WO2023124448A1 - 对象识别方法、系统、存储介质及程序 - Google Patents

对象识别方法、系统、存储介质及程序 Download PDF

Info

Publication number
WO2023124448A1
WO2023124448A1 PCT/CN2022/126822 CN2022126822W WO2023124448A1 WO 2023124448 A1 WO2023124448 A1 WO 2023124448A1 CN 2022126822 W CN2022126822 W CN 2022126822W WO 2023124448 A1 WO2023124448 A1 WO 2023124448A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
data
feature
storage object
storage
Prior art date
Application number
PCT/CN2022/126822
Other languages
English (en)
French (fr)
Inventor
黄王爵
林佩材
吴磊
张登奎
宋威
Original Assignee
上海商汤智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 上海商汤智能科技有限公司 filed Critical 上海商汤智能科技有限公司
Publication of WO2023124448A1 publication Critical patent/WO2023124448A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions

Definitions

  • Embodiments of the present disclosure relate to but are not limited to the field of information technology, and in particular, relate to an object recognition method, system, storage medium and program.
  • An embodiment of the present disclosure provides a technical solution for object recognition.
  • An embodiment of the present disclosure provides an object recognition method, which is applied to an object recognition system.
  • the object recognition system includes a management node and a service node; the management node is used to acquire a feature database of at least one storage object; the service node is used based on The feature database identifies the object to be identified, so as to obtain the identification result of the object to be identified.
  • the object recognition system firstly determines the feature database of each storage object based on the management node; Data is stored, which in turn can improve the privacy of data storage for each warehousing object.
  • the feature database of each stored object can be directly delivered to the corresponding application node, such as a service node, instead of sending the original data of each stored object, for example: Images or fingerprints, etc., in this way, the private data of each storage object can be transmitted in the form of characteristic data, that is, data privacy is guaranteed during the transmission process.
  • the service node is used to identify the object to be identified based on the characteristic database, so as to obtain the identification result of the object to be identified. In this way, the relevant information of the object can be stored and transmitted in the form of characteristic data, and the privacy in the process of data storage and transmission in the process of object recognition can be improved.
  • An embodiment of the present disclosure provides an object recognition system.
  • the object recognition system includes: a management node and a service node, wherein: the management node is configured to obtain at least one feature database of an object in storage; the service node is configured to It is configured to identify the object to be identified based on the feature database, so as to obtain the identification result of the object to be identified.
  • An embodiment of the present disclosure provides a computer storage medium, on which computer-executable instructions are stored. After the computer-executable instructions are executed, the above object recognition method can be implemented.
  • An embodiment of the present disclosure also provides a computer program, where the computer program includes computer readable codes, and when the computer readable codes run in an electronic device, the processor of the electronic device executes the program to implement the above-mentioned The object recognition method described above.
  • Embodiments of the present disclosure provide an object recognition method, system, storage medium, and program, wherein the object recognition method is applied to an object recognition system, and the object recognition system includes a management node and a service node, wherein, first, the The management node acquires the feature database of at least one warehousing object; in this way, the original data of each warehousing object stored in advance, such as images or fingerprints, can be converted into features and stored with feature data, thereby improving the quality of each warehousing object.
  • the privacy of the data storage of the warehousing object then, use the service node to obtain the feature database of the at least one warehousing object; in this way, the feature database of each warehousing object can be directly sent to the service node instead of sending
  • the original data of each warehousing object such as images or fingerprints, etc., so that the private data of each warehousing object can be transmitted in the form of characteristic data, that is, data privacy is guaranteed during the transmission process; finally, the service node is used based on The feature database identifies the object to be identified, so as to obtain the identification result of the object to be identified. In this way, during the process of object identification, the relevant information of the object can be stored and transmitted in the form of characteristic data, and the privacy in the process of data storage and transmission during the object identification process can be improved.
  • FIG. 1 is a schematic flowchart of a first object recognition method provided by an embodiment of the present disclosure
  • FIG. 2A shows a schematic diagram of a system architecture to which an object recognition method provided by an embodiment of the present disclosure can be applied;
  • FIG. 2B is a schematic flowchart of a second object recognition method provided by an embodiment of the present disclosure
  • FIG. 2C is a schematic flowchart of determining the first data of each storage object in the object recognition method provided by the embodiment of the present disclosure
  • FIG. 2D is a schematic flowchart of determining the target characteristics of each storage object in the object recognition method provided by the embodiment of the present disclosure
  • FIG. 2E is a schematic flow diagram of obtaining the identity of each storage object in the object recognition method provided by the embodiment of the present disclosure
  • FIG. 3A is a schematic flowchart of a third object recognition method provided by an embodiment of the present disclosure.
  • FIG. 3B is a schematic flowchart of determining the identity information of the object to be identified in the object identification method provided by the embodiment of the present disclosure
  • FIG. 4 is a schematic diagram of the system architecture of an object recognition system provided by an embodiment of the present disclosure.
  • FIG. 5 is a schematic flowchart of implementing an object recognition method based on the object recognition system provided by an embodiment of the present disclosure
  • FIG. 6 is a schematic diagram of the composition of an object recognition system provided by an embodiment of the present disclosure.
  • first ⁇ second ⁇ third is only used to distinguish similar objects, and does not represent a specific ordering of objects. Understandably, “first ⁇ second ⁇ third” Where permitted, the specific order or sequencing may be interchanged such that the embodiments of the disclosure described herein can be practiced in sequences other than those illustrated or described herein.
  • Face recognition It is a biometric technology for identification based on human facial feature information.
  • a series of related technologies that use a video camera or camera to collect images or video streams containing human faces, automatically detect and track human faces in the images, and then perform facial recognition on the detected faces, usually also called portrait recognition and facial recognition. .
  • the exemplary application of the object recognition system provided by the embodiment of the present disclosure is described below.
  • the object recognition system provided by the embodiment of the present disclosure can be implemented as a notebook computer, a tablet computer, a desktop computer, a camera, a mobile device (for example, a personal computer) with an image acquisition function
  • Various types of user terminals such as digital assistants, dedicated messaging devices, portable game devices, etc., can also be implemented as servers.
  • an exemplary application when the device is implemented as a terminal or a server will be described.
  • An embodiment of the present disclosure provides an object recognition method, which is applied to an object recognition system.
  • the object recognition system includes a management node and a service node; as shown in FIG. 1 , it shows the first object recognition method provided by the embodiment of the present disclosure.
  • Step S101 using the management node to obtain at least one feature database of the storage object.
  • the object recognition system may be a face recognition system, or a fingerprint recognition system.
  • the object recognition system may include multiple processing terminals or processing nodes, and multiple processing terminals may be located in the same area, or may be located in multiple different areas at the same time.
  • the object in the object recognition system can refer to a person, a terminal, an animal, etc.
  • the object recognition system can recognize a human face, fingerprint or voice.
  • the object recognition system includes a management node and a service node; where the management node can refer to the data center of the object recognition system, that is, the information processing center, which can be used to acquire and store relevant data information of objects to be stored .
  • the management node may have one data processing terminal, or may have multiple data processing terminals.
  • the warehousing object is the object that needs to store relevant information in advance in the database corresponding to the object recognition system.
  • the service node may refer to the application center of the object recognition system, that is, the service node may be used to realize the process of identifying related objects based on the feature database of at least one stored object.
  • the management node may include a data node and a feature node; wherein, the data node may be used to obtain the original data of each storage object, such as: the original image, original fingerprint information, and original sound information of each storage object , and at the same time, the feature node can be used to extract the features of the original data input by each storage object, so as to obtain the feature database of each storage object.
  • the number of features in the feature database of each warehousing object can be multiple; at the same time, the feature data in the feature database of each warehousing object can be the feature value corresponding to the original data of each warehousing object; wherein, the Eigenvalues can be represented using vectors.
  • the number of storage objects in the object recognition system may be one, or two or more, and when the number of storage objects is two or more, each two storage objects correspond to The feature data in the feature database for is different.
  • the storage objects are all the teachers and students of a certain school, correspondingly, the feature data in the feature database of the storage objects may be the feature data obtained after feature extraction is performed on the image information corresponding to a certain teacher and student;
  • the database objects are all employees of a certain company, and correspondingly, the feature data in the feature database of the stored objects may be the feature data obtained after feature extraction is performed on the image information corresponding to a certain employee.
  • Step S102 using the service node to identify the object to be identified based on the feature database, so as to obtain an identification result of the object to be identified.
  • the feature database of at least one warehousing object can be sent to the service node in the object recognition system, and then the service node can be used to identify the object to be identified based on the feature database, so as to obtain The recognition result of the recognized object.
  • the service node can be used to identify the object to be identified based on the feature database, so as to obtain The recognition result of the recognized object.
  • the service nodes when there are multiple service nodes, can be divided based on the attribute information of the service nodes, such as the interface attributes of the service nodes, such as: service nodes for performing attendance services, for performing Service nodes that monitor business, and each type of service node can include multiple sub-service nodes deployed in different regions.
  • the service node may include a plurality of application nodes with different attributes, and each application node is associated with a plurality of different terminal nodes; wherein, each terminal node is used to collect the data to be recognized input by the object to be recognized, and based on At least one feature database of the stored object is compared with the features of the data to be recognized to obtain the recognition result of the object to be recognized.
  • the object recognition system may use a service node to perform data recognition on the collected data to be recognized input by the object to be recognized based on the feature database, so as to obtain a recognition result of the object to be recognized.
  • the recognition result of the object to be recognized can be used to represent whether the object to be recognized belongs to one of the storage objects.
  • the service node can be an application node corresponding to a certain company's attendance system, which is used to identify the object to be identified entering the management area of the application node based on the pre-stored feature database of each employee of the company recognition to obtain the recognition result of the object to be recognized.
  • the recognition result is characterized as passing, it means that the object to be recognized is an employee of the company, and if the recognition result is represented as failing, it means that the object to be recognized is not an employee of the company.
  • the object recognition system determines the feature database of each storage object based on the management node; in this way, the original data of each storage object stored in advance, such as images or fingerprints, can be converted into features, Stored with feature data, which in turn can improve the privacy of data storage for each warehousing object.
  • the feature database of each stored object can be directly delivered to the corresponding application node, instead of the original data of each stored object, such as images or fingerprints, etc. In this way, the private data of each storage object can be transmitted in the form of characteristic data, that is, the data privacy in the transmission process can be improved.
  • the object recognition method is applied to an object recognition system, the object recognition system includes a management node and a service node, wherein the management node is used to obtain a feature database of at least one storage object, Then, the service node is used to obtain the feature database of the at least one storage object; finally, the service node is used to identify the object to be recognized based on the feature database, so as to obtain a recognition result of the object to be recognized.
  • the relevant information of the object can be stored and transmitted in the form of characteristic data, and the privacy in the process of data storage and transmission in the process of object recognition can be improved.
  • FIG. 2A shows a schematic diagram of a system architecture to which the object recognition method provided by an embodiment of the present disclosure can be applied; as shown in FIG. 2A , the system architecture includes: a management node 201 , a network 202 and a service node 203 .
  • the management terminal 201 and the service node 203 establish a communication connection through the network 202, and the management node 201 reports to the service node 203 through the network 202 the feature database of at least one storage object acquired by it; at the same time, the service node 203 Based on the feature database, the object to be identified is identified to obtain an identification result of the object to be identified.
  • the service node 203 can also upload the recognition result to the network 202 and send it to the management node 201 through the network 202 .
  • the management node when the management node includes a data node and a feature node, the data node is used to obtain the first data and identity of each storage object, and the feature node is used to obtain the first data of each storage object Perform feature extraction to obtain the target features of each warehousing object, and then determine the feature database of each warehousing object based on the target features and identity of each warehousing object; in this way, the correlation of each warehousing object can be
  • the data storage is converted into corresponding characteristic data in advance, so that the privacy of the relevant data of the storage object can be improved, that is, the above step S101 can be realized through the following steps.
  • FIG. 2B FIG. 2 is a flow realization diagram of the second object recognition method provided by the embodiment of the present disclosure; the following description is made in conjunction with the steps shown in FIG. 1 and FIG. 2B:
  • Step S201 using the data node to obtain the first data of each storage object in the at least one storage object.
  • a data node is used to obtain the first data of each storage object in at least one storage object; wherein, the first data may be base map data, fingerprint data or sound information of each storage object.
  • the data node can be used to collect the second data input by each storage object, and perform quality filtering on the second data to determine the first data of each storage object, that is, the first data is the data input to the storage object node and the data obtained after information filtering.
  • the quality inspection is first performed on the acquired second data of each storage object to obtain the corresponding detection result, and then the second data of each storage object is filtered based on the corresponding detection result , to obtain the first data of each warehousing object; thus, the quality of the obtained first data of each warehousing object can be improved; that is, the above step S201 can be realized through the following steps S2011 to S2013, as shown in Figure 2C , is a schematic flowchart of determining the first data of each storage object in the object recognition method provided by the embodiment of the present disclosure, wherein:
  • Step S2011 using the data node to obtain the second data of each storage object.
  • the data node can be used to collect and obtain the second data of each storage object, wherein the second data can be the original image, original fingerprint, original sound information, etc. of each storage object input to the data node .
  • the second data may include but not limited to: face image information, body image information, fingerprint information, voice information, etc. of the storage object, and the storage form of the second data may be pictures, audio, video, etc.
  • Step S2012 using the feature node to perform quality inspection on the second data of each storage object, and obtain the inspection result corresponding to the second data.
  • the obtained second data of each storage object can be transmitted to the feature node, so that the feature node can perform quality inspection on the second data of each storage object, that is, the feature node can be used to check the received
  • the quality inspection is performed on the second data of each storage object, and the inspection result corresponding to the second data is obtained; wherein, the inspection result corresponding to the second data may be the inspection passed or the inspection failed.
  • the face in the second data is blurred, there are multiple faces, the size of the image corresponding to the second data is too large or too small, and the person If there are many face occlusions, etc., correspondingly, the detection result corresponding to the second data is not passed; at the same time, the face definition in the second data is high and there is only one face, correspondingly, the detection result corresponding to the second data is for pass.
  • Step S2013 using the data node to filter the second data of each storage object based on the detection result, to obtain the first data of each storage object.
  • the detection result corresponding to the second data can be fed back to the data node, and the data node is used to filter the second data of each storage object based on the detection result, so as to obtain the The first data of ; wherein, the data in the first data of each storage object is the data that meets the requirements, that is, the data that passes the quality inspection.
  • the first data of each warehousing object may include a plurality of data, for example, may include the face image, fingerprint information and voice information of each warehousing object at the same time.
  • Step S202 using the feature node to perform feature extraction on the first data of each storage object, to obtain the target feature of each storage object.
  • the first data of each storage object obtained after quality filtering is transmitted from the data node to the feature node, and the feature node is used to perform feature extraction on the first data of each storage object, so as to Obtain the target feature of each stored object; for example, when the first data is a face image of the stored object, the target feature of each stored object is the face feature value of each stored object.
  • the target feature of each storage object is the sound feature of each storage object.
  • step S202 can be realized through the following steps S2021 and S2022, as shown in Figure 2D, which is a schematic flow chart of determining the target characteristics of each storage object in the object recognition method provided by the embodiment of the present disclosure, wherein:
  • Step S2021 using the data nodes to determine a target extraction model.
  • data nodes may be used to determine the algorithm model used for feature extraction, that is, the target extraction model.
  • the target extraction model can be determined by the data node based on the feature extraction model set currently stored in the feature node; in this way, the appropriate target extraction model can be selected through the feature extraction model set stored in the feature node. model, which can improve the accuracy and efficiency of the determined target lifting model; that is, the above step S2021 can be achieved through the following steps:
  • the feature node in response to the query request sent by the data node, the feature node is used to count the feature extraction models stored in the feature node to obtain a list of feature extraction models, and the list of feature extraction models is sent to the data node.
  • the data node initiates a query request to the feature node, and the feature node responds to the query request to collect statistics on the feature extraction models stored in the feature node to obtain a list of feature extraction models.
  • the feature extraction model list includes at least one feature extraction model, and the feature extraction model list may be list information corresponding to a set of available feature extraction models internally stored in the feature node.
  • the feature node is used to send the statistically obtained feature extraction model list to the data node, so that the data node knows the feature extraction models that can be used in the feature node.
  • the feature extraction model stored inside the feature node may be uploaded to the feature node by the data node in advance; in this way, relevant feature extraction operations can be performed based on the feature extraction model stored in advance, thereby improving the efficiency of extracting target features. That is, after executing the above "in response to the query request sent by the data node, use the feature node to count the feature extraction models stored in the feature node to obtain a list of feature extraction models, and send the list of feature extraction models to the Data Node", you can also perform the following procedures:
  • the data node is used to send the feature extraction model associated with the service node to the feature node.
  • the feature node is used to store the feature extraction model associated with the service node inside the feature node.
  • the data node in the object recognition system may obtain the feature extraction model associated with the service node in advance, and upload the feature extraction model to the feature node, so that the feature node internally stores the feature extraction model.
  • the feature extraction model associated with the service node may be a feature extraction model associated with a terminal node corresponding to the service node, that is, an available feature extraction model stored inside the terminal node.
  • the target extraction model is determined from the feature extraction model list by using the data node.
  • the data node can be used to determine the target extraction model from the feature extraction model list, so that the feature node can perform feature extraction on the first data of each storage object based on the determined target extraction model, and obtain each storage object The target feature corresponding to the object.
  • Step S2022 using the feature node to perform feature extraction on the first data of each storage object based on the target extraction model, to obtain the target feature of each storage object.
  • the feature node performs feature extraction on the first data of each storage object based on the target extraction model specified by the data node, to obtain the target feature of each storage object.
  • Step S203 using the data node to obtain the identity of each storage object.
  • the object recognition system can use the data node to obtain the identity of each stored object; wherein, the identity can be used to characterize the unique identifier of each stored object.
  • the identity of each storage object can be determined based on the identity information set of each storage object; in this way, the efficiency and accuracy of determining the identity of each storage object can be improved That is, the above step S203 can be realized through the following steps S2031 to S2033, as shown in FIG. 2E , which is a schematic flow diagram of obtaining the identity of each storage object in the object recognition method provided by the embodiment of the present disclosure, wherein:
  • Step S2031 using the data node to obtain the identity information set of each storage object.
  • the data node can be used to obtain the identity information set input by each storage object, wherein the identity information set of each storage object may include but not limited to: the name, ID card of each storage object Number, job number, student number, etc.
  • the second data and identity information set of each warehouse-in object can be input to the data node at the same time by each warehouse-in object.
  • Step S2032 using the service node to generate the identity of each stored object based on the identity information set of each stored object.
  • the identity information set of each incoming object can be delivered to the service node, so that the service node can save the identity information set of each incoming object, and at the same time, the service node can also base on the The identity information set generates the identity of each storage object.
  • Step S2033 using the service node to send the identity of each storage object to the data node.
  • the service node can send the identity of each stored object to the data node, so that the data node can modify, delete or add information to the relevant stored object based on the identity.
  • Step S204 using the data node to determine the feature database of the at least one object based on the target feature of each stored object and the identity of each stored object.
  • the target feature corresponding to each storage object is associated with the identity, so as to determine at least one storage object.
  • a feature database for each incoming object in the library object is associated with the identity, so as to determine at least one storage object.
  • the data node is used to obtain the identity information set and the second data of each stored object
  • the feature node is used to perform information filtering and feature extraction on the second data of each stored object to obtain The target characteristics of the library object, and use the service node to identify the identity information set of each object in the library to obtain the identity of each object in the library, and then use the data node to identify the object based on the identity and target of each object in the library feature, get the feature database of each object in storage. In this way, the accuracy and efficiency of the resulting feature database can be improved.
  • FIG. 3 is a flow realization diagram of the third object recognition method provided by the embodiment of the present disclosure; the following description is made in conjunction with the steps shown in FIG. 1 and FIG. 3A:
  • Step S301 using the at least one application node to determine the target feature database of the target object associated with the attribute information of the application node based on the feature database.
  • the object recognition system may determine the target feature database of the target object associated with the attribute information of the application node from the feature database of at least one incoming object based on the attribute information of different application nodes.
  • the target object associated with the attribute information of the application node may be the feature database of all students living in the school, that is, the target feature database;
  • the target characteristic database of the target object associated with the attribute information of the application node is determined, which may be the characteristic database of all teachers living in the school.
  • Step S302 using the at least one terminal node to identify the object to be identified based on the target feature database, and determine the identity information of the object to be identified.
  • the target feature database is acquired by using a terminal node associated with the application node, and after the target feature database is acquired, the object to be identified is identified to determine the identity information of the object to be identified.
  • the feature extraction model associated with the terminal node can be used to compare the features of the collected data to be identified based on the target feature data set library to determine the identity information of the object to be identified, that is It can directly perform feature comparison based on the acquired feature data and the collected to-be-identified data to obtain the identity information of the to-be-identified object.
  • the identity verification can be realized based on the characteristic data of the relevant object, and the privacy of the relevant information of the relevant object can be guaranteed, that is, the above step S302 can be realized through the following steps S3021 to S3023, as shown in FIG. 3B , which is implemented by the present disclosure.
  • a schematic flow chart of determining the identity information of the object to be identified in the object identification method provided by the example, wherein:
  • Step S3021 using the terminal node based on the feature extraction model associated with the terminal node to perform feature extraction on the collected data to be identified input by the object to be identified to obtain feature data to be compared.
  • the feature extraction model associated with the terminal node is used to perform feature extraction on the collected data to be identified input by the object to be identified to obtain feature data to be compared.
  • the data form of the feature data to be compared is the same as the data form of the feature data in the feature database of at least one storage object acquired by the management node.
  • Step S3022 using the terminal node to search for feature data similar to the feature data to be compared from the target feature database, and obtain a search result.
  • the terminal node is used to search the feature data similar to the feature data to be compared from the target feature database to obtain the search result; wherein, the search result can be represented by not found and found, or It can be represented by 0 and 1.
  • Step S3023 using the terminal node to determine the identity information of the object to be identified based on the search result.
  • the terminal node is used to determine the identity information of the object to be identified based on the search result, wherein, if the search result indicates that it is not found, it indicates that the identity information of the object to be identified cannot be identified, and the "error identity information” or “incorrect identity information” or “unrecognizable identity information”. If the search result indicates that it has been found, then use the identity corresponding to the found feature data to represent the object to be identified.
  • the terminal node can be used to upload the obtained identity information to the application node associated with the terminal node, that is, to realize timely uploading of the identification result.
  • the data to be identified input by the object to be identified can be deleted, which can improve the privacy of the relevant information of the object to be identified, that is, in the execution step After S3023, the object recognition method provided by the embodiment of the present disclosure may also perform the following process:
  • the terminal node determines the identity information of the object to be identified
  • the data to be identified is deleted.
  • the data to be identified inputted by the object to be identified to the terminal node is deleted, that is, the data to be identified is not saved. In this way, the privacy of the relevant information of the object to be identified can be improved.
  • Face recognition is a biometric technology for identity recognition based on human facial feature information.
  • the basic process of face recognition is: face detection, living body detection, face tracking, feature extraction, face comparison, output Results etc. Thanks to its advantages of non-contact, naturalness, and anti-counterfeiting, face recognition technology is very suitable for business scenarios that require frequent personnel identity verification, such as access control, check-in, and sign-in.
  • the common face recognition system composition structure is: face recognition terminal + management platform.
  • the face recognition terminal is mainly responsible for image acquisition, liveness detection, feature extraction, and feature comparison, and performs linkage operations (such as opening doors, etc.) according to the face recognition results;
  • the management platform provides users with face information management, face information download Send, record query and other functions.
  • Encrypted storage use a strong encryption algorithm to encrypt and store face information on the face recognition terminal and management platform;
  • the management platform can use encryption protocols such as Secure Sockets Layer (Secure Sockets Layer, SSL), Advanced Encryption Standard (Advanced Encryption Standard, AES) to realize Transport Layer Security (TLS) encryption.
  • SSL Secure Sockets Layer
  • AES Advanced Encryption Standard
  • TLS Transport Layer Security
  • the face recognition terminal can use the Hyper Text Transfer Protocol over Secure Socket Layer (HTTPS) interface to transmit data to the server to ensure that the entire link of the transmitted data is in an encrypted state.
  • HTTPS Hyper Text Transfer Protocol over Secure Socket Layer
  • Access control Provide permission setting function, and open face information access permission according to the minimum necessary principle.
  • Display restrictions De-identify face information, for example, face recognition terminals and/or management interfaces perform mosaic processing on displayed faces.
  • face information will still be stored in the face recognition terminal and management platform, and face information will be transmitted between the management platform and face recognition terminal.
  • face recognition system that is, the following requirements:
  • Transmission security During the communication process between the terminal of the face recognition system and the management platform, the face image is not directly transmitted.
  • the current common face recognition system privacy protection scheme has the following deficiencies:
  • the face image is still stored in the face recognition terminal and management platform, and there is still a risk of being leaked and cracked;
  • face images will be transmitted between the face recognition terminal and the management platform, which may be intercepted and cracked, resulting in leakage of private data.
  • an embodiment of the present disclosure provides an object recognition system, which is composed of "data nodes, feature nodes, application nodes, and terminal devices".
  • an object recognition system which is composed of "data nodes, feature nodes, application nodes, and terminal devices".
  • the minimum storage and transmission of face images in the entire face recognition system can be realized.
  • Figure 4 it is a schematic diagram of the system architecture of an object recognition system provided by an embodiment of the present disclosure; wherein, 401 is a data node, 402 is a feature node, 403 is a plurality of application nodes with different attributes, and 404 is an application node Multiple terminal devices corresponding to A. Its main technical implementations are:
  • Data node responsible for storing the user's original face base map data and basic information; and there is usually only one data node in a system architecture, and it is deployed in a trusted environment.
  • Feature node responsible for receiving the original face base map of the data node and converting it into a feature value that matches the face recognition terminal algorithm model. It does not store the original face image.
  • the number of feature nodes in the system structure can be one or more, with feature extraction capabilities of various algorithm models, and deployed in a trusted environment.
  • Application node responsible for receiving feature values and basic user information, and sending feature values and basic user information to designated face recognition terminals, which do not store original face images. Multiple sets of application nodes can be deployed according to actual needs (for example: face access control system, face control system, face attendance system, etc.), and can be deployed in an untrusted environment.
  • Terminal equipment the face recognition terminal responsible for on-site face image collection.
  • the terminal device locally uses facial features for identity authentication, and immediately deletes the original snapshot image after the authentication is completed.
  • the terminal device does not store or upload the original face base map and snapshot images.
  • FIG. 5 it is a schematic flow diagram of implementing an object recognition method based on the object recognition system provided by the embodiment of the present disclosure; it is mainly divided into five major steps: quality Detection, feature extraction, personnel warehousing, personnel distribution and identity authentication; among them:
  • the data node transmits the face image to the feature node, and the feature node performs quality inspection on the face image to obtain the corresponding quality inspection result; then the feature node will pass the quality inspection The result is fed back to the data node; the data node will be based on the quality detection result to the unqualified face picture (for example: no face, multiple faces, at least one of the size is too large and too small, the face is more occluded etc.) to filter and reject.
  • the unqualified face picture for example: no face, multiple faces, at least one of the size is too large and too small, the face is more occluded etc.
  • Feature extraction first upload the algorithm model, that is, execute step 504: the data node extracts the algorithm model that is consistent with the terminal device, and uploads it to the feature node; the feature node saves the algorithm model, that is, executes step 505 and feeds back the model upload result 506; wherein, the algorithm model is an algorithm model used by the feature node to extract features from the face picture.
  • the data node queries the algorithm model version: the data node queries the available algorithm model version from the feature node, and the feature node responds to the query model version request sent by the data node, and returns the list of available algorithm model versions to the data node; that is, the execution step 507 to step 509.
  • the data node can specify one or more available algorithm model versions based on the algorithm model version list, and pass the face pictures that meet the quality requirements of the storage to the feature node.
  • the feature node extracts the features of the face picture and returns the corresponding
  • the face feature of the data node saves the face feature; that is, step 510 to step 512 are executed.
  • Personnel warehousing first, basic information warehousing: the data node transmits the basic information of personnel (such as name, job number, student number, card number, ID number, etc.) to the application node, and the application node saves the personnel information and returns The corresponding personnel unique identification (this identification can be used for subsequent personnel information change/query/deletion); secondly, feature storage: the data node transfers the internally stored personnel unique identification, algorithm model version, and face features to the application node, and the application The node saves the relationship between the person and the facial features, that is, steps 513 to 515 are executed.
  • basic information warehousing the data node transmits the basic information of personnel (such as name, job number, student number, card number, ID number, etc.) to the application node, and the application node saves the personnel information and returns The corresponding personnel unique identification (this identification can be used for subsequent personnel information change/query/deletion);
  • feature storage the data node transfers the internally stored personnel unique identification, algorithm model version
  • Personnel distribution that is, the data node distributes the faces of different personnel to different devices by calling the personnel distribution interface of the application node (such as creating a new group, adding personnel to the group, and distributing the group to the device, etc.).
  • Feature value that is, execute steps 516 to 519: based on the different attribute information of the application nodes, send the face features of different people to the corresponding application nodes.
  • the face recognition terminal collects the face image, that is, executes step 520: the person scans the face, and after the person swipes the face, extracts the corresponding face features to be recognized and compares them with the local feature database, and completes the personnel identity authentication Upload the recognition result record to the application node; wherein, the local feature database is the personnel issued in step 519 .
  • the face recognition terminal does not store any face base map or face snapshot, and does not upload the face snapshot to the application node.
  • the "dual node" scheme that separates the feature node and the application node can be adopted, and the minimum storage of face image data in an untrusted environment can be realized. That is, it can be achieved:
  • Transmission security During the communication process between the terminal of the face recognition system and the management platform, the face image is not directly transmitted.
  • data nodes can be connected to multiple sets of face recognition application systems, maintain the face feature relationship required by different systems, and send face features to different terminal devices under the premise of ensuring privacy and security .
  • the object recognition method and object recognition system provided by the embodiments of the present disclosure can be applied to the deployment case of campus multi-face application system; wherein, in the campus scene, students’ personal information is usually managed by the The system is uniformly maintained by the corresponding administrative agencies.
  • the dormitory facial recognition access control system is managed and maintained by the logistics department
  • the campus facial recognition monitoring system is managed and maintained by the security department
  • the student facial recognition attendance system is managed and maintained by the academic affairs office; to ensure student information security
  • These application systems can use the object recognition system provided by the embodiments of the present disclosure for unified deployment and management, that is, the information center serves as a unified data center, deploying feature nodes in a trusted environment, converting the original image of the face into a feature value, and issuing them respectively
  • it can meet the normal use of various systems without leaking the original face data of students.
  • FIG. 6 is a schematic diagram of the structure and composition of the object recognition system provided by the embodiment of the present disclosure. As shown in FIG. 6 , the object recognition system 600 includes:
  • the management node 601 is configured to obtain at least one feature database of the storage object
  • the service node 602 is configured to identify the object to be identified based on the feature database, so as to obtain an identification result of the object to be identified.
  • the management node includes a data node and a feature node, and in the object recognition system 600, the data node is configured to acquire the first One data; the feature node is configured to perform feature extraction on the first data of each warehousing object to obtain the target feature of each warehousing object; the data node is also configured to obtain the The identity of each storage-in object; the data node is further configured to determine the at least one storage-in object based on the target feature of each storage-in object and the identity of each storage-in object feature database.
  • the data node is further configured to acquire the second data of each storage object; the feature node is further configured to obtain each of the Performing quality inspection on the second data of a storage object, and obtaining a detection result corresponding to the second data; the data node is further configured to perform a quality inspection on the second data of each storage object based on the detection result Filter to obtain the first data of each storage object.
  • the data node is further configured to determine a target extraction model; Feature extraction is performed on the first data of the object to obtain the target feature of each object in storage.
  • the data node is further configured to, in response to the query request sent by the data node, count the feature extraction models stored inside the feature node to obtain the feature extraction model list, and send the list of feature extraction models to the data node; the data node is further configured to determine the target extraction model from the list of feature extraction models.
  • the data node is further configured to send the feature extraction model associated with the service node to the feature node; the feature node is also configured to The feature extraction model associated with the service node is stored in the feature node.
  • the data node is further configured to obtain the identity information set of each object stored in the warehouse; the service node is further configured to The identity information set of the stored object generates the identity of each stored object; the service node is further configured to send the identity of each stored object to the data node.
  • the service node includes at least one application node and at least one terminal node associated with the application node, and in the object recognition system 600, the at least one application node is configured to be based on the A feature database, for determining a target feature database of a target object associated with the attribute information of the application node; the at least one terminal node is configured to identify the object to be identified based on the target feature database, and determine the target object to be identified Identity information that identifies an object.
  • the terminal node is further configured to, based on the feature extraction model associated with the terminal node, input the collected data to be recognized by the object to be recognized Perform feature extraction to obtain feature data to be compared; the terminal node is also configured to search for feature data similar to the feature data to be compared from the target feature database to obtain a search result; the terminal node is also configured It is configured to determine the identity information of the object to be identified based on the search result.
  • the terminal node is further configured to delete the data to be identified after determining the identity information of the object to be identified.
  • the above-mentioned object recognition method is implemented in the form of software function modules and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
  • the essence of the technical solutions of the embodiments of the present disclosure or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for A computer device (which may be a terminal, a server, etc.) is made to execute all or part of the methods described in various embodiments of the present disclosure.
  • the aforementioned storage media include: various media that can store program codes such as U disk, sports hard disk, read-only memory (Read Only Memory, ROM), magnetic disk or optical disk.
  • embodiments of the present disclosure are not limited to any specific combination of hardware and software.
  • the embodiments of the present disclosure further provide a computer program product, the computer program product includes computer executable instructions, and after the computer executable instructions are executed, the object recognition method provided by the embodiments of the present disclosure can be implemented.
  • an embodiment of the present disclosure further provides a computer storage medium, on which computer executable instructions are stored, and when the computer executable instructions are executed by a processor, the object recognition method provided in the above embodiments is implemented.
  • the computer storage medium may be a volatile storage medium or a non-volatile storage medium.
  • An embodiment of the present disclosure provides a computer program, the computer program includes computer-readable codes, and when the computer-readable codes run in an electronic device, the processor of the electronic device executes the program to implement the above-mentioned embodiments Provided object recognition methods.
  • the disclosed devices and methods may be implemented in other ways.
  • the device embodiments described above are only illustrative.
  • the division of the units is only a logical function division.
  • the coupling, or direct coupling, or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be electrical, mechanical or other forms of.
  • the units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed to multiple network units; Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
  • all the functional units in the embodiments of the present disclosure may be integrated into one processing unit, each unit may be used as a single unit, or two or more units may be integrated into one unit;
  • the above-mentioned integrated units can be implemented in the form of hardware, or in the form of hardware plus software functional units.
  • Those of ordinary skill in the art can understand that all or part of the steps to realize the above method embodiments can be completed by hardware related to program instructions, and the aforementioned programs can be stored in computer-readable storage media.
  • the execution includes: The steps in the foregoing method embodiments; and the aforementioned storage medium includes: various media capable of storing program codes such as removable storage devices, ROMs, magnetic disks, or optical disks.
  • the above-mentioned integrated units in the embodiments of the present disclosure are implemented in the form of software function modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.
  • a software product which is stored in a storage medium and includes several instructions for Make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the methods described in the various embodiments of the embodiments of the present disclosure.
  • the aforementioned storage medium includes various media capable of storing program codes such as removable storage devices, ROMs, magnetic disks or optical disks.
  • Embodiments of the present disclosure provide an object recognition method, system, storage medium, and program; wherein, the object recognition method is applied to an object recognition system, and the object recognition system includes a management node and a service node; using the management node Obtaining a feature database of at least one stored object; using the service node to identify the object to be identified based on the feature database, so as to obtain a recognition result of the object to be identified.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Image Analysis (AREA)

Abstract

本公开实施例提供了一种对象识别方法、系统、存储介质及程序；其中，所述对象识别方法，应用于对象识别系统，所述对象识别系统包括管理节点和服务节点；采用所述管理节点获取至少一个入库对象的特征数据库；采用所述服务节点基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。

Description

对象识别方法、系统、存储介质及程序
相关申请的交叉引用
本公开要求2021年12月27日提交的中国专利申请号为202111619572.4、申请人为深圳市商汤科技有限公司，申请名称为“对象识别方法、系统及存储介质”的优先权，该申请的全文以引用的方式并入本公开中。
技术领域
本公开实施例涉及但不限于信息技术领域，尤其涉及一种对象识别方法、系统、存储介质及程序。
背景技术
相关技术中，通常需将事先存储的人脸图像传输至人脸识别终端，并在人脸识别终端采集到待识别人脸图像的情况下，基于事先存储的人脸图像以实现人脸识别；这样，无法保证事先存储的人脸图像的隐私性。
发明内容
本公开实施例提供一种对象识别技术方案。
本公开实施例的技术方案是这样实现的：
本公开实施例提供一种对象识别方法，应用于对象识别系统，所述对象识别系统包括管理节点和服务节点；采用所述管理节点获取至少一个入库对象的特征数据库；采用所述服务节点基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。
这样，对象识别系统，首先基于管理节点，确定每一入库对象的特征数据库；如此，能够实现将事先存储的每一入库对象的原始数据，比如：图像或指纹等进行特征转换，以特征数据进行存储，进而能够提高每一入库对象的数据存储的隐私性。同时在基于该特征数据库进行相关应用的过程中，可以直接下发每一入库对象的特征数据库至对应的应用节点，如服务节点，而非下发每一入库对象的原始数据，比如：图像或指纹等，如此，能够实现每一入库对象的隐私数据以特征数据形式传输，即在传输过程保证数据隐私性。然后，采用服务节点基于该特征数据库，对待识别对象进行识别，以得到待识别对象的识别结果。如此，能够实现以特征数据的形式存储和传输对象的相关信息，进而能够提高对象识别过程中数据存储和传输过程中的隐私性。
本公开实施例提供一种对象识别系统，所述对象识别系统包括：管理节点和服务节点，其中：所述管理节点，被配置为获取至少一个入库对象的特征数据库；所述服务节点，被配置为基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。
本公开实施例提供一种计算机存储介质，所述计算机存储介质上存储有计算机可执行指令，所述计算机可执行指令被执行后，能够实现上述的对象识别方法。
本公开实施例还提供一种计算机程序，所述计算机程序包括计算机可读代码，在所述计算机可读代码在电子设备中运行的情况下，所述电子设备的处理器执行用于实现上述所述的对象识别方法。
本公开实施例提供一种对象识别方法、系统、存储介质及程序，其中，所述对象识 别方法应用于对象识别系统，所述对象识别系统包括管理节点和服务节点，其中，首先，采用所述管理节点获取至少一个入库对象的特征数据库；如此，能够实现将事先存储的每一入库对象的原始数据，比如：图像或指纹等进行特征转换，以特征数据进行存储，进而能够提高每一入库对象的数据存储的隐私性；然后，采用所述服务节点获取该至少一个入库对象的特征数据库；如此，可以直接下发每一入库对象的特征数据库至服务节点，而非下发每一入库对象的原始数据，比如：图像或指纹等，这样能够实现每一入库对象的隐私数据以特征数据形式传输，即在传输过程保证数据隐私性；最后，采用所述服务节点基于该特征数据库，对待识别对象进行识别，以得到待识别对象的识别结果。如此，能够在对象进行识别的过程中，实现以特征数据的形式存储和传输对象的相关信息，进而能够提高对象识别过程中数据存储和传输过程中的隐私性。
为使本公开的上述目的、特征和优点能更明显易懂，下文特举较佳实施例，并配合所附附图，作详细说明如下。
附图说明
为了更清楚地说明本公开实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本公开实施例的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的附图，其中：
图1为本公开实施例提供的第一种对象识别方法的流程示意图；
图2A示出可应用本公开实施例所提供的对象识别方法的一种系统架构示意图；
图2B为本公开实施例提供的第二种对象识别方法的流程示意图；
图2C为本公开实施例提供的对象识别方法中确定每一入库对象的第一数据的流程示意图；
图2D为本公开实施例提供的对象识别方法中确定每一入库对象的目标特征的流程示意图；
图2E为本公开实施例提供的对象识别方法中获取每一入库对象的身份标识的流程示意图；
图3A为本公开实施例提供的第三种对象识别方法的流程示意图；
图3B为本公开实施例提供的对象识别方法中确定待识别对象的身份信息的流程示意图；
图4为应用本公开实施例提供的一种对象识别系统的系统架构示意图；
图5为基于本公开实施例提供的对象识别系统来实现对象识别方法的一种流程示意图；
图6为本公开实施例提供的一种对象识别系统的组成示意图。
具体实施方式
为使本公开实施例的目的、技术方案和优点更加清楚，下面将结合本公开实施例中的附图，对发明的具体技术方案做进一步详细描述。以下实施例用于说明本公开实施例，但不用来限制本公开实施例的范围。
在以下的描述中，涉及到“一些实施例”，其描述了所有可能实施例的子集，但是可以理解，“一些实施例”可以是所有可能实施例的相同子集或不同子集，并且可以在不冲突的情况下相互结合。
在以下的描述中，所涉及的术语“第一\第二\第三”仅仅是是区别类似的对象，不代表针对对象的特定排序，可以理解地，“第一\第二\第三”在允许的情况下可以互换特定 的顺序或先后次序，以使这里描述的本公开实施例能够以除了在这里图示或描述的以外的顺序实施。
除非另有定义，本文所使用的所有的技术和科学术语与属于本公开实施例的技术领域的技术人员通常理解的含义相同。本文中所使用的术语只是为了描述本公开实施例的目的，不是旨在限制本公开实施例。
对本公开实施例进行进一步详细说明之前，对本公开实施例中涉及的名词和术语进行说明，本公开实施例中涉及的名词和术语适用于如下的解释。
人脸识别：是基于人的脸部特征信息进行身份识别的一种生物识别技术。用摄像机或摄像头采集含有人脸的图像或视频流，并自动在图像中检测和跟踪人脸，进而对检测到的人脸进行脸部识别的一系列相关技术，通常也叫做人像识别、面部识别。
下面说明本公开实施例提供的对象识别系统的示例性应用，本公开实施例提供的对象识别系统可以实施为具有图像采集功能的笔记本电脑，平板电脑，台式计算机，相机，移动设备(例如，个人数字助理，专用消息设备，便携式游戏设备)等各种类型的用户终端，也可以实施为服务器。下面，将说明设备实施为终端或服务器时示例性应用。
本公开实施例提供一种对象识别方法，应用于对象识别系统，所述对象识别系统包括管理节点和服务节点；如图1所示，示出本公开实施例提供的第一种对象识别方法的流程示意图；结合图1所示步骤进行说明：
步骤S101，采用所述管理节点获取至少一个入库对象的特征数据库。
在一些实施例中，对象识别系统可以是人脸识别系统，还可以是指纹识别的系统。其中，对象识别系统可以包括多个处理终端或处理节点，且多个处理终端之间可以位于同一区域，也可以同时位于多个不同区域。同时对象识别系统中的对象可以指代人、终端、动物等，同时在对象为人的情况下，对象识别系统中可以对人脸或指纹或声音进行识别。
在一些实施例中，对象识别系统包括管理节点和服务节点；其中，管理节点可以指代对象识别系统的数据中心，即信息处理中心，其可以用于获取以及存储需入库对象的相关数据信息。其中，管理节点可以有一个数据处理终端，也可以具有多个数据处理终端。同时入库对象即为需在对象识别系统对应的数据库中事先进行相关信息存储的对象。且服务节点可以指代对象识别系统的应用中心，即可采用该服务节点基于至少一个入库对象的特征数据库，以实现对相关对象进行识别的过程。
在一些实施例中，管理节点可以包括数据节点和特征节点；其中，可采用数据节点获取每一入库对象的原始数据，比如：每一入库对象的原始图像、原始指纹信息、原始声音信息，同时可以采用该特征节点对每一入库对象输入的原始数据进行特征提取，以得到每一入库对象的特征数据库。其中，每一入库对象的特征数据库中的特征数量可以有多个；同时每一入库对象的特征数据库中的特征数据可以是每一入库对象的原始数据对应的特征值；其中，该特征值可以使用向量进行表示。
在一些实施例中，对象识别系统中的入库对象的数量可以是一个，或两个及以上，同时在入库对象的数量为两个及以上的情况下，每两个入库对象各自对应的特征数据库中的特征数据不同。示例性地，入库对象为某一学校的全部师生，对应地，入库对象的特征数据库中的特征数据，可以是某位师生对应的图像信息进行特征提取后得到的特征数据；入库对象为某一公司的全部员工，对应地，入库对象的特征数据库中的特征数据，可以是某位员工对应的图像信息进行特征提取后得到的特征数据。
步骤S102，采用所述服务节点基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。
在一些实施例中，可以将至少一个入库对象的特征数据库，下发至对象识别系统中的服务节点，进而采用该服务节点基于特征数据库，对需要识别的待识别对象进行识别， 以得到待识别对象的识别结果。其中，服务节点的数量可以有多个。
在一些实施例中，在服务节点为多个的情况下，可以基于服务节点的属性信息，比如服务节点的接口属性对服务节点进行划分，比如：用于进行考勤业务的服务节点，用于进行监控业务的服务节点，同时每一类服务节点可以包括部署在不同区域的多个子服务节点。在一些实施例中，服务节点可以包括多个不同属性的应用节点，每一应用节点关联多个不同的终端节点；其中，每一终端节点用于采集待识别对象输入的待识别数据，并基于至少一个入库对象的特征数据库，对待识别数据进行特征比对，以得到待识别对象的识别结果。
在一些实施例中，对象识别系统可采用服务节点，基于特征数据库，对采集到的待识别对象输入的待识别数据进行数据识别，以得到待识别对象的识别结果。其中，待识别对象的识别结果，可以用于表征该待识别对象是否属于该入库对象中的一个。
在一些实施例中，服务节点可以为某一公司的考勤系统对应的应用节点，用于基于事先存储的该公司的每一员工的特征数据库，对进入该应用节点管理区域的待识别对象进行身份识别，以得到该待识别对象的识别结果。其中，识别结果表征为通过，则表征该待识别对象为该公司的某一员工，若识别结果表征为未通过，则表征该待识别对象非该公司的员工。
在一些实施例中，对象识别系统基于管理节点，确定每一入库对象的特征数据库；如此，能够实现将事先存储的每一入库对象的原始数据，比如：图像或指纹等进行特征转换，以特征数据进行存储，进而能够提高每一入库对象的数据存储的隐私性。同时在基于该特征数据库进行相关应用的过程中，可以直接下发每一入库对象的特征数据库至对应的应用节点，而非下发每一入库对象的原始数据，比如：图像或指纹等，如此，能够实现每一入库对象的隐私数据以特征数据形式传输，即能够提高传输过程数据隐私性。
本公开实施例提供的对象识别方法，所述对象识别方法应用于对象识别系统，所述对象识别系统包括管理节点和服务节点，其中，采用所述管理节点获取至少一个入库对象的特征数据库，然后采用所述服务节点获取该至少一个入库对象的特征数据库；最后采用所述服务节点基于该特征数据库，对待识别对象进行识别，以得到待识别对象的识别结果。如此，能够实现以特征数据的形式存储和传输对象的相关信息，进而能够提高对象识别过程中数据存储和传输过程中的隐私性。
图2A示出可应用本公开实施例所提供的对象识别方法的一种系统架构示意图；如图2A所示，该系统架构中包括：管理节点201、网络202和服务节点203。为实现支撑一个示例性应用，管理终端201和服务节点203通过网络202建立通信连接，管理节点201通过网络202向服务节点203上报其获取的至少一个入库对象的特征数据库；同时，服务节点203基于特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。最后，服务节点203还可将识别结果上传至网络202，并通过网络202发送给管理节点201。
在一些实施例中，在管理节点包括数据节点和特征节点的情况下，采用数据节点获取每一入库对象的第一数据和身份标识，并采用特征节点对每一入库对象的第一数据进行特征提取，得到每一入库对象的目标特征，进而基于每一入库对象的目标特征和身份标识，来确定每一入库对象的特征数据库；如此，能够将每一入库对象的相关数据存储事先转换为对应的特征数据，这样，能够提高入库对象的相关数据的隐私性，即上述步骤S101可以通过以下步骤来实现。如图2B所示，图2为本公开实施例提供的第二种对象识别方法的流程实现图；结合图1和图2B所示的步骤进行以下说明：
步骤S201，采用所述数据节点获取所述至少一个入库对象中每一入库对象的第一数据。
在一些实施例中，采用数据节点获取至少一个入库对象中每一入库对象的第一数据；其中，第一数据可以是每一入库对象的底图数据、指纹数据或声音信息等。同时可以是 采用数据节点采集每一入库对象输入的第二数据，并对第二数据进行质量过滤，以确定每一入库对象的第一数据，即第一数据是入库对象输入至数据节点且经过信息过滤后得到的数据。
在本公开的一些实施例中，首先对获取的每一入库对象的第二数据进行质量检测，得到对应的检测结果，进而基于对应的检测结果对每一入库对象的第二数据进行过滤，得到每一入库对象的第一数据；如此，能够提高得到的每一入库对象的第一数据的质量；即上述步骤S201可以通过以下步骤S2011至步骤S2013来实现，如图2C所示，为本公开实施例提供的对象识别方法中确定每一入库对象的第一数据的流程示意图，其中：
步骤S2011，采用所述数据节点获取所述每一入库对象的第二数据。
在一些实施例中，可以采用数据节点采集并获取每一入库对象的第二数据，其中，第二数据可以是每一入库对象输入至数据节点的原始图像、原始指纹、原始声音信息等。
其中，第二数据可以包括但不限于：入库对象的人脸图像信息、身体图像信息、指纹信息、声音信息等，同时第二数据的存储形式可以是图片、音频、视频等。
步骤S2012，采用所述特征节点对所述每一入库对象的第二数据进行质量检测，得到所述第二数据对应的检测结果。
在一些实施例中，可以将获取的每一入库对象的第二数据，传输至特征节点，以使特征节点对每一入库对象的第二数据进行质量检测，即采用特征节点对接收到的每一入库对象的第二数据进行质量检测，得到第二数据对应的检测结果；其中，第二数据对应的检测结果，可以是检测通过，或检测不通过。
示例性地，在第二数据为每一入库对象的人脸图像的情况下，第二数据中人脸模糊、存在多张人脸、第二数据对应的图像尺寸过大或过小，人脸遮挡较多等，对应地，第二数据对应的检测结果为不通过；同时第二数据中的人脸清晰度较高且仅存在一张人脸，对应地，第二数据对应的检测结果为通过。
步骤S2013，采用所述数据节点基于所述检测结果对所述每一入库对象的第二数据进行过滤，得到所述每一入库对象的第一数据。
在一些实施例中，可以将所述第二数据对应的检测结果反馈至数据节点，并采用数据节点基于该检测结果对每一入库对象的第二数据进行过滤，以得到每一入库对象的第一数据；其中，每一入库对象的第一数据中的数据即为符合要求，即质量检测通过的数据。
其中，每一入库对象的第一数据可以包括多个数据，比如，可以同时包括每一入库对象的人脸图像、指纹信息以及声音信息。
在一些实施例中，通过对每一入库对象的原始数据进行信息过滤，即对原始数据进行质量检测，以过滤掉质量检测结果为不符合预设标准的数据，进而使得存储至数据节点中的第一数据的质量更高，同时能够优化存储效率和存储空间。
步骤S202，采用所述特征节点对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征。
在一些实施例中，将进行质量过滤后得到的每一入库对象的第一数据，从数据节点传输至特征节点，并采用特征节点对每一入库对象的第一数据进行特征提取，以得到每一入库对象的目标特征；比如，第一数据为入库对象的人脸图像时，每一入库对象的目标特征即为每一入库对象的人脸特征值。第一数据为入库对象的声音信息时，每一入库对象的目标特征即为每一入库对象的声音特征。
在本公开的一些实施例中，基于事先设定好的目标提取模型对每一入库对象的第一数据进行特征提取，得到每一入库对象的目标特征；如此，通过事先设定好的目标提取模型，能够为后续进行特征提取提供基础，同时能够提高得到目标特征的效率。即上述步骤S202可以通过以下步骤S2021和步骤S2022来实现，如图2D所示，为本公开实施例提 供的对象识别方法中确定每一入库对象的目标特征的流程示意图，其中：
步骤S2021，采用所述数据节点确定目标提取模型。
在一些实施例中，可以采用数据节点确定进行特征提取所使用的算法模型，即目标提取模型。
在本公开的一些实施例中，目标提取模型可以是数据节点基于特征节点当前内部存储的特征提取模型集合来确定；如此，通过特征节点内部事先存储的特征提取模型集，来选取适当的目标提取模型，能够提高确定的目标提起模型的准确度和效率；即上述步骤S2021可以通过以下步骤过程来实现：
首先，响应于所述数据节点发送的查询请求，采用所述特征节点统计所述特征节点内部存储的特征提取模型得到特征提取模型列表，并将所述特征提取模型列表发送至所述数据节点。
在一些实施例中，数据节点发起一查询请求至特征节点，特征节点响应于该查询请求，以统计特征节点内部存储的特征提取模型，进而得到特征提取模型列表。其中，该特征提取模型列表中包括至少一个特征提取模型，该特征提取模型列表可以是特征节点中内部存储的可用的特征提取模型集合对应的列表信息。
在一些实施例中，采用特征节点将统计得到的特征提取模型列表发送至数据节点，以使数据节点获知特征节点中能够使用的特征提取模型。
这里，特征节点内部存储的特征提取模型可以是数据节点事先上传至特征节点的；如此，能够基于事先存储的特征提取模型执行相关特征提取操作，进而能够提高提取目标特征的效率。即在执行上述“响应于所述数据节点发送的查询请求，采用所述特征节点统计所述特征节点内部存储的特征提取模型得到特征提取模型列表，并将所述特征提取模型列表发送至所述数据节点”之前，还可以执行以下过程：
第一步，采用所述数据节点发送与所述服务节点关联的特征提取模型至所述特征节点。
第二步，采用所述特征节点，在所述特征节点内部存储所述服务节点关联的特征提取模型。
在一些实施例中，对象识别系统中的数据节点，可以事先获取与服务节点关联的特征提取模型，并将该特征提取模型上传至特征节点，以使特征节点内部存储该特征提取模型。其中，与服务节点关联的特征提取模型，可以是服务节点对应的终端节点关联的特征提取模型，即终端节点内部存储的可用的特征提取模型。
然后，采用所述数据节点从所述特征提取模型列表中确定所述目标提取模型。
在一些实施例中，可以采用数据节点从特征提取模型列表中确定目标提取模型，以便特征节点基于确定的目标提取模型，对每一入库对象的第一数据进行特征提取，得到每一入库对象对应的目标特征。
步骤S2022，采用所述特征节点基于所述目标提取模型对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征。
在一些实施例中，特征节点基于数据节点指定的目标提取模型对每一入库对象的第一数据进行特征提取，得到每一入库对象的目标特征。
步骤S203，采用所述数据节点获取所述每一入库对象的身份标识。
在一些实施例中，同时对象识别系统可以采用数据节点来获取每一入库对象的身份标识；其中，该身份标识可以用于表征每一入库对象的唯一标识。
在本公开的一些实施例中，可以基于每一入库对象的身份信息集，确定每一入库对象的身份标识；如此，能够提高确定的每一入库对象的身份标识的效率和准确度；即上述步骤S203可以通过以下步骤S2031至步骤S2033来实现，如图2E所示，为本公开实施例提供的对象识别方法中获取每一入库对象的身份标识的流程示意图，其中：
步骤S2031，采用所述数据节点获取所述每一入库对象的身份信息集。
在一些实施例中，可采用数据节点来获取每一入库对象输入的身份信息集，其中，每一入库对象的身份信息集中可以包括但不限于：每一入库对象的姓名、身份证号码、工号、学号等。这里每一入库对象的第二数据和身份信息集可以是每一入库对象同时输入至数据节点的。
步骤S2032，采用所述服务节点基于所述每一入库对象的身份信息集生成所述每一入库对象的身份标识。
在一些实施例中，可以将每一入库对象的身份信息集传递至服务节点，以使服务节点保存每一入库对象的身份信息集，同时该服务节点还可以基于每一入库对象的身份信息集，生成与每一入库对象的身份标识。
步骤S2033，采用所述服务节点将所述每一入库对象的身份标识发送至所述数据节点。
在一些实施例中，服务节点可以将每一入库对象的身份标识发送至数据节点，以使数据节点基于该身份标识对相关入库对象进行信息更改、删除或增加等。
步骤S204，采用所述数据节点基于所述每一入库对象的目标特征和所述每一入库对象的身份标识，确定所述至少一个对象的特征数据库。
在一些实施例中，基于数据节点确定的每一入库对象的目标特征和每一入库对象的身份标识，将每一入库对象对应的目标特征和身份标识记性关联，以确定至少一个入库对象中每一入库对象的特征数据库。
在一些实施例中，采用数据节点获取每一入库对象的身份信息集和第二数据，然后分别采用特征节点对每一入库对象的第二数据进行信息过滤以及特征提取，得到每一入库对象的目标特征，以及采用服务节点基于每一入库对象的身份信息集进行身份标识，以得到每一入库对象的身份标识，然后采用数据节点基于每一入库对象的身份标识和目标特征，得到每一入库对象的特征数据库。这样，能够提高得到的特征数据库的准确和效率。
在一些实施例中，在所述服务节点包括至少一个应用节点和与所述应用节点关联的至少一个终端节点的情况下，对采集到的待识别对象输入的待识别数据进行身份识别，确定所述待识别对象的身份信息，基于事先存储好的至少一个入库对象的特征数据库，对待识别对象进行身份识别，以确定待识别对象的身份信息，能够在进行相关身份验证的过程中，不涉及原始数据的图像信息等，仅涉及特征数据，能够提高原始数据的隐私性。如图3所示，图3为本公开实施例提供的第三种对象识别方法的流程实现图；结合图1和图3A所示的步骤进行以下说明：
步骤S301，采用所述至少一个应用节点基于所述特征数据库，确定与所述应用节点属性信息关联的目标对象的目标特征数据库。
在一些实施例中，对象识别系统可以基于不同应用节点的属性信息，从至少一个入库对象的特征数据库中，确定与应用节点属性信息关联的目标对象的目标特征数据库。
在一些实施例中，应用节点为学生宿舍门禁系统对应的节点的情况下，确定与应用节点属性信息关联的目标对象，可以是该校全部住校学生的特征数据库，即目标特征数据库；在应用节点为教师宿舍门禁系统对应的节点的情况下，确定与应用节点属性信息关联的目标对象的目标特征数据库，可以是该校全部住校老师的特征数据库。
步骤S302，采用所述至少一个终端节点基于所述目标特征数据库对所述待识别对象进行身份识别，确定所述待识别对象的身份信息。
在一些实施例中，采用与应用节点关联的终端节点，获取该目标特征数据库，并在获取到该目标特征数据库之后，对待识别对象进行身份识别，以确定待识别对象的身份信息。
在一些实施例中，可以采用与终端节点关联的特征提取模型，基于目标特征数据集库对采集到的待识别对象输入的待识别数据进行特征比对，以确定待识别对象的身份信息，即能够直接基于获取的特征数据与采集得到的待识别数据进行特征比对，得到待识别对象的身份信息。这里，即能够基于相关对象的特征数据实现身份验证，能够保证相关对象的相关信息的隐私性，即上述步骤S302可以通过以下步骤S3021至步骤S3023来实现，如图3B所示，为本公开实施例提供的对象识别方法中确定待识别对象的身份信息的流程示意图，其中：
步骤S3021，采用所述终端节点基于与所述终端节点关联的特征提取模型，对采集到的所述待识别对象输入的待识别数据进行特征提取，得到待比对特征数据。
在一些实施例中，采用与终端节点关联的特征提取模型，对采集到的待识别对象输入的待识别数据进行特征提取，得到待比对特征数据。这里，待比对特征数据的数据形式与管理节点获取的至少一个入库对象的特征数据库中特征数据的数据形式相同。
步骤S3022，采用所述终端节点，从所述目标特征数据库查找与所述待比对特征数据相似的特征数据，得到查找结果。
在一些实施例中，采用该终端节点，从目标特征数据库中查找与待比对特征数据相似的特征数据，得到查找结果；其中，该查找结果可以使用未查找到，和查找到进行表示，也可以是0和1进行表示。
步骤S3023，采用所述终端节点，基于所述查找结果确定所述待识别对象的身份信息。
在一些实施例中，采用该终端节点，基于查找结果确定待识别对象的身份信息，其中，若查找结果表征未查找到，则表明该待识别对象的身份信息无法被识别，则可以使用“错误身份信息”或“不正确的身份信息”或“无法识别的身份信息”进行表示。若查找结果表征查找到，则使用查找到的特征数据对应的身份标识来表征该待识别对象。
这里，可以采用所述终端节点将获取到的身份信息上传至与终端节点关联的应用节点，即实现识别结果及时上传。
在本公开的一些实施例中，采用终端节点对待识别对象进行身份信息确定之后，可以将待识别对象输入的待识别数据进行删除，能够提高待识别对象的相关信息的隐私性，即在执行步骤S3023之后，本公开实施例提供的对象识别方法还可以执行以下过程：
采用所述终端节点在确定所述待识别对象的身份信息之后，将所述待识别数据进行删除。
在一些实施例中，采用终端节点对待识别对象的身份信息进行确定之后，将待识别对象输入至终端节点的待识别数据进行删除，即不保存该待识别数据。如此，能够提高对待识别对象的相关信息的隐私性。
下面结合一个具体实施例对上述对象识别方法进行说明，然而值得注意的是，该具体实施例仅是为了更好地说明本公开实施例，并不构成对本公开实施例的不当限定。
人脸识别，是基于人的脸部特征信息进行身份识别的一种生物识别技术，人脸识别的基本过程为：人脸检测、活体检测、人脸跟踪、特征提取、人脸比对、输出结果等。得益于其非接触性、自然性、防伪性等优点，人脸识别技术非常适合用于门禁、打卡、签到等需要进行高频人员身份核验的业务场景。
其中，常见的人脸识别系统组成架构为：人脸识别终端+管理平台。其中，人脸识别终端主要负责图像采集、活体检测、特征提取、特征比对，并根据人脸识别结果执行联动操作(如开门等)；管理平台为用户提供人脸信息管理、人脸信息下发、记录查询等功能。同时随着人脸识别系统的发展，人脸识别系统的隐私安全愈发受到关注。目前业内常见的人脸识别系统的隐私保护方案和措施通常有以下几种：
1、加密存储：在人脸识别终端及管理平台采用强加密算法对人脸信息进行加密存储；
2、加密传输：管理平台可采用安全套接层(Secure Sockets Layer，SSL)、高级加密 标准(Advanced Encryption Standard，AES)等加密协议实现安全传输层协议(Transport Layer Security，TLS)加密。且人脸识别终端可以采用超文本传输安全协议(Hyper Text Transfer Protocol over Secure Socket Layer，HTTPS)接口向服务端传输数据，以保障传输数据全链路处于加密状态。
3、访问控制：提供权限设置功能，依据最小必要原则开放人脸信息访问权限。
4、展示限制：对人脸信息去标识化处理，例如，人脸识别终端和/或管理界面对显示的人脸进行马赛克处理。
然而，即便采取了上述几种措施，人脸信息依旧会存储于人脸识别终端及管理平台中，同时人脸信息会在管理平台和人脸识别终端之间进行传输。在政府或校园或部分企业等场景应用中，用户对人脸识别系统提出了“无痕”要求，即以下要求：
1、存储安全：人脸识别系统的终端及管理平台中，不存储任何人脸图像。
2、传输安全：人脸识别系统的终端与管理平台通信过程中，不直接传输人脸图像。
即目前常见的人脸识别系统隐私保护方案存在以下不足：
1、人脸图像仍旧存储在人脸识别终端和管理平台中，仍旧有被泄漏破解的风险；
2、在人脸图像下发、记录上传过程中，人脸识别终端和管理平台间会传输人脸图像，有可能被截取破解造成隐私数据泄漏。
基于此，本公开实施例提供一种对象识别系统，由“数据节点、特征节点、应用节点、终端设备”组成。依据“最小够用”原则，可以实现人脸图像在整个人脸识别系统的最小化存储和传输。如图4所示，为应用本公开实施例提供的一种对象识别系统的系统架构示意图；其中，401为数据节点，402为特征节点，403为多个不同属性的应用节点，404为应用节点A对应的多个终端设备。其主要技术实现为：
1、数据节点：负责存储用户的原始人脸底图数据及基本信息；且一个系统架构中数据节点通常只有一个，且部署在可信环境中。
2、特征节点：负责接收数据节点的原始人脸底图，将其转化为与人脸识别终端算法模型相匹配的特征值，其并不存储原始人脸图像。其中，系统结构中特征节点的数量可为一个或多个，具备多种算法模型的特征提取能力，部署在可信环境中。
3、应用节点：负责接收特征值以及用户的基本信息，并将特征值以及用户的基本信息下发到指定的人脸识别终端上，其不存储原始人脸图像。应用节点可以根据实际需要部署多套(例如：人脸门禁系统、人脸布控系统、人脸考勤系统等)，可部署在非可信环境中。
4、终端设备：负责现场人脸图像采集的人脸识别终端。终端设备本地采用人脸特征进行身份认证，并在认证完成后立即删除原始抓拍图像。终端设备不存储、不上传原始人脸底图及抓拍图图像。
基于图4提供的系统架构图，以实现以下步骤，如图5所示，为基于本公开实施例提供的对象识别系统来实现对象识别方法的一种流程示意图；其主要划分为五大步骤：质量检测、特征提取、人员入库、人员下发以及身份认证；其中：
1、质量检测：即执行步骤501至步骤503，数据节点将人脸图片传递给特征节点，特征节点对人脸图片进行质量检测，以得到对应的质量检测结果；然后将特征节点会将质量检测结果反馈至数据节点；数据节点会基于该质量检测结果对对不合格的人脸图片(例如：没有人脸、多人脸、尺寸过大和尺寸过小中的至少之一、人脸遮挡较多等情形)进行过滤和拒绝。
2、特征提取：首先上传算法模型，即执行步骤504：数据节点提取与终端设备一致的算法模型，并将其上传至特征节点；特征节点保存该算法模型，即执行步骤505以及反馈模型上传结果506；其中，该算法模型是特征节点对人脸图片进行特征提取时所采用的算法模型。
其次，数据节点查询算法模型版本：数据节点向特征节点查询可用的算法模型版本，特征节点响应于数据节点发送的查询模型版本的请求，并返回可用的算法模型版本列表至数据节点；即执行步骤507至步骤509。
最后，数据节点可基于该算法模型版本列表，指定一个或多个可用的算法模型版本，将满足入库质量要求的人脸图片传递给特征节点，特征节点对人脸图片进行特征提取，返回对应的人脸特征，数据节点保存该人脸特征；即执行步骤510至步骤512。
3、人员入库：首先，基本信息入库：数据节点将人员的基本信息(如姓名、工号、学号、卡号、身份证号等)传递给应用节点，应用节点保存人员信息，并返回对应的人员唯一标识(此标识可用于后续的人员信息变更/查询/删除)；其次，特征入库：数据节点将内部存储的人员唯一标识、算法模型版本、人脸特征传递给应用节点，应用节点保存人员与人脸特征的关联关系，即执行步骤513至步骤515。
4、人员下发：即数据节点通过调用应用节点的人员下发接口(如新建组、将人员添加到组、将组与下发给设备等)，向不同的设备下发不同人员的人脸特征值，即执行步骤516至步骤519：基于应用节点的属性信息不同，下发不同人员的人脸特征至对应的应用节点。
5、身份认证：人脸识别终端采集人脸图像，即执行步骤520：人员刷脸，并在人员刷脸之后提取对应的待识别人脸特征与本地特征库进行比对，完成人员身份认证后向应用节点上传识别结果记录；其中，本地特征库即步骤519中执行的人员下发。同时人脸识别终端不存储任何人脸底图或人脸抓拍图，且不上传人脸抓拍图给应用节点。
基于以上步骤，相比以往的人脸识别系，能够采用特征节点、应用节点分离的“双节点”方案，能够实现非可信环境下人脸图像数据的最小化存储。即可以实现：
1、存储安全：人脸识别系统的终端及管理平台中，不存储任何人脸图像。
2、传输安全：人脸识别系统的终端与管理平台通信过程中，不直接传输人脸图像。
3、传输效率高：相关技术中，人脸底图必须从数据节点通过网络依次传输到应用节点和终端设备，整体传输数据量较大，占用带宽高。基于以上步骤采用特征下发的方式，能够大大降低人脸下发这一过程的传输耗时。
4、拓展性强：基于以上步骤可实现数据节点对接多套人脸识别应用系统，维护不同系统所需的人脸特征关系，在保证隐私安全的前提下将人脸特征下发给不同终端设备。
本公开实施例提供的对象识别方法以及对象识别系统，可以应用于校园多人脸应用系统部署案例；其中，在校园场景内，学生个人信息通常由信息处统一管理，而各类人脸识别应用系统则由对应行政机构统一维护，例如：宿舍人脸门禁系统由后勤处管理维护；校园人脸监控系统由保卫处管理维护；学生人脸考勤系统由教务处管理维护；为保证学生信息安全，这些应用系统可以采用本公开实施例提供的对象识别系统进行统一部署管理，即信息处作为统一数据中心，在可信环境内部署特征节点，将人脸原始图片转化为特征值后，分别下发给不同应用系统，即可实现在不泄漏学生原始人脸数据的前提下满足各类系统的正常使用。
本公开实施例提供一种对象识别系统，图6为本公开实施例提供的对象识别系统的结构组成示意图，如图6所示，所述对象识别系统600包括：
所述管理节点601，被配置为获取至少一个入库对象的特征数据库；
所述服务节点602，被配置为基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。
在一些实施例中，所述管理节点包括数据节点和特征节点，在所述对象识别系统600中，所述数据节点，被配置为获取所述至少一个入库对象中每一入库对象的第一数据；所述特征节点，被配置为对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征；所述数据节点，还被配置为获取所述每一入库对象的身份标识； 所述数据节点，还被配置为基于所述每一入库对象的目标特征和所述每一入库对象的身份标识，确定所述至少一个入库对象的特征数据库。
在一些实施例中，在所述对象识别系统600中，所述数据节点，还被配置为获取所述每一入库对象的第二数据；所述特征节点，还被配置为对所述每一入库对象的第二数据进行质量检测，得到所述第二数据对应的检测结果；所述数据节点，还被配置为基于所述检测结果对所述每一入库对象的第二数据进行过滤，得到所述每一入库对象的第一数据。
在一些实施例中，在所述对象识别系统600中，所述数据节点，还被配置为确定目标提取模型；所述特征节点，还被配置为所述目标提取模型对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征。
在一些实施例中，在所述对象识别系统600中，所述数据节点，还被配置为响应于所述数据节点发送的查询请求，统计所述特征节点内部存储的特征提取模型得到特征提取模型列表，并将所述特征提取模型列表发送至所述数据节点；所述数据节点，还被配置为从所述特征提取模型列表中确定所述目标提取模型。
在一些实施例中，在所述对象识别系统600中，所述数据节点，还被配置为发送与所述服务节点关联的特征提取模型至所述特征节点；所述特征节点，还被配置为在所述特征节点内部存储所述服务节点关联的特征提取模型。
在一些实施例中，在所述对象识别系统600中，所述数据节点，还被配置为获取所述每一入库对象的身份信息集；所述服务节点，还被配置为基于所述每一入库对象的身份信息集生成所述每一入库对象的身份标识；所述服务节点，还被配置为将所述每一入库对象的身份标识发送至所述数据节点。
在一些实施例中，所述服务节点包括至少一个应用节点和与所述应用节点关联的至少一个终端节点，在所述对象识别系统600中，所述至少一个应用节点，被配置为基于所述特征数据库，确定与所述应用节点属性信息关联的目标对象的目标特征数据库；所述至少一个终端节点，被配置为基于所述目标特征数据库对所述待识别对象进行身份识别，确定所述待识别对象的身份信息。
在一些实施例中，在所述对象识别系统600中，所述终端节点，还被配置为基于与所述终端节点关联的特征提取模型，对采集到的所述待识别对象输入的待识别数据进行特征提取，得到待比对特征数据；所述终端节点，还被配置为从所述目标特征数据库查找与所述待比对特征数据相似的特征数据，得到查找结果；所述终端节点，还被配置为基于所述查找结果确定所述待识别对象的身份信息。
在一些实施例中，所述终端节点，还被配置为在确定所述待识别对象的身份信息之后，将所述待识别数据进行删除。
需要说明的是，以上系统实施例的描述，与上述方法实施例的描述是类似的，具有同方法实施例相似的有益效果。对于本公开系统实施例中未披露的技术细节，请参照本公开方法实施例的描述而理解。
需要说明的是，本公开实施例中，如果以软件功能模块的形式实现上述的对象识别方法，并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。基于这样的理解，本公开实施例的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是终端、服务器等)执行本公开各个实施例所述方法的全部或部分。而前述的存储介质包括：U盘、运动硬盘、只读存储器(Read Only Memory，ROM)、磁碟或者光盘等各种可以存储程序代码的介质。这样，本公开实施例不限制于任何特定的硬件和软件结合。
对应地，本公开实施例再提供一种计算机程序产品，所述计算机程序产品包括计算 机可执行指令，该计算机可执行指令被执行后，能够实现本公开实施例提供的对象识别方法。
相应的，本公开实施例再提供一种计算机存储介质，所述计算机存储介质上存储有计算机可执行指令，所述计算机可执行指令被处理器执行时实现上述实施例提供的对象识别方法。
其中，该计算机存储介质可以是易失性存储介质或非易失性存储介质。
本公开实施例提供一种计算机程序，所述计算机程序包括计算机可读代码，在所述计算机可读代码在电子设备中运行的情况下，所述电子设备的处理器执行用于实现上述实施例提供的对象识别方法。
以上对象识别装置、系统和存储介质实施例的描述，与上述方法实施例的描述是类似的，具有同相应方法实施例相似的技术描述和有益效果，限于篇幅，可按照上述方法实施例的记载，故在此不再赘述。对于本公开对象识别装置、系统、存储介质及程序实施例中未披露的技术细节，请参照本公开方法实施例的描述而理解。
应理解，说明书通篇中提到的“一个实施例”或“一实施例”意味着与实施例有关的特定特征、结构或特性包括在本公开实施例的至少一个实施例中。因此，在整个说明书各处出现的“在一个实施例中”或“在一实施例中”未必一定指相同的实施例。此外，这些特定的特征、结构或特性可以任意适合的方式结合在一个或多个实施例中。应理解，在本公开实施例的各种实施例中，上述各过程的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本公开实施例的实施过程构成任何限定。上述本公开实施例序号仅仅为了描述，不代表实施例的优劣。需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。
在本公开实施例所提供的几个实施例中，应该理解到，所揭露的设备和方法，可以通过其它的方式实现。以上所描述的设备实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，如：多个单元或组件可以结合，或可以集成到另一个系统，或一些特征可以忽略，或不执行。另外，所显示或讨论的各组成部分相互之间的耦合、或直接耦合、或通信连接可以是通过一些接口，设备或单元的间接耦合或通信连接，可以是电性的、机械的或其它形式的。
上述作为分离部件说明的单元可以是、或也可以不是物理上分开的，作为单元显示的部件可以是、或也可以不是物理单元；既可以位于一个地方，也可以分布到多个网络单元上；可以根据实际的需要选择其中的部分或全部单元来实现本实施例方案的目的。
另外，在本公开实施例各实施例中的各功能单元可以全部集成在一个处理单元中，也可以是各单元分别单独作为一个单元，也可以两个或两个以上单元集成在一个单元中；上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。本领域普通技术人员可以理解：实现上述方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成，前述的程序可以存储于计算机可读取存储介质中，该程序在执行时，执行包括上述方法实施例的步骤；而前述的存储介质包括：移动存储设备、ROM、磁碟或者光盘等各种可以存储程序代码的介质。
或者，本公开实施例上述集成的单元如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。基于这样的理解，本公开实施例的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算 机设备(可以是个人计算机、服务器、或者网络设备等)执行本公开实施例各个实施例所述方法的全部或部分。而前述的存储介质包括：移动存储设备、ROM、磁碟或者光盘等各种可以存储程序代码的介质。以上所述，仅为本公开实施例的具体实施方式，但本公开实施例的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本公开实施例揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本公开实施例的保护范围之内。因此，本公开实施例的保护范围应以所述权利要求的保护范围为准。
工业实用性
本公开实施例提供了一种对象识别方法、系统、存储介质及程序；其中，所述对象识别方法，应用于对象识别系统，所述对象识别系统包括管理节点和服务节点；采用所述管理节点获取至少一个入库对象的特征数据库；采用所述服务节点基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。

Claims (22)

  1. 一种对象识别方法，应用于对象识别系统，所述对象识别系统包括管理节点和服务节点；其中：
    采用所述管理节点获取至少一个入库对象的特征数据库；
    采用所述服务节点基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。
  2. 根据权利要求1所述的方法，其中，所述管理节点包括数据节点和特征节点，所述采用所述管理节点获取至少一个入库对象的特征数据库，包括：
    采用所述数据节点获取所述至少一个入库对象中每一入库对象的第一数据；
    采用所述特征节点对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征；
    采用所述数据节点获取所述每一入库对象的身份标识；
    采用所述数据节点基于所述每一入库对象的目标特征和所述每一入库对象的身份标识，确定所述至少一个入库对象的特征数据库。
  3. 根据权利要求2所述的方法，其中，所述采用所述数据节点获取所述至少一个入库对象中每一入库对象的第一数据，包括：
    采用所述数据节点获取所述每一入库对象的第二数据；
    采用所述特征节点对所述每一入库对象的第二数据进行质量检测，得到所述第二数据对应的检测结果；
    采用所述数据节点基于所述检测结果对所述每一入库对象的第二数据进行过滤，得到所述每一入库对象的第一数据。
  4. 根据权利要求2或3所述的方法，其中，所述采用所述特征节点对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征，包括：
    采用所述数据节点确定目标提取模型；
    采用所述特征节点基于所述目标提取模型对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征。
  5. 根据权利要求4所述的方法，其中，所述采用所述数据节点确定目标提取模型，包括：
    响应于所述数据节点发送的查询请求，采用所述特征节点统计所述特征节点内部存储的特征提取模型得到特征提取模型列表，并将所述特征提取模型列表发送至所述数据节点；
    采用所述数据节点从所述特征提取模型列表中确定所述目标提取模型。
  6. 根据权利要求5所述的方法，其中，所述响应于所述数据节点发送的查询请求，采用所述特征节点统计所述特征节点内部存储的特征提取模型得到特征提取模型列表，并将所述特征提取模型列表发送至所述数据节点之前，所述方法还包括：
    采用所述数据节点发送与所述服务节点关联的特征提取模型至所述特征节点；
    采用所述特征节点，在所述特征节点内部存储所述服务节点关联的特征提取模型。
  7. 根据权利要求2至6任一所述的方法，其中，所述采用所述数据节点获取所述每一入库对象的身份标识，包括：
    采用所述数据节点获取所述每一入库对象的身份信息集；
    采用所述服务节点基于所述每一入库对象的身份信息集生成所述每一入库对象的身份标识；
    采用所述服务节点将所述每一入库对象的身份标识发送至所述数据节点。
  8. 根据权利要求1至7任一所述的方法，其中，所述服务节点包括至少一个应用节 点和与所述应用节点关联的至少一个终端节点，所述采用所述服务节点基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果，包括：
    采用所述至少一个应用节点基于所述特征数据库，确定与所述应用节点属性信息关联的目标对象的目标特征数据库；
    采用所述至少一个终端节点基于所述目标特征数据库对所述待识别对象进行身份识别，确定所述待识别对象的身份信息。
  9. 根据权利要求8所述的方法，其中，所述采用所述至少一个终端节点基于所述目标特征数据库对所述待识别对象进行身份识别，确定所述待识别对象的身份信息，包括：
    采用所述终端节点基于与所述终端节点关联的特征提取模型，对采集到的所述待识别对象输入的待识别数据进行特征提取，得到待比对特征数据；
    采用所述终端节点，从所述目标特征数据库查找与所述待比对特征数据相似的特征数据，得到查找结果；
    采用所述终端节点，基于所述查找结果确定所述待识别对象的身份信息。
  10. 根据权利要求9所述的方法，其中，所述方法还包括：
    采用所述终端节点在确定所述待识别对象的身份信息之后，将所述待识别数据进行删除。
  11. 一种对象识别系统，所述对象识别系统包括：管理节点和服务节点，其中：
    所述管理节点，被配置为获取至少一个入库对象的特征数据库；
    所述服务节点，被配置为基于所述特征数据库，对待识别对象进行识别，以得到所述待识别对象的识别结果。
  12. 根据权利要求11所述的系统，其中，所述管理节点包括数据节点和特征节点，
    所述数据节点，被配置为获取所述至少一个入库对象中每一入库对象的第一数据；
    所述特征节点，被配置为对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征；
    所述数据节点，还被配置为获取所述每一入库对象的身份标识；
    所述数据节点，还被配置为基于所述每一入库对象的目标特征和所述每一入库对象的身份标识，确定所述至少一个入库对象的特征数据库。
  13. 根据权利要求12所述的系统，其中，
    所述数据节点，还被配置为获取所述每一入库对象的第二数据；
    所述特征节点，还被配置为对所述每一入库对象的第二数据进行质量检测，得到所述第二数据对应的检测结果；
    所述数据节点，还被配置为基于所述检测结果对所述每一入库对象的第二数据进行过滤，得到所述每一入库对象的第一数据。
  14. 根据权利要求12或13所述的系统，其中，
    所述数据节点，还被配置为确定目标提取模型；
    所述特征节点，还被配置为基于所述目标提取模型对所述每一入库对象的第一数据进行特征提取，得到所述每一入库对象的目标特征。
  15. 根据权利要求14所述的系统，其中，
    所述特征节点，还被配置为响应于所述数据节点发送的查询请求，统计所述特征节点内部存储的特征提取模型得到特征提取模型列表，并将所述特征提取模型列表发送至所述数据节点；
    所述数据节点，还被配置为从所述特征提取模型列表中确定所述目标提取模型。
  16. 根据权利要求15所述的系统，其中，
    所述数据节点，还被配置为发送与所述服务节点关联的特征提取模型至所述特征节点；
    所述特征节点，还被配置为在所述特征节点内部存储所述服务节点关联的特征提取模型。
  17. 根据权利要求12至16任一所述的系统，其中，
    所述数据节点，还被配置为获取所述每一入库对象的身份信息库；
    所述服务节点，还被配置为基于所述每一入库对象的身份信息集生成所述每一入库对象的身份标识；
    所述服务节点，还被配置为将所述每一入库对象的身份标识发送至所述数据节点。
  18. 根据权利要求11至17任一所述的系统，其中，所述服务节点包括至少一个应用节点和与所述应用节点关联的至少一个终端节点，
    所述至少一个应用节点，被配置为基于所述特征数据库，确定与所述应用节点属性信息关联的目标对象的目标特征数据库；
    所述至少一个终端节点，被配置为基于所述目标特征数据库对所述待识别对象进行身份识别，确定所述待识别对象的身份信息。
  19. 根据权利要求18所述的系统，其中，
    所述终端节点，还被配置为基于与所述终端节点关联的特征提取模型，对采集到的所述待识别对象输入的待识别数据进行特征提取，得到待比对特征数据；
    所述终端节点，还被配置为从所述目标特征数据库查找与所述待比对特征数据相似的特征数据，得到查找结果；
    所述终端节点，还被配置为基于所述查找结果确定所述待识别对象的身份信息。
  20. 根据权利要求19所述的系统，其中，
    所述终端节点，还被配置为在确定所述待识别对象的身份信息之后，将所述待识别数据进行删除。
  21. 一种计算机存储介质，所述计算机存储介质上存储有计算机可执行指令，该计算机可执行指令被执行后，能够实现权利要求1至10任一项所述的对象识别方法。
  22. 一种计算机程序，所述计算机程序包括计算机可读代码，在所述计算机可读代码在电子设备中运行的情况下，所述电子设备的处理器执行用于实现如权利要求1至10任一所述的对象识别方法。
PCT/CN2022/126822 2021-12-27 2022-10-21 对象识别方法、系统、存储介质及程序 WO2023124448A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111619572.4A CN114333005A (zh) 2021-12-27 2021-12-27 对象识别方法、系统及存储介质
CN202111619572.4 2021-12-27

Publications (1)

Publication Number Publication Date
WO2023124448A1 true WO2023124448A1 (zh) 2023-07-06

Family

ID=81014096

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/126822 WO2023124448A1 (zh) 2021-12-27 2022-10-21 对象识别方法、系统、存储介质及程序

Country Status (2)

Country Link
CN (1) CN114333005A (zh)
WO (1) WO2023124448A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114333005A (zh) * 2021-12-27 2022-04-12 深圳市商汤科技有限公司 对象识别方法、系统及存储介质
CN115080775A (zh) * 2022-08-24 2022-09-20 深圳市亲邻科技有限公司 门禁设备人脸数据的存储方法、装置、服务器及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160203367A1 (en) * 2013-08-23 2016-07-14 Nec Corporation Video processing apparatus, video processing method, and video processing program
CN113095430A (zh) * 2021-04-26 2021-07-09 北京瑞莱智慧科技有限公司 可保护隐私的模型更新方法、对象识别方法、系统、装置、介质和设备
CN113554046A (zh) * 2020-04-24 2021-10-26 阿里巴巴集团控股有限公司 图像处理方法和系统、存储介质及计算设备
CN114333005A (zh) * 2021-12-27 2022-04-12 深圳市商汤科技有限公司 对象识别方法、系统及存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160203367A1 (en) * 2013-08-23 2016-07-14 Nec Corporation Video processing apparatus, video processing method, and video processing program
CN113554046A (zh) * 2020-04-24 2021-10-26 阿里巴巴集团控股有限公司 图像处理方法和系统、存储介质及计算设备
CN113095430A (zh) * 2021-04-26 2021-07-09 北京瑞莱智慧科技有限公司 可保护隐私的模型更新方法、对象识别方法、系统、装置、介质和设备
CN114333005A (zh) * 2021-12-27 2022-04-12 深圳市商汤科技有限公司 对象识别方法、系统及存储介质

Also Published As

Publication number Publication date
CN114333005A (zh) 2022-04-12

Similar Documents

Publication Publication Date Title
US11611553B2 (en) Online identity verification platform and process
WO2023124448A1 (zh) 对象识别方法、系统、存储介质及程序
US10541806B2 (en) Authorizing account access via blinded identifiers
US20210166328A1 (en) Cross-blockchain interaction method, system, computer device, and storage medium
EA037018B1 (ru) Способ цифрового подписания электронного файла
KR101859433B1 (ko) 영상 모자이크 처리 방법 및 이를 위한 장치
US20220052998A1 (en) Proximity based data access restrictions
US11768957B2 (en) Privacy-preserving image distribution
US10802888B2 (en) Information processing device and cooperative distributed storage system
TWI746399B (zh) 基於區塊鏈的資訊讀寫方法以及裝置
KR102093145B1 (ko) 생체정보 인식 기반의 데이터 최적화를 위한 오브젝트 스토리지 클라우드 시스템
WO2019200737A1 (zh) 不动产资料上传方法、装置、计算机设备和存储介质
JP7236042B2 (ja) 準同型暗号を用いた顔認証のアプリケーション
CN111552822A (zh) 一种基于区块链节点通信的用户信息报告生成方法
US20230116631A1 (en) System and method for secure collection and display of sensitive data
CA3121338A1 (en) System and method for identity creation and assertion
Yasmeen et al. Intelligent systems powered hourly attendance capturing system
US11704398B1 (en) Real-world object-based image authentication method and system
CN112367314B (zh) 身份认证的方法、装置、计算设备和介质
Wong et al. A privacy-preserving framework for surveillance systems
CN108230514A (zh) 人员库更新方法、装置、电子设备和存储介质
KR100874354B1 (ko) Ｒｆｉｄ 인증 시스템 및 단말기, 방법
CN114171172A (zh) 一种医疗信息匹配与分发的系统和方法
CN112702623A (zh) 视频处理方法、装置、设备及存储介质
US20240179262A1 (en) Object data exchange

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22913711

Country of ref document: EP

Kind code of ref document: A1