picture library neural network constructed based on the retrieval records, and determine corresponding recommended images;

[0032] The display determination module is used to display the candidate images and recommended images, and determine the target retrieval image from the candidate images and recommended images according to the user's selection operation.

[0033] A third aspect of the present invention provides an image retrieval system, the system comprising at least one processor; and

[0034] a memory communicatively connected to the at least one processor; wherein,

[0035] The memory stores instructions that can be executed 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 the above-mentioned image retrieval method.

[0036] A fourth aspect of the present invention provides a non-volatile computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the one or more processors can execute the above-mentioned image retrieval method.

[0037] Beneficial effects: The present invention discloses a method, device, system and medium for image retrieval. Compared with the prior art, the embodiment of the present invention receives a retrieval keyword, wherein the retrieval keyword includes Chinese characters and / or pinyin letters; matches the retrieval keyword with the image content stored in the search server to determine the candidate images that match the retrieval keyword; performs retrieval recommendation based on the candidate images and a gallery neural network constructed based on the retrieval records to determine the corresponding recommended images; displays the candidate images and recommended images, and determines the target retrieval image from the candidate images and recommended images according to the user's selection operation. By matching the image content with the retrieval keyword, and performing intelligent image retrieval recommendation based on the retrieved candidate images and the pre-constructed gallery neural network, the user can directly and efficiently retrieve the most relevant images by entering keywords, thereby effectively improving the efficiency and accuracy of image retrieval. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0039] Figure 1 A flow chart of a picture retrieval method provided by an embodiment of the present invention;

[0040] Figure 2 A flow chart of image storage in the image retrieval method provided in an embodiment of the present invention;

[0041] Figure 3 A flow chart of step S102 in the image retrieval method provided in an embodiment of the present invention;

[0042] Figure 4 A schematic diagram of the functional modules of the image retrieval device provided by an embodiment of the present invention;

[0043] Figure 5 A schematic diagram of the hardware structure of the image retrieval system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. The embodiments of the present invention are described below in conjunction with the accompanying drawings.

[0045] Image retrieval technology has many applications in the financial field, especially in banking scenarios. For example, image retrieval technology can be used to recommend similar or related financial products based on the pictures of financial products viewed by users on the bank's website or app; or image retrieval can be used during internal training to quickly locate relevant pictures in training materials to improve training efficiency.

[0046] However, traditional image retrieval only stores images on the server. If users want to quickly find images, they need to record the address where the images are stored, or manually add notes when storing images, which affects the efficiency of image upload and storage. It is difficult to efficiently locate the images to be retrieved through the image content, which increases the workload.

[0047] In order to solve the above problems, the present invention proposes a picture retrieval method, please refer to Figure 1 , Figure 1 The flowchart of an embodiment of the image retrieval method provided by the present invention. The system architecture of the image retrieval method provided by this embodiment may include a first terminal device, a second terminal device, a third terminal device, a network and a server. The network is used to provide a medium for a communication link between the first terminal device, the second terminal device, the third terminal device and the server. The network may include various connection types, such as wired and / or wireless communication links, etc.

[0048] The user may use the first terminal device, the second terminal device, and the third terminal device to interact with the server through the network to receive or send messages, etc. Various communication client applications may be installed on the first terminal device, the second terminal device, and the third terminal device, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only as examples).

[0049] The first terminal device, the second terminal device, and the third terminal device may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0050] The server may be a server that provides various services, such as a background management server that provides support for the content browsed by the user using the first terminal device, the second terminal device, and the third terminal device (for example only). The background management server may analyze and process the received data such as the user request, and feed back the processing results (such as web pages, information, or data obtained or generated according to the user request) to the terminal device. The server may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server for a distributed system, or a server combined with a blockchain.

[0051] It should be noted that the image retrieval method provided in the embodiment of the present invention can generally be executed by the first terminal device, the second terminal device or the third terminal device. Correspondingly, the image retrieval device provided in the embodiment of the present invention can also be set in the first terminal device, the second terminal device or the third terminal device.

[0052] Alternatively, the image retrieval method provided by the embodiment of the present invention may also be generally executed by a server. Accordingly, the image retrieval device provided by the embodiment of the present invention may generally be arranged in a server. The image retrieval method provided by the embodiment of the present invention may also be executed by a server or a server cluster that is different from the server and can communicate with the first terminal device, the second terminal device, the third terminal device and / or the server. Accordingly, the image retrieval device provided by the embodiment of the present invention may also be arranged in a server or a server cluster that is different from the server and can communicate with the first terminal device, the second terminal device, the third terminal device and / or the server.

[0053] It should be understood that the numbers of the above terminal devices, networks and servers are only illustrative. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0054] like Figure 1 As shown, the image retrieval method provided in this embodiment specifically includes the following steps:

[0055] S101, receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters.

[0056] In this embodiment, when searching for pictures, the user can enter the corresponding search keyword through the front-end interactive interface. The keyword can be Chinese characters, or pinyin letters, or Chinese characters and pinyin letters, thereby supporting multiple forms of picture retrieval. Specifically, the user's search request can be received through a text box or voice input function on the interactive interface. For example, the user can enter the search keyword through a physical keyboard input, a touch keyboard input, voice input, etc., which is not limited in this embodiment. Through multiple input forms, the user can quickly and conveniently enter the search keyword, and support for Chinese characters and pinyin increases the flexibility and accuracy of picture retrieval.

[0057] S102: Match the search keyword with the picture content stored in the search server to determine candidate pictures that match the search keyword.

[0058] After receiving the search keyword, it is matched with the image content stored in the search server, that is, the text content in the image is recognized in advance in the search server through recognition technologies such as OCR, and converted into searchable text data, so that when the user retrieves the image, the search keyword input by the user can be matched with the image content through the search engine algorithm to determine the candidate images that match the search keyword. The matching method of the search keyword and the image content can be specifically implemented by a simple string matching algorithm, or by using natural language processing technology to understand the context and semantics of the keyword for matching. The similarity or correlation between the image content and the key search term can also be calculated, such as Jaccard similarity, cosine similarity and other algorithms, and the matched images are ranked according to the similarity score, and the highest score images are selected as the candidate images. By identifying and storing the image content, the user can realize the image search through automatic keyword matching when searching for the image, which reduces the time for the user to manually search for the image and improves the search efficiency.

[0059] S103: performing search and recommendation based on the candidate images and a picture library neural network constructed based on the search records to determine corresponding recommended images.

[0060] In addition to matching image content based on search keywords, intelligent recommendations are also made through a pre-built gallery neural network. A corresponding user search gallery neural network is formed based on the search records of all users. That is, whenever a user searches for an image, the system records the user's search keywords, selected images, and search time. The gallery neural network is constructed and trained with these data to enable users to further perform search recommendations and recommend related images through matched alternative images when searching for images. The personalized recommendation method based on the gallery neural network can provide images that better meet user needs and enhance user experience.

[0061] In one embodiment, the searching and recommending based on the candidate images and the image library neural network constructed based on the search records to determine the corresponding recommended images specifically includes:

[0062] The gallery neural network is called according to the candidate images, and a number of images with the highest number of clicks are obtained from the gallery neural network according to a three-degree relationship algorithm as the recommended images.

[0063] In this embodiment, when intelligent recommendation is performed based on candidate images retrieved by the user, a library neural network such as the neo4j library is called according to the candidate images, and several images with the highest number of clicks are obtained from the library neural network according to the three-degree relationship algorithm as corresponding recommended images for intelligent recommendation. For example, the current user retrieves image A, and after calling the library neural network, it is analyzed that the current user's colleagues A and B have also retrieved image A and image B, and the number of clicks of image B ranks among the top three of all images that meet the previous conditions. In this case, image B is recommended as a recommended image, thereby performing intelligent and efficient retrieval recommendation through the retrieval relationship between the user and the image, providing the user with more abundant choices for image retrieval.

[0064] S104: Display the candidate images and recommended images, and determine a target search image from the candidate images and recommended images according to a user's selection operation.

[0065] The search results are displayed on the front-end interactive interface, including the matched candidate images and the images recommended by the neural network, providing users with a wider range of choices and increasing the possibility of users retrieving satisfactory images. Preferably, filtering and sorting functions can also be used, such as users can choose to display only candidate images, or the candidate images can be displayed in the first order, etc., to meet the personalized needs of different users.

[0066] The image retrieval method provided in this embodiment reduces the time users spend searching for images by matching image content with search keywords through an automated search process. At the same time, intelligent image retrieval recommendations are made based on the retrieved candidate images and a pre-built image library neural network, so that users can directly and efficiently retrieve the most relevant images quickly and efficiently by entering keywords, thereby effectively improving the efficiency and accuracy of image retrieval.

[0067] In one embodiment, Figure 2 As shown, before receiving the search keyword, the method also includes:

[0068] S201, obtaining a picture to be stored;

[0069] S202, performing image and text recognition on the image to be stored to obtain corresponding image content;

[0070] S203, uploading the image to be stored to a distributed file system, and uploading the image content and storage address to a search server;

[0071] S204: creating a node for the image to be stored in the image library neural network, wherein the image library neural network is used to maintain a retrieval relationship between a retrieval user node and an image node according to a retrieval record.

[0072] In this embodiment, during the image storage stage, the user uploads the image to be stored through the image upload system, and calls the server backend application through the API interface. The backend server application calls the OCR function to perform image and text recognition on the image to be stored, obtain the corresponding image content, and convert the unstructured image data into searchable structured text data to improve the retrieval efficiency; the backend server application uploads the image to be stored to the FastDFS distributed file storage system to ensure the high availability and fault tolerance of the data, and uploads the identified image content and the FastDFS storage address to the elasticsearch search server for subsequent efficient image retrieval. At the same time, when receiving the newly stored image, a node of the image to be stored is also created in the gallery neural network. The gallery neural network is used to maintain the retrieval relationship between the retrieval user node and the image node according to the retrieval record, that is, a node is created for each image in the gallery neural network, and the characteristics and metadata of the image are recorded, and a relationship network between the image and the user is constructed, which is convenient for complex queries and recommendations.

[0073] Preferably, after uploading the image content and storage address to the search server, the search server is also configured with the retrieval data type supported, which is set to support retrieval of Chinese characters and / or pinyin letters, and the detailed data of the image is stored in the MySQL database to ensure the integrity of the image data storage.

[0074] In one embodiment, Figure 3 As shown, when the search keyword includes Chinese characters and pinyin letters, matching the search keyword with the picture content stored in the search server to determine the candidate pictures matching the search keyword includes:

[0075] S301, matching the Chinese characters in the search keyword with the picture content stored in the search server to obtain a first search result;

[0076] S302, converting Chinese characters and pinyin letters in the search keyword into pinyin and matching them with the picture content stored in the search server to obtain a second search result;

[0077] S303: Merge the first search result and the second search result to determine a merged candidate image.

[0078] In this embodiment, when the search keywords input by the user are all Chinese characters or all pinyin letters, they are directly matched and searched with the picture content in the search server, and the candidates are sorted according to factors such as content matching score and the number of searches from large to small. When the search keywords input by the user include both Chinese characters and pinyin letters, the Chinese search keywords input by the user are first matched with the Chinese characters in the picture content, and a string matching algorithm or a text similarity algorithm can be used to obtain the first search result; then the Chinese characters in the search keywords are converted into pinyin, and the pinyin in the search keywords are combined with the pinyin letters in the picture content to obtain the second search result. Since the first search result and the second search result may overlap, the first search result and the second search are merged, and the two search results are merged into a list, and a set operation is used to remove duplicates, and then the merged candidate pictures are determined. Through the dual matching of Chinese characters and pinyin, the user's search intention can be more comprehensively covered, and the accuracy of the search results can be improved. Users may make spelling errors when entering search terms, and pinyin matching can reduce the impact of such errors on the search results. Merging two search results can also increase the number of alternative images, providing users with more choices, better meeting the needs of different users, and providing richer and more accurate search results.

[0079] In one embodiment, after displaying the candidate images and the recommended images and determining the target search image from the candidate images and the recommended images according to the user's selection operation, the method further comprises:

[0080] Update the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network.

[0081] In this embodiment, after the user selects the target retrieval image he needs, the relationship between the user and the image is maintained in the gallery neural network. Specifically, the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network is updated, and the number of clicks on the target retrieval image by the current user in the gallery neural network is increased by 1. That is, after each user search, the alternative images and recommended images obtained according to the search keywords are displayed on the user interface. When the user clicks on a certain image, the system records the click event, finds the current user node and the node of the target retrieval image in the gallery neural network, and increases the number of clicks on the target retrieval image by 1. The number of clicks on the target retrieval image by the user is maintained and updated according to each search result, so that when other users perform image searches subsequently, they can make associated retrieval recommendations based on the retrieval relationship between each user and each image, and recommend the most relevant images to the searching user, thereby improving the relevance and personalization level of the search.

[0082] It should be noted that there is not necessarily a certain order between the above-mentioned steps. A person skilled in the art can understand, based on the description of the embodiments of the present invention, that in different embodiments, the above-mentioned steps may have different execution orders, that is, they may be executed in parallel, may be executed interchangeably, and so on.

[0083] Another embodiment of the present invention provides a picture retrieval device, such as Figure 4 As shown, the device 1 comprises:

[0084] A receiving module 11 is used to receive a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters;

[0085] A matching module 12, used to match the search keyword with the image content stored in the search server, and determine the candidate images matching the search keyword;

[0086] A recommendation module 13 is used to perform search and recommendation based on the candidate images and a picture library neural network constructed based on the search records, and determine a corresponding recommended image;

[0087] The display determination module 14 is used to display the candidate images and recommended images, and determine the target search image from the candidate images and recommended images according to the user's selection operation.

[0088] The module referred to in the present invention refers to a series of computer program instruction segments that can perform specific functions, which are more suitable for describing the image retrieval execution process than programs. For the specific implementation of each module, please refer to the corresponding method embodiment above, which will not be repeated here.

[0089] In one embodiment, the device 1 further comprises:

[0090] A picture acquisition module, used to acquire pictures to be stored;

[0091] An image and text recognition module is used to perform image and text recognition on the image to be stored to obtain the corresponding image content;

[0092] An uploading module, used to upload the image to be stored to a distributed file system, and upload the image content and storage address to a search server;

[0093] The picture library creation module is used to create the node of the picture to be stored in the picture library neural network, and the picture library neural network is used to maintain the retrieval relationship between the retrieval user node and the picture node according to the retrieval record.

[0094] In one embodiment, the device 1 further comprises:

[0095] The configuration module is used to configure the search server to support retrieval data types as Chinese characters and / or pinyin letters.

[0096] In one embodiment, the matching module 12 includes:

[0097] A first matching unit, configured to match the Chinese characters in the search keyword with the image content stored in the search server to obtain a first search result;

[0098] A second matching unit is used to convert the Chinese characters and pinyin letters in the search keyword into pinyin and then match them with the picture content stored in the search server to obtain a second search result;

[0099] A merging unit is used to merge the first search result and the second search result to determine a merged candidate image.

[0100] In one embodiment, the recommendation module 13 is specifically used to:

[0101] The gallery neural network is called according to the candidate images, and a number of images with the highest number of clicks are obtained from the gallery neural network according to a three-degree relationship algorithm as the recommended images.

[0102] In one embodiment, the device 1 further comprises:

[0103] An updating module is used to update the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network.

[0104] In one embodiment, the update module is specifically used to:

[0105] The number of clicks on the target retrieval image by the current user in the gallery neural network is increased by 1.

[0106] Another embodiment of the present invention provides an image retrieval system. Figure 5 As shown, the system 10 includes:

[0107] One or more processors 110 and memory 120, Figure 5 A processor 110 is used as an example for description. The processor 110 and the memory 120 may be connected via a bus or other means. Figure 5 The example of connecting through bus is taken in the following.

[0108] The processor 110 is used to complete various control logics of the system 10, and it can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, a discrete gate or transistor logic, a discrete hardware component, or any combination of these components. In addition, the processor 110 can also be any traditional processor, microprocessor or state machine. The processor 110 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a DSP and / or any other such configuration.

[0109] The memory 120 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions corresponding to the image retrieval method in the embodiment of the present invention. The processor 110 executes various functional applications and data processing of the system 10 by running the non-volatile software programs, instructions and units stored in the memory 120, that is, implements the image retrieval method in the above method embodiment.

[0110] The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function; the data storage area may store data created according to the use of the system 10, etc. In addition, the memory 120 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 120 may optionally include a memory remotely arranged relative to the processor 110, and these remote memories may be connected to the system 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0111] One or more units are stored in the memory 120, and when executed by one or more processors 110, the image retrieval method in any of the above method embodiments is executed, for example, the following steps are executed:

[0112] Receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters;

[0113] Matching the search keyword with the picture content stored in the search server to determine candidate pictures that match the search keyword;

[0114] Performing retrieval recommendation based on the candidate images and the image library neural network constructed based on the retrieval records to determine the corresponding recommended images;

[0115] The candidate images and recommended images are displayed, and a target retrieval image is determined from the candidate images and recommended images according to a user's selection operation.

[0116] In one embodiment, before receiving the search keyword, the method further includes:

[0117] Get the picture to be stored;

[0118] Performing image and text recognition on the image to be stored to obtain corresponding image content;

[0119] Uploading the image to be stored to the distributed file system, and uploading the image content and storage address to the search server;

[0120] A node for the image to be stored is created in the image library neural network, and the image library neural network is used to maintain a retrieval relationship between a retrieval user node and an image node according to a retrieval record.

[0121] In one embodiment, after uploading the image to be stored to the distributed file system and uploading the image content and storage address to the search server, the method further includes:

[0122] The search server is configured to support Chinese characters and / or pinyin letters as the retrieval data type.

[0123] In one embodiment, when the search keyword includes Chinese characters and pinyin letters, matching the search keyword with the image content stored in the search server to determine the candidate image matching the search keyword includes:

[0124] Matching the Chinese characters in the search keyword with the image content stored in the search server to obtain a first search result;

[0125] Convert the Chinese characters and pinyin letters in the search keyword into pinyin and then match them with the picture content stored in the search server to obtain a second search result;

[0126] The first search result and the second search result are combined to determine a combined candidate image.

[0127] In one embodiment, the searching and recommending based on the candidate images and the image library neural network constructed based on the search records to determine the corresponding recommended images specifically includes:

[0128] The gallery neural network is called according to the candidate images, and a number of images with the highest number of clicks are obtained from the gallery neural network according to a three-degree relationship algorithm as the recommended images.

[0129] In one embodiment, after displaying the candidate images and the recommended images and determining the target search image from the candidate images and the recommended images according to the user's selection operation, the method further comprises:

[0130] Update the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network.

[0131] In one embodiment, updating the search relationship between the current user node and the node of the target search image in the gallery neural network specifically includes:

[0132] The number of clicks on the target retrieval image by the current user in the gallery neural network is increased by 1.

[0133] An embodiment of the present invention provides a non-volatile computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the image retrieval method in any of the above method embodiments is executed, for example, the following steps are performed:

[0134] Receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters;

[0135] Matching the search keyword with the picture content stored in the search server to determine candidate pictures that match the search keyword;

[0136] Performing retrieval recommendation based on the candidate images and the image library neural network constructed based on the retrieval records to determine the corresponding recommended images;

[0137] The candidate images and recommended images are displayed, and a target retrieval image is determined from the candidate images and recommended images according to a user's selection operation.

[0138] In one embodiment, before receiving the search keyword, the method further includes:

[0139] Get the picture to be stored;

[0140] Performing image and text recognition on the image to be stored to obtain corresponding image content;

[0141] Uploading the image to be stored to the distributed file system, and uploading the image content and storage address to the search server;

[0142] A node for the image to be stored is created in the image library neural network, and the image library neural network is used to maintain a retrieval relationship between a retrieval user node and an image node according to a retrieval record.

[0143] In one embodiment, after uploading the image to be stored to the distributed file system and uploading the image content and storage address to the search server, the method further includes:

[0144] The search server is configured to support Chinese characters and / or pinyin letters as the retrieval data type.

[0145] In one embodiment, when the search keyword includes Chinese characters and pinyin letters, matching the search keyword with the image content stored in the search server to determine the candidate image matching the search keyword includes:

[0146] Matching the Chinese characters in the search keyword with the image content stored in the search server to obtain a first search result;

[0147] Convert the Chinese characters and pinyin letters in the search keyword into pinyin and then match them with the picture content stored in the search server to obtain a second search result;

[0148] The first search result and the second search result are combined to determine a combined candidate image.

[0149] In one embodiment, the searching and recommending based on the candidate images and the image library neural network constructed based on the search records to determine the corresponding recommended images specifically includes:

[0150] The gallery neural network is called according to the candidate images, and a number of images with the highest number of clicks are obtained from the gallery neural network according to a three-degree relationship algorithm as the recommended images.

[0151] In one embodiment, after displaying the candidate images and the recommended images and determining the target search image from the candidate images and the recommended images according to the user's selection operation, the method further comprises:

[0152] Update the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network.

[0153] In one embodiment, updating the search relationship between the current user node and the node of the target search image in the gallery neural network specifically includes:

[0154] The number of clicks on the target retrieval image by the current user in the gallery neural network is increased by 1.

[0155] As an example, non-volatile storage media can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) as external cache memory. By way of illustration and not limitation, RAM can be obtained in many forms such as synchronous RAM (SRAM), dynamic RAM, (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable types of memory.

[0156] In summary, the present invention discloses a method, device, system and medium for image retrieval, wherein the method includes: receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters; matching the search keyword with the image content stored in the search server to determine the candidate images that match the search keyword; performing search and recommendation based on the candidate images and a gallery neural network constructed based on the search records to determine the corresponding recommended images; displaying the candidate images and the recommended images, and determining the target search image from the candidate images and the recommended images based on the user's selection operation. By matching the image content with the search keyword, and performing intelligent image search and recommendation based on the retrieved candidate images and the pre-constructed gallery neural network, the user can directly and efficiently retrieve the most relevant images by entering keywords, thereby effectively improving the efficiency and accuracy of image retrieval.

[0157] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium, and the computer program can include the processes of the above-mentioned method embodiments when executed. The storage medium can be a memory, a disk, a floppy disk, a flash memory, an optical storage device, etc.

[0158] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A picture retrieval method, characterized in that: include: Receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters; Matching the search keyword with the picture content stored in the search server to determine candidate pictures that match the search keyword; Performing retrieval recommendation based on the candidate images and the image library neural network constructed based on the retrieval records to determine the corresponding recommended images; The candidate images and recommended images are displayed, and a target retrieval image is determined from the candidate images and recommended images according to a user's selection operation.

2. The image retrieval method according to claim 1, characterized in that: Before receiving the search keyword, the method further includes: Get the picture to be stored; Performing image and text recognition on the image to be stored to obtain corresponding image content; Uploading the image to be stored to the distributed file system, and uploading the image content and storage address to the search server; A node for the image to be stored is created in the image library neural network, and the image library neural network is used to maintain a retrieval relationship between a retrieval user node and an image node according to a retrieval record.

3. The image retrieval method according to claim 2, characterized in that: After uploading the image to be stored to the distributed file system, and uploading the image content and storage address to the search server, the method further includes: The search server is configured to support Chinese characters and / or pinyin letters as the retrieval data type.

4. The image retrieval method according to claim 1, characterized in that: When the search keyword includes Chinese characters and pinyin letters, matching the search keyword with the picture content stored in the search server to determine the candidate pictures matching the search keyword includes: Matching the Chinese characters in the search keyword with the image content stored in the search server to obtain a first search result; Convert the Chinese characters and pinyin letters in the search keyword into pinyin and then match them with the picture content stored in the search server to obtain a second search result; The first search result and the second search result are combined to determine a combined candidate image.

5. The image retrieval method according to claim 1, characterized in that: The performing retrieval and recommendation based on the candidate images and the image library neural network constructed based on the retrieval records to determine the corresponding recommended images specifically includes: The gallery neural network is called according to the candidate images, and a number of images with the highest number of clicks are obtained from the gallery neural network according to a three-degree relationship algorithm as the recommended images.

6. The image retrieval method according to claim 1, characterized in that: After displaying the candidate images and the recommended images, and determining the target search image from the candidate images and the recommended images according to the user's selection operation, the method further includes: Update the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network.

7. The image retrieval method according to claim 6, characterized in that: The updating of the retrieval relationship between the current user node and the node of the target retrieval image in the gallery neural network specifically includes: The number of clicks on the target retrieval image by the current user in the gallery neural network is increased by 1.

8. A picture retrieval device, characterized in that: include: A receiving module, used for receiving a search keyword, wherein the search keyword includes Chinese characters and / or pinyin letters; A matching module, used to match the search keyword with the image content stored in the search server, and determine the candidate images that match the search keyword; A recommendation module, used to perform retrieval and recommendation based on the candidate images and a picture library neural network constructed based on the retrieval records, and determine corresponding recommended images; The display determination module is used to display the candidate images and recommended images, and determine the target retrieval image from the candidate images and recommended images according to the user's selection operation.

9. An image retrieval system, characterized in that: The system includes at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed 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 the image retrieval method described in any one of claims 1-7.

10. A non-volatile computer-readable storage medium, characterized in that: The non-volatile computer-readable storage medium stores computer-executable instructions, which, when executed by one or more processors, enable the one or more processors to execute the image retrieval method described in any one of claims 1-7.

Images