WO2023284183A1

WO2023284183A1 - Image recognition method, electronic device, and computer-readable storage medium

Info

Publication number: WO2023284183A1
Application number: PCT/CN2021/128516
Authority: WO
Inventors: Dening DI; Jun Yin; Huadong PAN; Jingsong HAO; Shulei ZHU
Original assignee: Zhejiang Dahua Technology Co., Ltd.
Priority date: 2021-07-13
Filing date: 2021-11-03
Publication date: 2023-01-19
Also published as: CN113255703A

Abstract

The present application provides an image recognition method, an electronic device, and a computer-readable storage medium. The image recognition method includes: obtaining an image to be recognized, wherein the image to be recognized comprises a target object; in response to an archive corresponding to the target object being present in historic archives, determining whether identity information of the target object is present in the archive; determining the identity information as a recognition result of the image to be recognized in response to the identity information of the target object being present; and updating the image to be recognized and the recognition result into the archive. In this way, an efficiency and a recall rate of image recognition is improved.

Description

IMAGE RECOGNITION METHOD, ELECTRONIC DEVICE, AND COMPUTER-READABLE STORAGE MEDIUM

The present application claims priority of Chinese Patent Application No. 202110791380.5, archived on July 13, 2021, in China National Intellectual Property Administration, the entire contents of which are hereby incorporated by reference in their entireties.

TECHNICAL FIELD

The present disclosure relates to the field of image processing, and in particular to an image recognition method, an electronic device, and a computer readable storage medium.

BACKGROUND

In recent years, as the mobile internet and the internet of things develops, and living concepts of people are improved and changed, technologies of image recognition are widely applied. In detail, the image recognition is highly valued and applied in various fields, such as security, finance, human-computer interaction, information, education, and so on, and therefore, the image recognition has a promising future.

Currently, while recognizing images, the images are generally recognized in real time by means of a deployment mode. Further, the images are clustered to be one-person-one-archive and analyzed and judged. However, in the art, the computing amount and resource consumption of the deployment mode and the one-person-one-archive clustering are huge and inefficient.

Therefore, a method to reduce the computing amount and the resource consumption of the deployment mode and the one-person-per-archive clustering while recognizing images needs to be addressed.

SUMMARY OF THE DISCLOSURE

The present disclosure provides an image recognition method, an electronic device, and a computer readable storage medium, to solve the problem of an image recognition method in the art being inefficient and having huge computing amount.

According to a first aspect of the present disclosure, an image recognition method includes: obtaining an image to be recognized, wherein the image to be recognized comprises a target object; in response to an archive corresponding to the target object being present in historic archives, determining whether identity information of the target object is present in the archive; determining the identity information as a recognition result of the image to be recognized in response to the identity information of the target object being present; and updating the image to be recognized and the recognition result into the archive.

In some embodiments, the method further includes: recognizing an identity of the target object based on an identity information database in response to the archive corresponding to the target object being present in the historical archives, and the identity information of the target object being not present in the historical archives, or the archive corresponding to the target object being not present in the historical archives; and obtaining the recognition result of the image to be recognized.

In some embodiments, the method further includes: responding to a predetermined number of images being obtained; clustering the predetermined number of images based on target objects in the predetermined number of images to combine images having a same target object into a same archive to obtain the historical archives.

In some embodiments, after the clustering the predetermined number of images based on target objects in the predetermined number of images to combine images having a same target object into a same archive to obtain the historical archives, the method further includes: in response to the identity information of the target object being obtained, adding the identity information to the archive of the target object.

In some embodiments, the recognizing an identity of the target object based on an identity information database, and the obtaining the recognition result of the image to be recognized, includes: extracting a feature vector of the target object of the image to be recognized; comparing the feature vector with a plurality of standard feature vectors in the identity information database to obtain a similarity value between the feature vector and each of the plurality of standard feature vectors; and determining identity information of a standard feature vector corresponding to a maximum similarity value as the identity information of the target object to obtain the recognition result.

In some embodiments, the operation of in response to an archive corresponding to the target object being present in historic archives, determining whether identity information of the target object is present in the archive, comprises: comparing the target object of the image to be recognized with a historical archive portion until determining that the archive corresponding to the target object is present in the historic archives or traversing the historic archives; wherein the historic archives are divided into a plurality of historical archive portions based on a predetermined rule, and each historical archive portion comprises archives of a plurality of target objects.

In some embodiments, the comparing the target object of the image to be recognized with a historical archive portion until determining that the archive corresponding to the target object is present in the historic archives or traversing the historic archives, includes: comparing the image to be recognized to each of the plurality of historical archive portions successively in a cascade manner based on correspondence between the image to be recognized and each of the plurality of historical archive portions, until determining that the archive corresponding to the target object is present in the historical archives or traversing the historical archives.

In some embodiments, the updating the image to be recognized and the recognition result into the archive, includes: in response to a predetermined number of images to be recognized being obtained, updating the historical archives based on the predetermined number of images to be recognized and recognition results of the predetermined number of images.

According to a second aspect of the present disclosure, an electronic device includes a memory and a processor coupled to the memory. The processor is configured to execute program instructions stored in the memory to perform the image recognition method according to any one of above embodiments.

According to a third aspect of the present disclosure, a computer-readable storage medium stores program instructions. The program instructions, when being executed by a processor, implement the image recognition method according to any one of above embodiments.

According to the present disclosure, in response to an archive corresponding to a target object being present in historical archives, it may be determined whether identification information of the target object is present in the archive. In response to the identification information of the target object being present in the archive, the identity information is determined as a recognition result of an image to be recognized, such that the identity information of the target object may be determined directly based on the archive corresponding to the target object. A process, which compares the image to be recognized with a plurality of target objects in an identity information database to obtain the identity information of the target object, is simplified to some extent, reducing recognition workload. At last, the image to be recognized and the recognition result are updated into the archive, such that the image and the identity information of the archive may be updated, facilitating subsequent recognition of a new image to be recognized.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a flow chart of an image recognition method according to an embodiment of the present disclosure.

FIG. 2 is a flow chart of an image recognition method according to another embodiment of the present disclosure.

FIG. 3 is a flow chart of an image recognition method according to still another embodiment of the present disclosure.

FIG. 4 is a structural schematic view of an electronic device according to an embodiment of the present disclosure.

FIG. 5 is a structural schematic view of a computer-readable storage medium according to an embodiment of the present disclosure.

DETAILED DESCRIPTION

Technical solutions of the embodiments of the present disclosure will be clearly and completely described by referring to the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some, but not all, of the embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by an ordinary skilled person in the art without making creative work shall fall within the scope of the present disclosure.

As shown in FIG. 1, FIG. 1 is a flow chart of an image recognition method according to an embodiment of the present disclosure.

In an operation S11, an image to be recognized is obtained, and the image to be recognized includes a target object.

The image to be recognized is obtained, and the image to be recognized includes the target object. The target object includes a human body, a human face, a gait, a body pose, a vehicle, a license plate, an animal or other objects. The target object may be determined based on an actual image recognition object.

In an application scenario, the target object may be photographed by various types of cameras to obtain images to be recognized that include the target object. In another application scenario, the image to be recognized may be obtained by processing an initial image, for example, the processing may include heat map generation, feature extraction, processing by a deep learning model and other image processing operations.

In an operation S12, in response to an archive corresponding to the target object being present in historic archives, it may be determined whether identity information of the target object is present in the archive.

In response to the archive corresponding to the target object being present in the historical archives, it is further determined whether the identity information of the target object is present in the corresponding archive.

The archive corresponding to the target object being present in the historical archives indicates that it is not a first time that the target object is recognized. It is further determined whether the identity information of the target object is present in the archive corresponding to the target object.

The historical archives refer to an archive library obtained by archiving all images, which are obtained previously and are to be recognized, while the images are being recognized. In an application scenario, when the image recognition method is applied to recognize a person who is on street and photographed, the historical archives refer to an archive library obtained by archiving all images that are captured previously and include all pedestrians. The archive is a record combining and archiving various images to be recognized based on the target object.

In an application scenario, the archive corresponding to the target object being not present in the historical archives indicates that the target object is recognized for the first time. A new archive for the target object may be established based on the image to be recognized, and the target object may be recognized. In another application scenario, when the archive corresponding to the target object is not present in the historical archives, the target object may be identified directly, and the archive may be established subsequently, such that the identification information may be recognized in real time.

In an operation S13, in response to the identity information of the target object being present, the identity information is determined as a recognition result of the image to be recognized.

When the identity information of the target object is present in the archive corresponding to the target object, the identity information is determined as the identity information of the target object, i.e. the recognition result of the image to be recognized. In this way, the identification information of the target object may be determined directly based on the archive corresponding to the target object. A recognition process, which compares the image to be recognized with a plurality of objects in an identity information database to obtain the identity information of the target object, may be reduced to some extent, such that a recognition workload may be reduced.

In an application scenario, in response to the identity information of the target object being not present, the image to be recognized is compared with a plurality of target objects in the identity information database to obtain the identity information of the target object, obtaining the recognition result. In another application scenario, in response to the identity information of the target object being not present, a manual mark of the target object is received to obtain the identity information of the target object, obtaining the recognition result. A specific means of obtaining the identity information is not limited by the present disclosure. The above method may be applied in scenarios such as personnel statistics and big data analysis. In another application scenario, in response to the identity information of the target object being not present in the identity information database, no response may be made. The above method may be applied in scenarios such as monitoring specific persons, checking class attendance, and other scenarios of identifying specific persons.

In an operation S14, the image to be recognized and the recognition result are updated into the archive.

The image to be recognized and the recognition result of the target object are updated to the archive corresponding to the target object, such that images and the identity information of the archive may be updated, facilitating subsequent recognition of a new image to be recognized.

According to the image recognition method of the present disclosure, in response to the archive corresponding to the target object being present in the historical archives, it may be determined whether the identification information of the target object is present in the archive. In response to the identity information of the target object being present, the identity information is determined as the recognition result of the image to be recognized, such that the identity information of the target object may be determined directly based on the archive corresponding to the target object. The recognition process, which compares the image to be recognized with a plurality of target objects in the identity information database to obtain the identity information of the target object, may be reduced to some extent, such that the recognition workload may be reduced. At last, the image to be recognized and the recognition result are updated to the archive. In this way, the image and the identity information of the archive are updated, facilitating subsequent recognition of the new image to be recognized.

As shown in FIG. 2, FIG. 2 is a flow chart of an image recognition method according to another embodiment of the present disclosure.

When the image recognition method is just initiated, a predetermined number of images are obtained. The predetermined number of images are clustered based on the target object in the images, such that images having a same target object are combined into a same archive. In this way, the historical archives that include all archived archives of the target object are obtained. The images may include an image to be recognized and an image that has been recognized. A clustering method of the preset embodiment includes a K-means clustering method, a HAC hierarchical cohesive clustering method or a maximum and minimum distance clustering algorithm, and so on. The present disclosure does not limit the clustering method.

The clustering process of the present embodiment is performed in response to the number of images satisfying a predetermined number. The predetermined number may be 100, 500, 1000, and the like, which may be determined based on an actual situation.

The operation of obtaining the historical archives may be performed before or simultaneously with the operation of obtaining the images to be recognized. When the images to be recognized are obtained, the images to be recognized are recognized based on the historical archives. When the images to be recognized are obtained, and when the total number of the images to be recognized does not reach the predetermined number, that is, when the historical archives are established by clustering, the images to be recognized may be compared with a plurality of target objects in the identity information database to obtain the identity information of the target object to obtain the recognition result. When the number of the images reaches the predetermined number, and the historical archives are obtained, subsequent recognition of any image to be recognized may be performed based on the historical archives.

After establishing the historical archives, in response to the identity information of the target object in the historical archives being obtained, the identity information and the corresponding image are added to the archive of the target object to complete the operation of archiving each target object. The identity information of the target object in the historical archives may be obtained from the recognition result of the image to be recognized in the process of recognizing the image. For example, when the identity information of the target object to be recognized is obtained by comparing the image to be recognized with a plurality of target objects in the identity information database, the identity information obtained above is added directly to the archive of the target object after the number of images to be recognized and/or the number of images that have been recognized meets the predetermined number, and the historical archives are established.

In an operation S21, the image to be recognized is obtained, and the image to be recognized includes the target object.

The operation is the same as the operation S11. The present operation may be referred to the above description, and will not be repeated.

In an operation S22, in response to the archive corresponding to the target object being present in the historical archives, it may be determined whether the identity information of the target object is present in the archive.

It may be determined whether the archive corresponding to the target object of the image to be recognized is present in the historical archives. In response to the archive corresponding to the target object of the image to be recognized being present in the historical archives, it may further be determined whether the identity information of the target object is present in the archive. In response to the archive corresponding to the target object of the image to be recognized being not present in the historical archives, an operation S24 may be performed.

The operation of determining whether the archive corresponding to the target object of the image to be recognized is present in the historical archives includes: comparing the target object of the image to be recognized with partial historical archives, until determining the archive corresponding to the target object being present in the historical archives or traversing the historical archives. The historical archives are divided into a plurality of history archive portions based on a predetermined rule. Each historical archive portion includes archives of a plurality of target objects.

While comparing, the image to be recognized may be compared to each historical archive portion successively in a cascade manner based on correspondence between the image to be recognized and the plurality of historical archive portions until it is determined that the archive corresponding to the target object is present in the historical archives or the historical archives are traversed.

The predetermined rule may refer to dividing the historical archives into the plurality of historical archive portions based on an attribute of each target object in each historical archive, a location at which the image corresponding to each target object is obtained, a time point at which the image corresponding to each target object is obtained, and other attributes. In an application scenario, the present image recognition is applied for recognizing persons on streets. When three streets are present, the historical archives may be divided into three historical archive portions based on an area of each of the three streets. Each historical archive portion corresponds to images captured for a respective street and target objects captured for the respective street. While comparing, the image to be recognized may be compared to the historical archive portions corresponding to the streets successively, starting from a historical archive portion corresponding to a nearest street to a historical archive portion corresponding to a furthest street based on location information of the current image to be recognized. Regional distance constraint may be gradually expanded in a plurality of steps until comparison is successful or all the historical archive portions are compared.

The correspondence between the image to be recognized and the plurality of historical archive portions may be determined based on the rule for dividing the historical archives, such as, dividing the historical archives based on actual attributes such as a time point, a location, and the like.

In an application, appearance features of the target object, a moving track of the target object or a location at which the target object appears often have a certain rule. The historical archives may be divided based on the certain rule. In this way, relevance of the historical archive portions to the target object may be improved. In an application scenario, the method is applied for recognizing persons on streets. The target object often passes through a same street every day. Therefore, the historical archives may be divided into a plurality of historical archive portions according to an area of each street. In this way, based on location correspondence between the image to be recognized and each historical archive portion, the image to be recognized may be firstly compared to the historical archive portion nearest to the image to be recognized, and subsequently compared to other historical archive portions based on the distance in the cascade manner to determine the archive. In this way, a comparison recall rate and a comparison efficiency may be improved to some extent, such that comparison consumption caused by traversing all historical archives may be reduced. In another application scenario, as the images stored in the historical archives are generally recent images of the target object, a height, a body type, a hair style, a dressing style and other features in the images may be similar to the real target object. Therefore, the historical archives may be divided into a plurality of historical archives portions based on time periods. In this way, based on time-period correspondence between the image to be recognized and each historical archive portion, the image to be recognized may be firstly compared to a latest historical archive portion, and subsequently compared to earlier historical archive portions, in a cascade manner, to determine the archive. In this way, the comparison consumption caused by traversing the historical archives may be reduced. In another application scenario, the historical archives may be divided based on both time periods and locations. In this way, both the time-period correspondence and the locational correspondence between the image to be recognized and the historical archive portions may be considered simultaneously for cascade comparison.

Therefore, the present operation may be performed to recognize the target object based on the inherent rule of the target object. In this way, a recognition recall rate and a recognition efficiency may be improved specifically based on the inherent feature rule of the target object. The inherent rule of the target object can hardly be expressed by a standard identification photo stored in the identity information database. Therefore, after obtaining the image to be recognized, the image to be recognized may firstly be compared to the historical archive portions to determine the archive. In this way, the comparison recall rate and the comparison efficiency may be improved, such that the efficiency of recognizing the image may be improved.

In an operation S23, in response to the identity information of the target object being present, the identity information is determined as the recognition result of the image to be recognized.

In response to the identity information of the target object being present in the archive, the identity information is determined as the recognition result of the image to be recognized. In this way, the recognition result of the image to be recognized may be directly determined based on the information in the archive. In this way, a computing process of recognition comparison based on features of the target object may be avoided, reducing the recognition workload and improving the recognition efficiency.

In response to the identity information of the target object being not present in the archive, the operation S24 may be performed.

In the operation S24, in response to the archive corresponding to the target object being present in the historical archives, and the identity information of the target object being not present in the archive, or in response to the archive corresponding to the target object being not present in the historical archives, the identity of the target object is recognized based on the identity information database, and the recognition result of the image to be recognized is obtained.

When the archive corresponding to the target object is present in the historical archives, but the identity information of the target object is not present in the archive, or when the archive corresponding to the target object is not present in the historical archives, the identity of the target object may be recognized based on the identity information database, and the recognition result of the image to be recognized is obtained. The identity information database is a base library that includes a large number of reference images of the target object and the identity information corresponding to the target object, such as a standard ID photo, including an identification photo, an employee identification photo, and the like. That is, in the process of recognizing the image, when the target object is recognized, the recognition result of a newly-obtained image to be recognized corresponding to the target object may be obtained based on the archive of the target object. In this way, the process of comparing and recognizing the same target object may not be repeated, improving the efficiency of image recognition.

The operation of recognizing the identity of the target object based on the identity information database may be performed as follows. A feature vector of the target object of the image to be recognized may be extracted. The feature vector may be compared with a plurality of standard feature vectors in the identity information database to obtain a similarity value between the feature vector and each standard feature vector. At last, identity information of a standard feature vector corresponding to a maximum similarity value is determined as the identity information of the target object, such that the recognition result of the target object is obtained. As the identity information database stores the standard ID photo of each target object, the standard feature vector of each target object may be extracted based on each standard ID photo and stored for subsequent feature comparison.

The feature vector of each image may be extracted by a convolutional neural network, an hourglass network or other deep neural networks.

When the identity information corresponding to the target object is not present in the identity information database, the image to be recognized corresponding to the target object is updated to the historical archives, and manual recognition and marking performed on the target object is received to obtain the recognition result. Alternatively, the process may stand by to wait for the identity information database to be updated, or no response is made. The operations to be performed may be determined based on the application scenario at which the image recognition method is applied.

In an operation S25, in response to the predetermined number of images to be recognized being obtained, the historical archives are updated based on the predetermined number of images to be recognized and the recognition results of the images.

In the process of image recognition, when the number of images to be recognized reaches the predetermined number, the historical archives may be updated as a batch based on the predetermined number of images to be recognized and the recognition results of the images. In this way, the historical archives may be updated gradually and iteratively in the process of applying image recognition, reducing an updating consumption. The smaller the predetermined number is, the more real time the historical archives are. The greater the predetermined number is, the less the updating consumption of the historical archives is, and the higher the updating efficiency is.

The operations S21-S24 are performed immediately after each image to be recognized is obtained. The operation S25 is performed only when the number of images to be recognized reaches the predetermined number.

As each image to be recognized is recognized immediately after the image is obtained, the recognition result of each image to be recognized, i.e., identity information, is obtained. While updating the historical archives as a batch, the recognition result may be directly taken to update the archive that does not include the identity information. In this way, the process, which recognizes the target object repeatedly while archiving the image to be recognized into the historical archives for updating, may be avoided, repeated computing may be reduced, and the efficiency of updating the archive may be improved.

According to the image recognition method of the present embodiment, the recognition results are provided for image recognition based on historical archives, and the historical archives are updated as a batch based on the recognition results. In this way, the process of image recognition and the process of archive updating may be integrated as an entirety. One process may take the result of the other process to facilitate the present process to perform relevant operations, and vice versa. In this way, repeated recognition may be reduced, and the recognition efficiency may be improved.

As long as the image to be recognized is recognized in real time, the historical archives are updated and archived efficiently. Further, in the present embodiment, the image to be recognized is compared to each historical archive portion successively in a cascade manner to determine the archive of the target object of the image to be recognized based on the correspondence between the image to be recognized and the plurality of historical archive portions. Therefore, the image to be recognized is compared to the historical archive portions successively based on a portion priority, such that an efficiency and a recall rate of the archive comparison may be improved.

As shown in FIG. 3, FIG. 3 is a flow chart of an image recognition method according to still another embodiment of the present disclosure.

In an operation S301, the image to be recognized is obtained.

The image to be recognized is obtained, and the image to be recognized includes the target object.

In an application, the target object may be photographed by various types of cameras to obtain the image to be recognized that includes the target object. In another application scenario, the image to be recognized may be obtained by processing the initial image, such as the processing may include heat map generation, feature extraction, processing by a deep learning model, and other image processing operations.

In an operation S302, it may be determined whether the archive corresponding to the target object in the image to be recognized being present in the historical archives.

It may be determined whether the archive corresponding to the target object in the image to be recognized being present in the historical archives. The operation of determining whether the archive corresponding to the target object being present in the historical archives may be similar to and referred to the operation S22 as described in the above embodiment, and will not be repeated herein.

The historical archives are obtained by clustering and archiving the predetermined number of images based on the target objects in the images in response to the predetermined number of images being obtained, when the image recognition method is just applied.

When the archive corresponding to the target object of the image to be recognized is present in the historical archives, an operation S303 is performed. When the archive corresponding to the target object of the image to be recognized is not present in the historical archives, an operation S305 is performed.

In the operation S303: it is determined whether the identity information is present in the archive.

When the archive corresponding to the target object of the image to be recognized is present in the historical archives, it is further determined whether the identity information is present in the archive.

The operation of determining whether the identity information is present in the archive is the same as and referred to the operation S23 as described in the previous embodiment, and will not be repeated herein.

When the identity information corresponding to the target object of the image to be recognized is present in the archive, an operation S304 is performed. When the identity information corresponding to the target object of the image to be recognized is not present in the archive, the operation S305 is performed.

In the operation S304, the identify information is obtained.

When the identity information corresponding to the target object of the image to be recognized is present in the archive, the identity information is obtained and is taken as the recognition result of the image, such that real time image recognition may be completed.

When the image recognition method is applied for security of a certain person, a real time alarm may be generated immediately after the identity information of the target object is obtained, improving a security efficiency.

In the operation S305, it is determined whether the identity information corresponding to the target object of the image to be recognized is present in the identity information database.

When the archive corresponding to the target object of the image to be recognized is not present in the historical archives, it is difficult to achieve image recognition rapidly based on records of the historical archives. Therefore, it is determined whether the identity information database includes the identity information corresponding to the target object of the image to be recognized.

The operation of recognizing the identity of the target object through the identity information base is the same as and referred to the operation S24 as described in the previous embodiment, and will not be repeated.

In an operation S306: No response is made.

When the identity information corresponding to the target object is not present in the identity information database, the real time recognition of the image does not respond to the target object, i.e. the recognition result is unsuccessful recognition. Subsequently, the image to be recognized corresponding to the target object may be updated into the historical archives. The manual recognition and marking for the target object may be received to obtain the recognition result. Alternatively, the process may stand by, waiting for the identity information database to be updated. An actual operation to be performed may be determined based on an actual application scenario of the image recognition.

In an operation S307, in response to the predetermined number of images to be recognized being obtained, data processing is performed.

Operations S301-306 are operations related to real time recognition in the process of image recognition, and operations S307-S312 are non-real-time operations for subsequent archive processing.

When the number of newly-obtained images to be recognized reaches the predetermined number, data processing is performed for the predetermined number of newly-obtained images to be recognized.

In an application scenario, clustering may be performed on the target object in the predetermined number of newly-obtained images to be recognized, such that images, which are to be recognized and are for a same target object, may be clustered together. In this way, the images, which are to be recognized and are clustered together, are archived and processed, improving a processing efficiency. In an application scenario, the predetermined number of newly-obtained images to be recognized may further be processed based on the recognition results obtained in operations S301-306. As the operations S301-306 are firstly performed to recognize the images to be recognized in real time, while clustering the target objects in the predetermined number of images to be recognized, the recognition results and the corresponding archives are known. Therefore, the predetermined number of images to be recognized are processed based on the known recognition results and the known corresponding archives. In this way, recognizing the images to be recognized and determining the archive for the images to be recognized may not be repeated, improving a processing efficiency.

In an operation S308, in response to the identity information of the target object being known, it is determined whether the target object is a known archive based on the identity information.

In detail, the processing operation may further include determining whether the identity information of the target object in each of the images to be recognized is known. As the image to be recognized is recognized immediately after being obtained, each image to be recognized has a respective recognition result, i.e. the identity information. Therefore, it is determined whether the archive of the target object is known based on the identity information. In detail, it is searched whether the identity information has the corresponding archive. The present operation can be performed based on the determination result of the operation S302, such that comparison with the historical archives by traversing the historical archives may be reduced. In response to the identity information having the corresponding archive, the archive of the target object is known, and an operation S310 is performed. In response to the identity information not having the corresponding archive, the archive of the target object is unknown, and an operation S309 is performed.

In the operation S309, an archive is established.

The archive of the target object being unknown indicates that, in the operation S302, the archive of the target object is not found in the historical archives. In this case, the image to be recognized corresponding to the target object may be compared to the historical archives by traversing the historical archives to find the archive of the target object. In response to the archive of the target object being found in the historical archives, the operation S310 is performed. In response to the archive of the target object being not found in the historical archives, a new archive is established based on the image to be recognized and the identity information of the target object, and is stored in the historical archives.

In an operation S310, the archive is archived.

In response to the archive of the image to be recognized being known, the image to be recognized and the identity information may be combined directly into the archive, such that the computing operation for comparing to each of the historical archives may be omitted.

After the operations S309 and S310, updating the historical archives may be completed. In this way, after a next image to be recognized is obtained, the image may be recognized based on the updated historical archives, such that subsequent image recognition may be performed based on an updated result of the historical archives, improving a success rate and an efficiency of image recognition.

According to the image recognition method of the present embodiment, recognition results are provided for image recognition based on the historical archives, and the historical archives are updated as a batch based on the recognition results. In this way, the process of image recognition and the process of archive clustering and updating may be integrated, one process may take the application result of the other process to facilitate the present process to perform relevant operations, and vice versa, such that the recognition process may not be repeatedly performed, and the recognition efficiency may be improved. The images to be recognized may be recognized in real time, while the historical archives may be updated and archived efficiently.

Based on a same inventive concept, the present disclosure further provides an electronic device which is capable of implementing the image recognition method of any of the above embodiments, as shown in FIG. 4, FIG. 4 is a structural schematic view of an electronic device according to an embodiment of the present disclosure. The electronic device includes a processor 41 and a memory 42 coupled to the processor 41.

The processor 41 is configured to execute the program instructions stored in the memory 42 to implement the operations of the image recognition method of any one of the above embodiments. In an application scenario, the electronic device may include, but is not limited to, a microcomputer, a server. In addition, the electronic device may include a mobile device such as a laptop computer, a tablet computer, and the like. The present disclosure dose not limit the type of the electronic device.

In detail, the processor 41 is configured to control the processor 41 and the memory 42 to implement the operations of the image recognition method of any one of the above embodiments. The processor 41 may also be referred to as a Central Processing Unit (CPU) . The processor 41 may be an integrated circuit chip with signal processing capability. The processor 41 may also be a general purpose processor, a Digital Signal Processor (DSP) , an Application Specific Integrated Circuit (ASIC) , a Field-Programmable Gate Array (FPGA) or other programmable logic devices, a discrete gate or transistor logic device, a discrete hardware component. The general purpose processor may be a microprocessor or any conventional processor, and the like. Alternatively, the processor 41 may be implemented by an integrated circuit chip.

According to the above technical solutions, the efficiency and the recall rate of image recognition may be improved.

Based on a same inventive concept, the present disclosure further provides a computer-readable storage medium. As shown in FIG. 5, FIG. 5 is a structural schematic view of a computer-readable storage medium according to an embodiment of the present disclosure. The computer-readable storage medium 50 stores at least one program data 51. The program data 51 is configured to implement any of the above image recognition methods. In one embodiment, the computer-readable storage medium 50 includes: a universal serial bus disk, a portable hard disk, a read-only memory (ROM) , a random access memory (RAM) , a magnetic disk or an optical disk, and various other media that can store program codes.

According to various embodiments of the present disclosure, it shall be understood that the disclosed methods and apparatus may be achieved in other ways. For example, the above described device is only exemplary. For example, the modules or the units are divided based on logical functions, but may be divided by other means when actually implemented. For example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted or not implemented. On another hand, mutual coupling or direct coupling or communicative connection shown or discussed may be indirect coupling or communicative connection via some interfaces, devices or units, which may be electrical, mechanical or in other forms.

Units illustrated as separate components may or may not be physically separated. Components displayed as units may or may not be physical units. That is, the units may be located in one place or may be distributed to a plurality of network units. Some or all of these units may be selected according to practical needs to achieve the purpose of the present disclosure.

In addition, various functional units in the various embodiments of the present disclosure may be integrated in one processing unit, or the various units may be physically configured separately, or two or more units may be integrated in one unit. The above integrated units may be implemented either in the form of hardware or in the form of software functional units.

The integrated unit may be implemented by being stored in a computer-readable storage medium as a software functional unit and sold or used as an independent product. Based on the understanding, the essence of a part of the technical solution of the present disclosure which contributes to the prior art may be implemented in the form of a software product. The software product is stored in a storage medium.

The above description shows only an embodiment of the present disclosure and does not limit the scope of the present disclosure. Any equivalent structure or equivalent process transformation performed based on the specification and the accompanying drawings of the present disclosure, directly or indirectly applied in other related art, shall be equally included in the scope of the present disclosure.

Claims

An image recognition method, comprising:

obtaining an image to be recognized, wherein the image to be recognized comprises a target object;

in response to an archive corresponding to the target object being present in historic archives, determining whether identity information of the target object is present in the archive;

determining the identity information as a recognition result of the image to be recognized in response to the identity information of the target object being present; and

updating the image to be recognized and the recognition result into the archive.
The image recognition method according to claim 1, further comprising:

recognizing an identity of the target object based on an identity information database in response to the archive corresponding to the target object being present in the historical archives, and the identity information of the target object being not present in the historical archives, or the archive corresponding to the target object being not present in the historical archives; and

obtaining the recognition result of the image to be recognized.
The image recognition method according to claim 1 or 2, further comprising:

responding to a predetermined number of images being obtained;

clustering the predetermined number of images based on target objects in the predetermined number of images to combine images having a same target object into a same archive to obtain the historical archives.
The image recognition method according to claim 3, wherein after the clustering the predetermined number of images based on target objects in the predetermined number of images to combine images having a same target object into a same archive to obtain the historical archives, the method further comprises:

in response to the identity information of the target object being obtained, adding the identity information to the archive of the target object.
The image recognition method according to claim 2, wherein the recognizing an identity of the target object based on an identity information database, and the obtaining the recognition result of the image to be recognized, comprise:

extracting a feature vector of the target object of the image to be recognized;

comparing the feature vector with a plurality of standard feature vectors in the identity information database to obtain a similarity value between the feature vector and each of the plurality of standard feature vectors; and

determining identity information of a standard feature vector corresponding to a maximum similarity value as the identity information of the target object to obtain the recognition result.
The image recognition method according to claim 1, wherein the operation of in response to an archive corresponding to the target object being present in historic archives, determining whether identity information of the target object is present in the archive, comprises:

comparing the target object of the image to be recognized with a historical archive portion until determining that the archive corresponding to the target object is present in the historic archives or traversing the historic archives;

wherein the historic archives are divided into a plurality of historical archive portions based on a predetermined rule, and each historical archive portion comprises archives of a plurality of target objects.
The image recognition method according to claim 6, wherein the comparing the target object of the image to be recognized with a historical archive portion until determining that the archive corresponding to the target object is present in the historic archives or traversing the historic archives, comprises:

comparing the image to be recognized to each of the plurality of historical archive portions successively in a cascade manner based on correspondence between the image to be recognized and each of the plurality of historical archive portions, until determining that the archive corresponding to the target object is present in the historical archives or traversing the historical archives.
The image recognition method according to claim 1, wherein the updating the image to be recognized and the recognition result into the archive, comprises:

in response to a predetermined number of images to be recognized being obtained, updating the historical archives based on the predetermined number of images to be recognized and recognition results of the predetermined number of images.
An electronic device, comprising a memory and a processor coupled to the memory, wherein the processor is configured to execute program instructions stored in the memory to perform the image recognition method according to any one of claims 1 to 8.
A computer-readable storage medium, storing program instructions, wherein the program instructions, when being executed by a processor, implement the image recognition method according to any one of claims 1 to 8.