Searching method and device

Abstract

The invention discloses a search method and device. The search method comprises the steps of obtaining a target search request for target content; searching in a first target image set based on the target content to obtain a first image set; obtaining a reference image obtained by searching in a second target image set by a second device based on the target content; searching in the first target image set based on the reference image to obtain a second image set; based on the first image set and the second image set, obtaining a target image set for the target search request; wherein the first target image set is an image set corresponding to the search range of the first equipment; the second target image set is an image set corresponding to the search range of the second equipment.

Description

Technical Field

[0001] This application relates to search technology, and more specifically, to a search method and device. Background Art

[0002] Currently, when searching for images through artificial intelligence (AI) technology, most use the CLIP (Contrastive Language-Image Pre-Training, an image-text multimodal model based on contrastive learning) model. When two interconnected electronic devices perform a global search based on the CLIP model, the search results of the two devices often have inconsistent search quality and standards. Summary of the Invention

[0003] In view of this, this application provides the following technical solutions:

[0004] In the first aspect of this application, a search method for a first device is provided, including:

[0005] Obtain a target search request for target content;

[0006] Search in a first target image set based on the target content to obtain a first image set;

[0007] Obtain a reference image obtained by a second device searching in a second target image set based on the target content;

[0008] Search in the first target image set based on the reference image to obtain a second image set;

[0009] Based on the first image set and the second image set, obtain a target image set for the target search request;

[0010] Wherein, the first target image set is the image set corresponding to the search scope of the first device; the second target image set is the image set corresponding to the search scope of the second device.

[0011] In a possible implementation, the searching in the first target image set based on the target content to obtain a first image set includes:

[0012] Obtain a first vector to be matched corresponding to the target content;

[0013] Match the first vector to be matched with the image vectors corresponding to each image in the first target image set to obtain a first image set whose matching degree meets a first matching condition;

[0014] The searching in the first target image set based on the reference image to obtain a second image set includes:

[0015] Obtain a second vector to be matched corresponding to the reference image;

[0016] Match the second vector to be matched with the image vectors corresponding to the images in the first target image set to obtain a second image set whose matching degree meets the second matching condition.

[0017] In a possible implementation, the obtaining of the first vector to be matched corresponding to the target content includes:

[0018] Input the target content into the first intelligent engine to obtain the first vector to be matched corresponding to the target content;

[0019] The obtaining of the second vector to be matched corresponding to the reference image includes:

[0020] Input the reference image into the second intelligent engine to obtain the second vector to be matched corresponding to the reference image.

[0021] In a possible implementation, the reference image is an image searched by the second device in the second target image set and having a matching degree higher than a set threshold with respect to the target content.

[0022] In a possible implementation, the target search request is a search request input locally by the first device or a search request received from the second device;

[0023] If the target search request is a search request input locally by the first device, the method further includes:

[0024] Send the obtained target search request to the second device so that the second device performs a picture search in the second target image set based on the target search request.

[0025] In a possible implementation, if the target search request is a search request input locally by the first device, it further includes:

[0026] Display the images included in the target image set;

[0027] If the target search request is a search request received from the second device, the method further includes:

[0028] Send the target image set to the second device so that the second device displays and outputs the target image set.

[0029] In a possible implementation, the displaying and outputting of the images included in the target image set includes:

[0030] Display the images included in the target image set in a target order;

[0031] Sending the target image set to the second device includes:

[0032] Sending the target image set in the target order to the second device, so that the second device displays the target image set in the target order;

[0033] Wherein, the target order is determined according to the matching degree between the image vectors of the respective images in the target image set and the first vector to be matched of the target content.

[0034] In a possible implementation manner, it further includes:

[0035] Determining a sub-image set in the second target image set excluding the images in the first target image set;

[0036] Matching the first vector to be matched with the image vectors corresponding to the respective images included in the determined sub-graphic set to obtain the matching degrees corresponding to the respective images in the sub-graphic set.

[0037] In a possible implementation manner, displaying the images included in the target image set in the target order includes:

[0038] Displaying the target image set corresponding to the first device in the first area;

[0039] Displaying the target image set corresponding to the second device in the second area.

[0040] A second aspect of this application provides a search device, which is applied to the first device and includes:

[0041] A request obtaining module, configured to obtain a target search request for the target content;

[0042] A first image search module, configured to search in the first target image set based on the target content to obtain a first image set;

[0043] A reference obtaining module, configured to obtain a reference image obtained by the second device searching in the second target image set based on the target content;

[0044] A second image search module, configured to search in the first target image set based on the reference image to obtain a second image set;

[0045] A target image set obtaining module, configured to obtain a target image set for the target search request based on the first image set and the second image set;

[0046] Wherein, the first target image set is the image set corresponding to the search range of the first device; the second target image set is the image set corresponding to the search range of the second device. Description of the Drawings

[0047] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the provided drawings.

[0048] Figure 1 It is a flowchart of a search method disclosed in an embodiment of the present application;

[0049] Figure 2 It is a flowchart of obtaining a first image set disclosed in an embodiment of the present application;

[0050] Figure 3 It is a flowchart of obtaining a second image set disclosed in an embodiment of the present application;

[0051] Figure 4 It is a schematic diagram of the implementation process of the search solution disclosed in an embodiment of the present application;

[0052] Figure 5 It is a schematic structural diagram of a search device disclosed in an embodiment of the present application;

[0053] Figure 6 It is a schematic structural diagram of an electronic device disclosed in an embodiment of the present application. Detailed implementation manners

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0055] In many current scenarios, users will interconnect two or more electronic devices and then trigger a global search on one of the electronic devices. That is, a search request is input on one electronic device, and this search request not only performs relevant searches on the local device, but also on the sides of other electronic devices interconnected with this device. Finally, the search results on each interconnected device can be aggregated and output to the user.

[0056] In practical applications, due to differences in computing power or other resources among different electronic devices, even if they are equipped with the same type of search model, the quality and standards of the search results obtained by performing searches will be different. For example, for the CLIP model installed on different electronic devices, based on the same search conditions, one device may be able to search for the target picture, while another device may not be able to search for the target picture.

[0057] In view of the above problems, the present application proposes a solution aiming to improve the comprehensiveness of the search model in performing searches and enhance the accuracy of search results. Embodiments of the present application can be applied to electronic devices, and the present application does not limit the product form of the electronic device, which may include but is not limited to smart phones, tablet computers, wearable devices, personal computers (PCs), netbooks, etc., and can be selected according to application requirements.

[0058] Figure 1 It is a flowchart of a search method disclosed in an embodiment of the present application. Figure 1 The method shown is applied to a first device, and the first device has a connection relationship with a second device. Here, the connection relationship can be a wired connection or a wireless connection. Refer to Figure 1 As shown, the search method may include:

[0059] Step 101: Obtain a target search request for target content.

[0060] In the present application, the target search request is a request for searching for pictures related to the target content. In practical applications, the target search request may be a text search request, that is, an image search request based on a piece of text; correspondingly, in the implementation, the target content is text content, and this text content can record the conditions that the searched image needs to meet. For example, if it is "a white cat is sleeping", then the searched image needs to meet the conditions of "white cat" and "sleeping". Or, the target search request may also be an image search request, that is, an image search request based on a picture; correspondingly, in the implementation, the target content is a picture of a brown puppy, then the searched image needs to meet the condition of "brown puppy".

[0061] The target search request may be a search request input by the user on the first device side, or may also be a search request received from other devices, such as the second device. The present application does not impose a fixed limit on this. In the present application, there is a connection relationship between the first device and the second device, and there is a communication channel between the two. The second device can transmit the target search request input on the second device side to the first device through the communication channel, and the first device can also send the target search request input locally to the second device through the communication channel, so that both the first device and the second device can perform search tasks based on the target search request, that is, to achieve "global search". Here, "global search" refers to the search of the storage areas corresponding to at least two electronic devices with a connection relationship.

[0062] Step 102: Search in a first target image set based on the target content to obtain a first image set.

[0063] Among them, the first target image set is the image set corresponding to the search range of the first device; the search range of the first device includes, but is not limited to, the local storage device of the first device or other storage devices connected to the first device. Therefore, the first target image set at least includes all the images stored locally on the first device, and may also include images on other devices, such as images in a storage disk or memory card inserted into the first device.

[0064] The implementation of step 102 can be completed based on a search model in the field, such as the CLIP model. Since the target content is text content, the implementation of step 102 is also a process of searching for images by text. The matching degree between the images in the first image set and the target content meets the set conditions, such as images with a matching degree greater than 80%. There are many implementation methods for obtaining the first image set in the related art, which will not be elaborated one by one in this disclosure. A search process shown in this disclosure will be introduced later.

[0065] Step 103: Obtain a reference image obtained by the second device searching in the second target image set based on the target content.

[0066] The second target image set is the image set corresponding to the search range of the second device; the search range of the second device includes, but is not limited to, the local storage device of the second device or other storage devices connected to the second device.

[0067] Since this application is a global search for the mutually connected first device and second device, in addition to the first device searching based on the target content, the second device will also search based on the target content. The implementation of the second device searching based on the target content is the same as that of the first device searching based on the target content. The reference image is an image in the second target image set searched by the second device whose matching degree with the target content meets the set requirements. In one example, the reference image is the image with the highest matching degree with the target content in the second image set.

[0068] Step 104: Search in the first target image set based on the reference image to obtain a second image set.

[0069] The implementation of step 104 can also be completed based on the CLIP model. Since the reference image is an image, the implementation of step 104 is also a process of searching for images by image. In the CLIP model, the accuracy of image-based search is higher than that of text-based search. Therefore, in this application, in addition to searching for images based on the target content (text content), a second image search will be further performed in the first target image set corresponding to the first device based on the reference image to obtain a second image set. Subsequently, the results of the first image set and the second image set can be integrated to ensure the comprehensiveness and accuracy of the search results.

[0070] Step 105: Obtain a target image set for the target search request based on the first image set and the second image set.

[0071] After obtaining the first image set and the second image set, a target image set can be obtained based on processing the two. For example, perform a merging and duplicate removal operation on the first image set and the second image set to obtain the target image set. Since the target image set combines the first image set and the second image set, that is, combines the results of text search images and image search images, the coverage of images that "match" the target content in the first target image set is more comprehensive, and the search result accuracy is higher.

[0072] For example, in a scenario where a mobile phone is connected to a laptop via wireless connection or data cable, and a target search request is triggered on the mobile phone. The mobile phone will perform a first search locally. The laptop searches for the target search request on the computer side, obtains a reference image, and sends the reference image to the mobile phone. Then the mobile phone performs a second search locally based on the received reference image, and finally combines the search results of the two searches to obtain the target image set.

[0073] In the search method of this embodiment, after the first device obtains the target search request for the target content, it will not only perform a first search based on the target content, but also further perform a second search according to the reference image obtained by the second device based on the target content. The target image set of the target search request is obtained by combining the search results of the two searches; since two searches are performed using different search methods, the comprehensiveness and accuracy of the search can be effectively guaranteed.

[0074] Figure 2 This is a flowchart of obtaining the first image set disclosed in the embodiments of the present application. Refer to Figure 2 As shown, in the above embodiment content, the searching in the first target image set based on the target content to obtain the first image set may include:

[0075] Step 201: Obtain a first vector to be matched corresponding to the target content.

[0076] In a search model, whether obtaining text content or image content, it is necessary to first extract the corresponding vector, that is, the "language" that the machine can recognize, and then perform subsequent searches based on the vector. In the embodiments of the present application, after obtaining the target search request for the target content, first parse the request to obtain the target content therein, and perform processing such as splitting and recognition on the target content to obtain the first vector to be matched corresponding to the target content.

[0077] Step 202: Match the first vector to be matched with the image vectors corresponding to the images in the first target image set to obtain a first image set whose matching degree meets the first matching condition.

[0078] Beforehand, the search model can extract the image vectors of all the images within the search range of the first device, and store the image vectors of each image within the search range in the image vector library. In this way, after obtaining the first vector to be matched, the first vector to be matched can be directly matched with the image vectors corresponding to the images in the first target image set, that is, the first vector to be matched is matched with the image vectors of each image in the image vector library to obtain a first image set whose matching degree meets the first matching condition. Among them, the first matching condition has a default value, such as 70%, and the user can also configure it according to the application scenario requirements.

[0079] Figure 3 It is a flowchart for obtaining the second image set disclosed in the embodiments of the present application. Refer to Figure 3 As shown, the searching in the first target image set based on the reference image to obtain the second image set may include:

[0080] Step 301: Obtain a second vector to be matched corresponding to the reference image.

[0081] The first vector to be matched is obtained by extracting from the target content (text content), and the second vector to be matched is obtained by extracting from the reference image. In one implementation, both the first vector to be matched and the second vector to be matched are one-dimensional vectors, and the image vectors extracted by the search model from each image (the images in the first target image set) within the search range of the first device are also one-dimensional vectors.

[0082] Step 302: Match the second vector to be matched with the image vectors corresponding to the images in the first target image set to obtain a second image set whose matching degree meets the second matching condition.

[0083] After obtaining the second vector to be matched, the second vector to be matched is compared with the image vectors of each image in the first target image set in sequence to determine the matching degree between each image and the second vector to be matched, and the set of all images that meet the second matching condition is determined as the second image set.

[0084] Specifically, the obtaining of the first vector to be matched corresponding to the target content may include: inputting the target content into the first intelligent engine to obtain the first vector to be matched corresponding to the target content. The obtaining of the second vector to be matched corresponding to the reference image may include: inputting the reference image into the second intelligent engine to obtain the second vector to be matched corresponding to the reference image.

[0085] When the target content is text content, the first intelligent engine extracts a first vector to be matched from the text, and the second intelligent engine extracts a second vector to be matched from the image. The first intelligent engine and the second intelligent engine are independent intelligent engines with different functions. When the target content is an image, both the first intelligent engine and the second intelligent engine are used to extract the second vector to be matched from the image, and the first intelligent engine and the second intelligent engine are the same intelligent engine.

[0086] The above content introduces the specific implementation of obtaining the first image set and the second image set, which is convenient for those skilled in the art to better understand and implement the technical solution of this application.

[0087] In one implementation, the reference image may be a thumbnail. Since the technical solution of this application is only used for image search rather than image processing, in order to save some unnecessary resource overheads, the reference image may be a thumbnail. Of course, the resolution of the thumbnail needs to ensure that the second intelligent engine can extract image vectors from it that can meet the subsequent matching and comparison requirements.

[0088] The implementation of using a thumbnail as the reference image saves the resource overhead of image transmission and processing on the one hand, and can also improve the overall image search speed on the other hand.

[0089] The reference image may be an image in the second target image set searched by the second device with a matching degree higher than a set threshold with respect to the target content. Based on different set thresholds, the number of images in the second target image set that can be used as reference images is different. In one example, the image with the highest matching degree with the reference image in the second target image set can be determined as the reference image. Here, the matching degree is the matching degree between the image vector and the second vector to be matched described above.

[0090] When the reference image is an image in the second target image set searched by the second device with a matching degree higher than a set threshold with respect to the target content, by configuring the set threshold, a reference image highly matching the target content can be obtained, indirectly improving the accuracy of the second search by the first device based on the reference image.

[0091] In the content of the foregoing embodiments, the target search request may be a search request input locally by the first device, or may also be a search request received from the second device. For example, in one scenario, the user connects the mobile phone to the TV through a wireless connection. The user can input the target content on the mobile phone side and click the search instruction to trigger the generation of the target search request; or the user can input the target content on the TV through the remote control and click the search instruction to trigger the generation of the target search request.

[0092] In one implementation, if the target search request is a search request locally input by the first device, the search method may further include: sending the obtained target search request to the second device, so that the second device performs a picture search in the second target image set based on the target search request. The reference image is an image in the image set obtained by the second device based on the target search request.

[0093] It should be noted that the search model on the second device and the search model on the first device may be the same search model or different search models; the difference here includes different types or different versions. Thus, the processing of obtaining the image set by the second device's search is the same as or different from the processing of obtaining the image set by the first device's search.

[0094] Based on the foregoing disclosure, if the target search request is a search request locally input by the first device, the method may further include: displaying and outputting the images included in the target image set. If the target search request is a search request received from the second device, the method may further include: sending the target image set to the second device, so that the second device displays and outputs the target image set.

[0095] That is, initially, on which device the user inputs the target content and performs the image search, the final obtained target image set will be displayed on that device. On which device the user inputs the target content probably indicates that the user is currently using that device. Therefore, after obtaining the image search result, it is necessary to control the display of the image search result on the device side where the user inputs the search request.

[0096] The displaying and outputting the images included in the target image set may include: displaying the images included in the target image set in a target order. Wherein, the target order is determined according to the matching degree between the image vectors of each image in the target image set and the first vector to be matched of the target content. During the process of determining the search result, the images are matched with the first vector to be matched of the target content. Therefore, each image can record the matching degree with the first vector to be matched. When displaying the final target image set, the images can be sorted in descending order of the matching degree to obtain the target order of all the images in the target image set. When displaying the target image set, the images included in the target image set may be displayed in the target order.

[0097] If the target image set needs to be sent to the second device for display, the target image set in the target order may also be sent to the second device, so that the second device displays the target image set in the target order.

[0098] Since the target order is determined according to the matching degree between each image and the first vector to be matched of the target content, the higher the matching degree of the image, the more likely it is to conform to the user's search intention. Therefore, displaying the images in the target image set according to the target order can better meet the user's needs.

[0099] In other implementations, the search method may further include: determining a subset of images in the second target image set that excludes the first target image set; matching the first vector to be matched with the image vectors corresponding to the images included in the determined subset of images to obtain the matching degrees corresponding to the images in the subset of images.

[0100] Specifically, when displaying the images included in the target image set according to the target order, the target order depends on the matching degree between the target content and the images. In other words, the higher the matching degree between the image and the target content, the more the image conforms to the search request, and the higher its display order.

[0101] When the first device performs the first round of search based on the target content (i.e., the process of searching for the first image set in the first target image set based on the target content), during the search process, since it is necessary to match the first vector to be matched with the image vectors of the images in the first target image set, the matching degrees between the image vectors of each image and the first vector to be matched, that is, the matching degrees between the images and the target content, can be obtained during the search process, and the sorting of each image in the first image set can be obtained based on the matching degrees. When performing the second round of search based on the reference image (i.e., the process of searching for the second image set in the first target image set based on the reference image), the matching degrees between the reference image and the images are obtained during the search process. For the partial images where the second-round search results do not overlap with the first-round search results, there is no matching degree between the target content and the images. Therefore, it is necessary to re-match the non-overlapping partial images with the target content to obtain the matching degrees between these images and the target content. The premise of the above implementation is to determine the display order based on the matching degree between the target content and the images, or in other words, to determine which image better meets the requirements based on the matching degree between the image and the target content.

[0102] In practical applications, there may be duplicate images in the search results corresponding to the two searches performed by the first device. Therefore, when determining the matching degrees between the images in the second image set and the first vector to be matched, it is not necessary to determine the matching degrees between all the images and the first vector to be matched, but only to determine the matching degrees between the target content and the partial images in the second image set that do not overlap with the first image set. In this embodiment, first, a subset of images in the second target image set that excludes the images in the first target image set is determined, and then only for the images in the subset that do not have a matching degree with the first vector to be matched, the matching degree with the first vector to be matched is determined, which can effectively reduce the amount of calculation and improve the overall image search efficiency.

[0103] Alternatively, in another implementation, after obtaining the second image set, instead of performing another matching between the images and the target content, the images in the target image set are comprehensively sorted directly according to the matching degrees between the images in the first image set and the target content during the first-round search and the matching degrees between the images in the second image set and the reference image during the second-round search. In other words, the display order of the target images is based on the matching degrees obtained during their respective search processes.

[0104] For the overlapping images in the first image set and the second image set, since they have both the matching degree with the target content and the matching degree with the reference image, when sorting the images in the target image set, the overlapping images can be sorted according to their matching degrees with the target content. That is, for the overlapping images in the first image set and the second image set, their sorting is determined based on the matching degrees determined by the first device during the first-round search. Alternatively, for the overlapping images in the first image set and the second image set, when displaying the images in the target image set, these overlapping images can be output first, and then the other images in the first data set and the second data set are sorted according to the matching degrees. Among them, the overlapping images can also be sorted according to the matching degrees determined for each image during the first-round search when output, or sorted according to the matching degrees determined for each image during the second-round search. This application does not impose a fixed limit on this.

[0105] In one implementation, the displaying of the images included in the target image set according to the target order may include: displaying the target image set corresponding to the first device in the first region; and displaying the target image set corresponding to the second device in the second region.

[0106] In this implementation, when displaying the target image set on the device, it is necessary to display not only the target image set determined by the first device through search, but also the target image set determined by the second device through search. The target image sets of different devices are displayed in separate regions to facilitate the user to better distinguish which device the target image set comes from.

[0107] Among them, the method for obtaining the target image set corresponding to the second device may be the same as that for the first device to obtain the target image set. For example, first perform a first search based on the target content, and then perform a second search based on the reference picture sent by the first device, and merge and deduplicate the results of the two searches to obtain the target image set. The method for obtaining the target image set corresponding to the second device may also be different from that for the first device to obtain the target image set. For example, the second device only performs one search and directly determines the target image set based on the result of the one search.

[0108] The display order of the target image set corresponding to the second device is the same as that of the first device, and is sorted from high to low according to the matching degrees between the image vectors of the images in the target image set and the first vector to be matched.

[0109] Figure 4 This is a schematic diagram of the implementation process of the search solution disclosed in the embodiments of the present application. In combination with Figure 4 as shown, the mobile phone is connected to the computer. When the user inputs the target content (text content):

[0110] 1. The mobile phone and the computer simultaneously start searching for relevant pictures on their own devices in the way of searching pictures by text, and each obtains the corresponding search results (corresponding to the first image set);

[0111] 2. The mobile phone obtains the picture search results of the computer, searches for similar pictures in the way of searching pictures by picture, and obtains the search results (corresponding to the second image set); after integrating and removing duplicates with the search results on the mobile phone itself, a final picture list (corresponding to the target image set) is generated;

[0112] 3. The computer obtains the picture search results of the mobile phone, searches for similar pictures in the way of searching pictures by picture, and obtains the search results (corresponding to the second image set); after integrating and removing duplicates with the search results on the computer itself, a final picture list (corresponding to the target image set) is generated.

[0113] The search solution disclosed in the present application can improve the accuracy of the search. The implementation of the solution does not require adding new hardware devices, and the product is easy to use and promote.

[0114] For the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0115] In the foregoing embodiments disclosed in the present application, the method is described in detail. The method of the present application can be implemented by means of devices in various forms. Therefore, the present application also discloses a device, and specific embodiments are given below for detailed description.

[0116] Figure 5 This is a schematic diagram of the structure of a search device disclosed in the embodiments of the present application. Refer to Figure 5 as shown, the search device 50 may include:

[0117] A request acquisition module 501, configured to acquire a target search request for the target content.

[0118] A first image search module 502, configured to search in the first target image set based on the target content to obtain a first image set.

[0119] A reference acquisition module 503 is configured to acquire a reference image obtained by the second device through searching in a second target image set based on the target content.

[0120] A second image search module 504 is configured to search in the first target image set based on the reference image to obtain a second image set.

[0121] A target image set acquisition module 505 is configured to obtain a target image set for the target search request based on the first image set and the second image set.

[0122] Wherein, the first target image set is an image set corresponding to the search range of the first device; the second target image set is an image set corresponding to the search range of the second device.

[0123] After obtaining a target search request for target content, the search device in this embodiment not only performs a first search based on the target content, but also further performs a second search according to the reference image obtained by the second device through searching based on the target content, and obtains a target image set for the target search request by combining the search results of the two searches; since two searches are performed using different search methods, the comprehensiveness and accuracy of the search can be effectively ensured.

[0124] In one implementation, the first image search module may include: a first vector acquisition module configured to acquire a first vector to be matched corresponding to the target content; a first matching module configured to match the first vector to be matched with image vectors corresponding to each image in the first target image set to obtain a first image set whose matching degree meets a first matching condition; the second image search module may include: a second vector acquisition module configured to acquire a second vector to be matched corresponding to the reference image; a second matching module configured to match the second vector to be matched with image vectors corresponding to each image in the first target image set to obtain a second image set whose matching degree meets a second matching condition.

[0125] In one implementation, the first vector acquisition module is specifically configured to: input the target content into a first intelligent engine to obtain a first vector to be matched corresponding to the target content; the second vector acquisition module is specifically configured to: input the reference image into a second intelligent engine to obtain a second vector to be matched corresponding to the reference image.

[0126] In one implementation, the reference image is a thumbnail.

[0127] In one implementation, the reference image is an image in the second target image set searched by the second device and having a matching degree with the target content higher than a set threshold.

[0128] In one implementation, the target search request is a search request input locally by the first device or a search request received from the second device. If the target search request is a search request input locally by the first device, the apparatus may further include: a request sending module, configured to send the obtained target search request to the second device, so that the second device performs a picture search in the second target image set based on the target search request.

[0129] In one implementation, if the target search request is a search request input locally by the first device, the apparatus may further include: a display control module, configured to display and output the images included in the target image set. If the target search request is a search request received from the second device, the apparatus may further include: an image sending module, configured to send the target image set to the second device, so that the second device displays and outputs the target image set.

[0130] In one implementation, the display control module may specifically be configured to: display the images included in the target image set in a target order. The image sending module may specifically be configured to: send the target image set in the target order to the second device, so that the second device displays the target image set in the target order. The target order is determined according to the matching degree between the image vectors of the respective images in the target image set and the first vector to be matched of the target content.

[0131] In one implementation, the apparatus may further include: a sub-image set processing module, configured to determine a sub-image set in the second target image set that excludes the images in the first target image set; match the first vector to be matched with the image vectors corresponding to the respective images included in the determined sub-graphic set to obtain the matching degrees corresponding to the respective images in the sub-graphic set.

[0132] In one implementation, the display control module may specifically be configured to: display the target image set corresponding to the first device in a first area; display the target image set corresponding to the second device in a second area.

[0133] For the specific implementation of the above search apparatus and each module included therein, reference may be made to the corresponding content introduction in the method embodiment, which will not be repeated here.

[0134] Any one of the search apparatuses in the above embodiments includes a processor and a memory. The request obtaining module, the first image search module, the reference obtaining module, the second image search module, the target image set obtaining module, the first vector obtaining module, the first matching module, the second vector obtaining module, the second matching module, etc. in the above embodiments are all stored in the memory as program modules, and the corresponding functions are implemented by the processor executing the above program modules stored in the memory.

[0135] The processor includes a kernel, which retrieves the corresponding program module from the memory. One or more kernels can be set, and the processing of the access data can be realized by adjusting the kernel parameters.

[0136] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0137] In an exemplary embodiment, a computer-readable storage medium is also provided, which can be directly loaded into the internal memory of a computer and contains software code. After being loaded and executed by a computer, the computer program can implement the steps shown in any embodiment of the above-mentioned search method.

[0138] In an exemplary embodiment, a computer program product is also provided, which can be directly loaded into the internal memory of a computer, wherein the computer program contains software codes, and after being loaded and executed by a computer, the computer program can implement the steps shown in any embodiment of the search method described above.

[0139] Furthermore, an embodiment of the present application provides an electronic device. Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present application. Figure 6 As shown, the electronic device 60 includes at least one processor 601, and at least one memory 602 and a bus 603 connected to the processor; wherein the processor and the memory communicate with each other via the bus; the processor is used to call the program instructions in the memory to execute the above-mentioned search method.

[0140] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0141] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0142] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0143] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A search method, which should be used in a first device, includes: Obtaining a target search request for target content; Searching in the first target image set based on the target content to obtain a first image set; Obtain a reference image obtained by searching the second target image set by the second device based on the target content; Searching in the first target image set based on the reference image to obtain a second image set; Based on the first image set and the second image set, obtaining a target image set for the target search request; The first target image set is an image set corresponding to the search range of the first device; and the second target image set is an image set corresponding to the search range of the second device.

2. The search method according to claim 1, wherein searching in the first target image set based on the target content to obtain the first image set comprises: Obtaining a first to-be-matched vector corresponding to the target content; Matching the first to-be-matched vector with image vectors corresponding to each image in a first target image set to obtain a first image set whose matching degree satisfies a first matching condition; The step of searching in the first target image set based on the reference image to obtain a second image set comprises: Obtaining a second to-be-matched vector corresponding to the reference image; The second to-be-matched vector is matched with the image vector corresponding to each image in the first target image set to obtain a second image set whose matching degree satisfies a second matching condition.

3. The search method according to claim 2, wherein obtaining the first to-be-matched vector corresponding to the target content comprises: Inputting the target content into a first intelligent engine to obtain a first to-be-matched vector corresponding to the target content; The obtaining of a second to-be-matched vector corresponding to the reference image includes: The reference image is input into a second intelligent engine to obtain a second vector to be matched corresponding to the reference image.

4. The search method according to claim 1, wherein: The reference image is an image found by the second device in the second target image set and having a matching degree with the target content higher than a set threshold.

5. The search method according to claim 1, wherein: The target search request is a search request locally input by the first device, or a search request received from the second device; If the target search request is a search request locally input by the first device, the method further includes: The obtained target search request is sent to the second device, so that the second device performs an image search in the second target image set based on the target search request.

6. The search method according to claim 5, if the target search request is a search request locally input by the first device, further comprising: Display the images contained in the output target image set; If the target search request is a search request received from the second device, the method further includes: The target image set is sent to the second device, so that the second device displays and outputs the target image set.

7. The search method according to claim 6, wherein the images contained in the output target image set are displayed, comprising: Displaying the images included in the target image set in a target order; Sending the target image set to the second device includes: sending the target image set in the target order to the second device, so that the second device displays the target image set in the target order; The target order is determined according to the matching degree between the image vector of each image in the target image set and the first to-be-matched vector of the target content.

8. The search method according to claim 6, further comprising: Determine a sub-image set in the second target image set excluding the images in the first target image set; The first vector to be matched is matched with the image vectors corresponding to each image included in the determined sub-graphics set to obtain the matching degree corresponding to each image in the sub-graphics set.

9. The search method according to claim 7, wherein the step of displaying the images included in the target image set in target order comprises: Displaying a target image set corresponding to the first device in the first area; The target image set corresponding to the second device is displayed in the second area.

10. A search device, applied to a first device, comprising: A request obtaining module, used for obtaining a target search request for target content; A first image search module, configured to search in a first target image set based on the target content to obtain a first image set; A reference obtaining module, used to obtain a reference image searched by a second device in a second target image set based on the target content; A second image search module, configured to search in the first target image set based on the reference image to obtain a second image set; A target image set obtaining module, configured to obtain a target image set for the target search request based on the first image set and the second image set; The first target image set is an image set corresponding to the search range of the first device; and the second target image set is an image set corresponding to the search range of the second device.

Images