CN109241111B - Distributed face recognition system and method based on memory database - Google Patents
Distributed face recognition system and method based on memory database Download PDFInfo
- Publication number
- CN109241111B CN109241111B CN201810981766.0A CN201810981766A CN109241111B CN 109241111 B CN109241111 B CN 109241111B CN 201810981766 A CN201810981766 A CN 201810981766A CN 109241111 B CN109241111 B CN 109241111B
- Authority
- CN
- China
- Prior art keywords
- face
- redis
- kafka
- picture
- message
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
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
- G06V40/161—Detection; Localisation; Normalisation
-
- 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
- G06V40/168—Feature extraction; Face representation
Landscapes
- Engineering & Computer Science (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Processing Or Creating Images (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of face recognition, and discloses a distributed face recognition system and a method based on a memory database, wherein the system comprises a plurality of network cameras, a plurality of face detection modules, a plurality of face feature extraction modules, a plurality of face feature comparison modules, Redis, KAFKA, a database and a platform management module, and the modules are connected with one another through a network; the method comprises the steps that binary data such as picture streams, human face characteristic values and the like are forwarded by using a middle Key memory database Redis, a Key value stored in the Redis is attached to a KAFKA message for transmission, and a module receiving the message directly acquires available picture streams and characteristic values from the Redis through a Redis Key in the KAFKA message; the method provided by the invention avoids the conversion and inverse conversion needed when directly transmitting through KAFKA, thus reducing the occupation of network resources; and the storage and extraction speed of the binarized data in the memory database is higher than that of KAFKA, so that the processing capacity of the server is improved.
Description
Technical Field
The invention belongs to the technical field of face recognition, and particularly relates to a distributed face recognition system and method based on a memory database.
Background
The face recognition is a research hotspot of the current artificial intelligence and mode recognition, the initial application is from the filing management and criminal investigation and case solving of criminal photos by the public security department, and the requirements of quick, efficient and automatic face recognition are increasingly urgent along with the development of science and technology. The technology has good application in the fields of certificate verification, criminal investigation and case solving, entrance control, information safety, video monitoring and the like.
In the actual engineering application process, the face recognition system is often subjected to dynamic increase and decrease of the network cameras caused by change of project quantities. When the number of the added cameras exceeds the upper limit of the processing capacity of the server, the server with higher performance needs to be replaced, a plurality of maintainability problems such as system reinstallation and user data migration are faced in the server replacement process, and the number of the network cameras supported by the system is influenced by the processing capacity of the server.
The system adopting the distributed architecture can support the number of theoretically unlimited network cameras and can continuously expand the capacity according to the increase of server hardware; however, although the advantages of distributed systems in terms of capacity expansion and maintenance are clear, there are many complex places to implement. For example, the distributed system uses KAFKA as a distributed message queue, the messages supported by KAFKA are all of a character string type, pictures are transmitted between processes, feature values are transmitted, and the conversion is required to be Base64, and users receive the messages and convert the messages from Base64 into pictures or feature value types, which not only has overhead on the performance of programs in conversion and reverse conversion. But also the overhead for the network is very large.
Disclosure of Invention
Aiming at the defects or improvement requirements in the prior art, the invention provides a distributed face recognition system and a distributed face recognition method based on a memory database, which adopt a distributed architecture and forward binary data such as a picture stream, a characteristic value and the like by using a middle key memory database, thereby reducing network overhead and improving the processing capacity of a server.
In order to achieve the above object, according to one aspect of the present invention, a distributed face recognition system based on a memory database is provided, which includes a plurality of webcams, a plurality of face detection modules, a plurality of face feature extraction modules, a plurality of face feature comparison modules, a memory database module (Redis), a distributed publish-subscribe message queue module (KAFKA), a database, and a platform management module;
the system comprises a platform management module, a network camera distribution module, a human face registration module and a registration information storage module, wherein the platform management module receives information set by a user, generates control information, and stores the control information, the network camera distribution module, the human face registration module and the registration information storage module;
the network camera module is used for providing real-time video stream as a video source for face recognition;
the face detection module is used for acquiring video frames from the real-time video stream to perform face detection and detecting the video frames with the human images; and carrying out face detection on the face registration picture;
the face feature extraction module is used for extracting features of video frames with the human images and extracting features of face registration pictures with the human images to form a registration face feature value set;
the face feature comparison module is used for comparing the extracted face feature value with a registered face feature value set;
the database is used for storing face registration information and registering a characteristic value set; redis is used for storing pictures and face characteristic values; KAFKA is used for message transmission among modules, the message has a Key value of a picture and a face characteristic value stored in Redis, the Key value is attached to the KAFKA message for transmission, and the module receiving the message acquires the picture and the face characteristic value which can be directly used from the Redis through the Key value in the KAFKA.
Preferably, in the distributed face recognition system based on the memory database, the face detection module stores the detected video frame with the portrait by means of Redis, and sends the message with the Redis Key value of the video frame with the portrait to the KAFKA queue by means of the face detection module, so that the plurality of face feature extraction modules can consume the message in a balanced manner.
Preferably, in the distributed face recognition system based on the memory database, after the face detection module detects the video frame with the face image, the face image in the video frame is stored in the Redis.
Preferably, in the distributed face recognition system based on the memory database, the face feature extraction module acquires face picture data from the Redis through the Redis Key value, extracts face features from the face picture data, and stores the extracted feature values in the Redis; and sending the message with the face feature value and the Redis Key to a KAFKA queue through a face feature extraction module so that a plurality of face feature comparison modules can consume the message in a balanced manner.
Preferably, in the distributed face recognition system based on the memory database, the face feature comparison module obtains the face feature value from Redis through Redis Key, compares the face feature value with the registered face feature value set stored in the database, and sends the comparison result to the platform management module through KAFKA.
Preferably, in the distributed face recognition system based on the memory database, the platform management module has a function of specifying a matching relationship between the face detection module and the network camera, and is configured to allocate the face detection module to the network camera, and the specified face detection module obtains a video frame from a real-time video stream provided by the network camera to perform face detection.
Preferably, in the distributed face recognition system based on the memory database, the platform management module is configured to generate a deployment control alarm according to the comparison result of the face feature comparison module and the deployment control information, and/or push the deployment control alarm and snapshot information to a WEB user; the snapshot information preferably comprises a face picture, a snapshot panoramic picture, a snapshot camera and snapshot time; the deployment information includes a camera designated for the snapshot, a designated time period, a designated database, and a feature comparison threshold.
To achieve the above objects, according to another aspect of the present invention, there is provided a memory database-based face recognition method,
the method for forwarding the binary data comprising the picture stream and the characteristic value by using a Redis internal memory database is characterized in that a Key value stored in the Redis is attached to a KAFKA message for transmission, so that a module receiving the message acquires the picture stream and the characteristic value which can be directly used from the Redis through the Redis Key value carried by the KAFKA message;
the method avoids the conversion from the picture or characteristic value to the Base64 and the reverse conversion from the Base64 to the picture or characteristic value, which are required when the data are directly transmitted through the KAFKA, and the saving and extracting speed of the binarized data of the picture and the characteristic value in the intermediate key memory database Redis is higher than that of the KAFKA, so that the method can also reduce the occupation of network resources and improve the processing capacity of a server.
Preferably, the memory database-based face recognition method specifically includes the following steps:
(1) acquiring and analyzing video stream, converting video frames into pictures, and carrying out face detection on the pictures;
(2) cutting a face picture from an original image of a video frame with a detected face, and storing the original panoramic picture and the face picture into Redis;
(3) sending a human face feature extraction request message with a panoramic image and a Redis Key corresponding to the human face image to a KAFKA human face feature extraction request message queue;
(4) the feature extraction module extracts request message queue subscription information from the KAFKA face features, and after the face detection module generates the face feature extraction request message, the feature extraction module receives the face feature extraction request message;
(5) acquiring a face picture from Redis through a Key corresponding to Redis in the face feature extraction request message;
(6) extracting the face features of the face picture, and storing the extracted face feature values into Redis;
(7) sending a feature extraction result message with a face feature value, a face picture and a panoramic picture Redis Key to a KAFKA face feature comparison request message queue;
(8) acquiring a face characteristic value from Redis through a Key corresponding to Redis in the face characteristic comparison request message, comparing the characteristic value with a registered person characteristic value acquired from a database, and sending a characteristic comparison result message to a KAFKA face characteristic comparison result message queue;
(9) and acquiring a face feature comparison result from the KAFKA face feature comparison result message queue.
Preferably, the memory database-based face recognition method further includes a face registration step, which is specifically as follows:
(a) receiving registration information uploaded by a user, wherein the registration information comprises personnel information and face pictures;
(b) storing the face picture in a database; uploading the uploaded face picture to Redis, and sending a face detection request message with a Key of Redis corresponding to the face picture to a KAFKA face detection request message queue;
(c) subscribing information from a KAFKA face detection request information queue, and acquiring a face picture from Redis through a Key corresponding to Redis in the face detection request information when a platform management module generates the face detection request information;
(d) carrying out face detection on the face picture, cutting the detected face from the picture and storing the cut face into Redis;
(e) sending a human face feature extraction request message with a Redis Key corresponding to the human face photo to a KAFKA human face feature extraction request message queue;
(f) the feature extraction module extracts a request message queue subscription message from the KAFKA face feature, and acquires a face picture from Redis through a Key corresponding to Redis in the face feature extraction request message;
(g) extracting the face features of the face picture; storing the extracted face characteristic value into Redis; sending a feature extraction result message with a Redis Key corresponding to the face feature value and a Redis Key corresponding to the face picture to a KAFKA face feature extraction result message queue;
(h) acquiring a face picture and a face characteristic value from Redis through a Key corresponding to Redis in the face characteristic extraction result message;
(i) storing the human face picture and the human image registration information into a database; and returning the registration result information to the user.
In general, compared with the prior art, the above technical solution contemplated by the present invention can achieve the following beneficial effects:
(1) the distributed face recognition system based on the memory database adopts a distributed architecture, and can perform dynamic expansion according to project requirements; the adopted expansion mode is a soft expansion mode, the task of the original system is not required to be changed, any module is not required to be restarted, the server is only required to be added according to the actual engineering requirement, a new module is deployed on the new server, the module configuration file is modified, the module is started, and the module is automatically added into the distributed system; then, a new network camera analysis task is added to a new module through the platform management module, namely, the expansion of the network camera is dynamically completed; the whole expansion process is very smooth, and the operation and maintenance are very convenient; because a distributed architecture is adopted, the data of the network camera supported by the system theoretically has no upper limit, and the capacity can be continuously expanded according to the increase of server hardware.
(2) The invention provides a distributed face recognition system and method based on a memory database, wherein binary data such as picture stream and characteristic value are forwarded by using a middle Key memory database Redis, a Key value stored in the Redis is attached to a KAFKA message for transmission, and a module receiving the message acquires the picture stream and the characteristic value from the Redis through the Redis Key value in the KAFKA message;
because the messages supported by KAFKA are all of character string types, the transmission of pictures and characteristic values between processes needs to be converted into Base64 firstly, and the messages are converted into the picture or characteristic value types from the Base64 by reverse conversion after being received by the processes; the method of the invention transmits through the Key value in Redis, thereby avoiding the program overhead caused by direct KAFKA transmission on conversion and reverse conversion, occupying network resources, improving the processing capacity of the server and reducing the network overhead; and the storing and extracting speed of the binarized data in the memory database is higher than that of KAFKA, so the processing speed can be further improved by adopting the method of the invention.
Drawings
FIG. 1 is a schematic diagram of one embodiment of a distributed face recognition system provided by the present invention;
FIG. 2 is a schematic diagram of a portrait registration process according to an embodiment of the present invention;
fig. 3 is a schematic flow chart of real-time video face recognition in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The invention discloses an embodiment of a distributed face recognition system based on a memory database, which comprises a plurality of network cameras, a plurality of face detection modules, a plurality of face feature extraction modules, a plurality of face feature comparison modules, a memory database module (Redis), a distributed publish-subscribe message queue module (KAFKA), a MySql database and a platform management module, wherein the face detection modules are connected with the face feature extraction modules; all modules are connected through a network, and communication is carried out by adopting TCP/IP;
the platform management module receives information set by a user, generates deployment and control information, performs network camera allocation, face registration and registration information storage, allocates a face detection module for the network camera, and acquires a real-time video stream from the network camera by the specified face detection module to perform face detection; the detected face picture is stored through Redis, the message with the Redis Key of the face picture is sent to a KAFKA queue through a face detection module, and the message is consumed in a balanced mode through a plurality of face feature extraction modules.
The face feature extraction module acquires face picture data from Redis through Redis Key, extracts face features from the face picture data, and stores the extracted feature values into Redis.
The information with the face feature value and the Redis Key is sent to a KAFKA queue through a face feature extraction module, and the information is consumed in a balanced mode through a plurality of face feature comparison modules.
The face feature comparison module acquires a face feature value from Redis through a Redis Key and compares the face feature value with a face feature set registered in a database; and sending the comparison result to the platform management module through KAFKA.
In a preferred embodiment, the platform management module confirms whether to generate an alarm or not through the user control information, and pushes alarm result information and snapshot information to the WEB user; the snapshot information comprises a face picture, information of a snapshot camera and snapshot time.
The following describes the above embodiment, modules in the embodiment, and signal interaction between the modules in detail with reference to fig. 1.
The network camera is used as the front-end equipment of the face recognition system, is a camera combining the traditional camera and the network technology, is used for the face recognition at present, is a network camera based on picture flow, and has the face detection function; the other is a universal network camera based on video stream, and the camera provides real-time video stream for a face detection module to carry out face detection; in the embodiment, a second camera for face detection based on video stream is adopted.
In the distributed face recognition system provided in the embodiment, the face detection module is used for face detection of a video stream on the one hand and face detection of a face registration picture on the other hand. The face detection of the video stream mainly comprises the steps of converting video frames into image data through a standard RTSP network video stream, an offline video file and a mainstream network camera video stream, detecting whether a face exists in an image or not by using an algorithm, feeding back coordinates of the face in the image, coordinates of eyes, coordinates of a mouth and coordinates of a nose, and transmitting the image including face information and the face information to a feature extraction module through KAFKA and Redis. For the face detection of the face registration picture, the algorithm is directly used for carrying out the face detection of the image, whether the face exists in the image is detected, and the coordinates of the face in the image, the coordinates of eyes, the coordinates of mouth and the coordinates of nose are fed back; and transmitting the image with the face information and the face information to a feature extraction module through KAFKA and Redis.
The feature extraction module acquires face information detected by video streaming and face information detected by face registration from KAFKA and Redis, and extracts a feature value of a face; and transmitting the characteristic value and the face picture information to a face characteristic comparison module and a platform management module through KAFKA and Redis.
The face feature comparison module loads registered face feature values from a database, obtains face feature values detected by video streaming from KAFKA and Redis, compares the face feature values detected by the real-time video streaming with a face feature value set loaded in the database, and determines identity information corresponding to a face through comparison of the feature values.
The platform management module is used for managing other modules in the system, registering human faces and controlling cameras; generating a control alarm through the comparison result of the face feature comparison module and control information, and pushing alarm result information and snapshot information to a WEB user; the snapshot information preferably includes a face picture, a snapshot panoramic picture, snapshot camera information, and a snapshot time. (ii) a The control information comprises a camera appointed for snapshot, an appointed time period, an appointed database and a characteristic comparison threshold; comparing the portrait captured by the appointed camera in the appointed time period with the registered face characteristic value set stored in the appointed database, and generating alarm information when the set threshold value is reached; the control information is issued to the face feature comparison module by the platform management module.
Redis is a database of memory key value pairs, used as a network transmission intermediate key for picture information and characteristic value information in the distributed system of the present invention.
KAFKA is a distributed publish-subscribe message queue module, which is used for receiving and distributing messages among modules to realize load balance of message processing; the message only supports a text format, and the transmission of picture files and binary data is inconvenient and can be carried out after the Base64 conversion.
And the MySQL database is used for storing configuration information, face registration information and control information in the system.
The method for face recognition based on the memory database distributed face recognition system provided by the embodiment is to forward binary data comprising a picture stream and a characteristic value by using a middle Key memory database Redis, attach a Key value stored in the Redis to a KAFKA message for transmission, so that a module receiving the message can directly acquire the usable picture stream and the characteristic value from the Redis through the Redis Key value carried by the KAFKA message.
Referring to fig. 2, the method for registering a face by using the face recognition system according to the embodiment includes the following steps:
(1) receiving registration information uploaded by a user through a platform management module, wherein the registration information comprises personnel information and face pictures;
(2) the platform management module stores the uploaded face picture into a MySQL database;
(3) the platform management module uploads the uploaded face picture to Redis and sends a face detection request message with Key of Redis corresponding to the face picture to a KAFKA face detection request message queue;
(4) the face detection module subscribes messages from the KAFKA face detection request message queue, and when the platform management module generates the face detection request message, the face detection module receives the face detection request message; the face detection module acquires a face picture from Redis through a Key corresponding to Redis in the face detection request message;
(5) the face detection module carries out face detection on the face picture through a preset face detection algorithm, cuts out the detected face from the picture and stores the cut face into Redis;
(6) the face detection module sends a face feature extraction request message with a Redis Key corresponding to the face photo to a KAFKA face feature extraction request message queue;
(7) the feature extraction module extracts request message queue subscription information from the KAFKA face features, and after the face detection module generates the face feature extraction request message, the feature extraction module receives the face feature extraction request message; the feature extraction module acquires a face picture from Redis through a Key corresponding to Redis in the face feature extraction request message;
(8) the feature extraction module extracts the face features through a face picture and a preset face feature extraction algorithm; storing the extracted face characteristic value into Redis; sending a feature extraction result message with a Redis Key corresponding to the face feature value and a Redis Key corresponding to the face picture to a KAFKA face feature extraction result message queue;
(9) the platform management module subscribes information from the KAFKA facial feature extraction result information queue, and receives the facial feature extraction result information after the facial feature extraction module generates the facial feature extraction result information;
(10) the platform management module acquires a face picture and a face characteristic value from Redis through a Key corresponding to Redis in the face characteristic extraction result message;
(11) the platform management module stores the face picture and the face registration information into a MySQL database; and returning the registration result information to the user.
Referring to fig. 3, in the embodiment, based on the distributed face recognition system based on the memory database provided in the embodiment, the method for performing face recognition based on the real-time snapshot video includes the following steps:
(1) the human image detection module acquires and analyzes video stream from the network camera, converts video frames into pictures, and performs human face detection on the pictures through a preset human face recognition algorithm;
(2) the human image detection module does not process the video frames without human faces; cutting a face picture from an original image of a video frame with a detected face, and storing the original panoramic picture and the face picture into Redis;
(3) the face detection module sends a face feature extraction request message with a panoramic image and a Redis Key corresponding to the face image to a KAFKA face feature extraction request message queue;
(4) the feature extraction module extracts request message queue subscription information from the KAFKA face features, and after the face detection module generates the face feature extraction request message, the feature extraction module receives the face feature extraction request message;
(5) the feature extraction module acquires a face picture from Redis through a Key corresponding to Redis in the face feature extraction request message;
(6) the feature extraction module extracts the face features through a face picture and a face feature extraction algorithm; storing the extracted face characteristic value into Redis;
(7) the face feature extraction module sends a feature extraction result message with a face feature value, a face picture and a panoramic picture RedisKey to a KAFKA face feature comparison request message queue;
(8) acquiring a registered personnel characteristic value and related information from MySQL when the face characteristic comparison module is initialized; the face feature comparison module subscribes messages from the KAFKA face feature comparison request message queue, and after the face feature extraction module produces the face feature comparison request messages, the feature extraction module receives the face feature comparison request messages;
(9) the face feature comparison module acquires a face feature value from Redis through a Key corresponding to Redis in the face feature comparison request message, and compares the feature value with a feature value of a registered person; sending the feature comparison result message to a KAFKA face feature comparison result message queue;
(10) the platform management module subscribes information from the KAFKA human face feature comparison result message queue, and the platform service module receives the human face feature comparison result message after the human face feature comparison module produces the human face feature comparison result message.
By the system and the method, binary data such as the picture stream, the characteristic value and the like are forwarded by using the Redis, so that the conversion and the inverse conversion required when the binary data are directly transmitted through KAFKA are avoided, the occupation of network resources can be reduced, and the processing capacity of a server is improved.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A distributed face recognition system based on a memory database is characterized by comprising a plurality of network cameras, a plurality of face detection modules, a plurality of face feature extraction modules, a plurality of face feature comparison modules, Redis, KAFKA, a database and a platform management module;
the platform management module receives user settings, generates deployment and control information, and performs network camera allocation, face registration and registration information storage;
the network camera module is used for providing real-time video stream as a video source for face recognition;
the face detection module is used for acquiring video frames from the real-time video stream to perform face detection and detecting the video frames with the human images; and carrying out face detection on the face registration picture;
the face feature extraction module is used for extracting features of video frames with the human images and extracting features of face registration pictures with the human images to form a registration face feature value set;
the face feature comparison module is used for comparing the extracted face feature value with a registered face feature value set;
the database is used for storing face registration information and a registration characteristic value set; redis is used for storing pictures and face characteristic values; KAFKA is used for message transmission among modules, and the message has a Key value of a picture and a face characteristic value stored in Redis; and attaching the Key value to the KAFKA message for transmission, and acquiring the directly usable picture and the face characteristic value from the Redis through the Key value in the KAFKA message by a module receiving the message.
2. The distributed face recognition system of claim 1, wherein the face detection module stores the detected video frames with faces via Redis, and sends a message with Redis Key value of the video frames with faces via the face detection module to a KAFKA queue for a plurality of face feature extraction modules to balance consumption messages.
3. The distributed face recognition system of claim 1 or 2, wherein the face detection module stores face images in a video frame into Redis after detecting the video frame with the face images.
4. The distributed face recognition system of claim 3, wherein the face feature extraction module obtains face picture data from Redis through a Redis Key value and extracts face features from the face picture data, and stores the extracted feature values into Redis; and sending the message with the face feature value and the Redis Key to a KAFKA queue through a face feature extraction module so that a plurality of face feature comparison modules can consume the message in a balanced manner.
5. The distributed face recognition system of claim 3, wherein the face feature comparison module obtains a face feature value from Redis via Redis Key, compares the face feature value with a set of registered face feature values stored in a database, and sends the comparison result to the platform management module via KAFKA.
6. The distributed face recognition system according to claim 1 or 2, wherein the platform management module further has a function of specifying a matching relationship between the detection module and the network camera, and is configured to assign the face detection module to the network camera, and the specified face detection module obtains video frames from a real-time video stream provided by the network camera for face detection.
7. The distributed face recognition system according to claim 1 or 2, wherein the platform management module is configured to generate a deployment control alarm according to the comparison result of the face feature comparison module and the deployment control information, and/or push the deployment control alarm and the snapshot information to a WEB user.
8. A face recognition method based on memory database is characterized in that binary data including picture stream and characteristic value are forwarded by using a middle Key memory database Redis, a Key value stored in Redis is attached to a KAFKA message for transmission, so that a module receiving the message acquires the picture stream and the characteristic value which can be directly used from the Redis through the Redis Key value carried by the KAFKA message, and conversion and inverse conversion required when the message is directly transmitted through KAFKA are avoided.
9. A face recognition method based on a memory database is characterized by comprising the following steps:
(1) acquiring and analyzing video stream, converting video frames into pictures, and carrying out face detection on the pictures;
(2) cutting a face picture from an original image of a video frame with a detected face, and storing the original panoramic picture and the face picture into Redis;
(3) sending a human face feature extraction request message with a panoramic image and a Redis Key corresponding to the human face image to a KAFKA human face feature extraction request message queue;
(4) the feature extraction module extracts request message queue subscription information from the KAFKA face features, and after the face detection module generates the face feature extraction request message, the feature extraction module receives the face feature extraction request message;
(5) acquiring a face picture from Redis through a Key corresponding to Redis in the face feature extraction request message;
(6) extracting the face features of the face picture, and storing the extracted face feature values into Redis;
(7) sending a feature extraction result message with a face feature value, a face picture and a panoramic picture Redis Key to a KAFKA face feature comparison request message queue;
(8) acquiring a face characteristic value from Redis through a Key corresponding to Redis in the face characteristic comparison request message, comparing the characteristic value with a registered person characteristic value acquired from a database, and sending a characteristic comparison result message to a KAFKA face characteristic comparison result message queue;
(9) and acquiring a face feature comparison result from the KAFKA face feature comparison result message queue.
10. A face recognition method based on memory database is characterized by also comprising face registration, comprising the following steps:
(a) receiving registration information uploaded by a user, wherein the registration information comprises personnel information and face pictures;
(b) storing the face picture in a database; uploading the uploaded face picture to Redis, and sending a face detection request message with a Key of Redis corresponding to the face picture to a KAFKA face detection request message queue;
(c) subscribing information from a KAFKA face detection request information queue, and acquiring a face picture from Redis through a Key corresponding to Redis in the face detection request information when a platform management module generates the face detection request information;
(d) carrying out face detection on the face picture, cutting the detected face from the picture and storing the cut face into Redis;
(e) sending a human face feature extraction request message with a Redis Key corresponding to the human face photo to a KAFKA human face feature extraction request message queue;
(f) the feature extraction module extracts a request message queue subscription message from the KAFKA face feature, and acquires a face picture from Redis through a Key corresponding to Redis in the face feature extraction request message;
(g) extracting the face features of the face picture; storing the extracted face characteristic value into Redis; sending a feature extraction result message with a Redis Key corresponding to the face feature value and a Redis Key corresponding to the face picture to a KAFKA face feature extraction result message queue;
(h) acquiring a face picture and a face characteristic value from Redis through a Key corresponding to Redis in the face characteristic extraction result message;
(i) storing the human face picture and the human image registration information into a database; and returning the registration result information to the user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810981766.0A CN109241111B (en) | 2018-08-27 | 2018-08-27 | Distributed face recognition system and method based on memory database |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810981766.0A CN109241111B (en) | 2018-08-27 | 2018-08-27 | Distributed face recognition system and method based on memory database |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109241111A CN109241111A (en) | 2019-01-18 |
| CN109241111B true CN109241111B (en) | 2020-10-23 |
Family
ID=65069251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810981766.0A Active CN109241111B (en) | 2018-08-27 | 2018-08-27 | Distributed face recognition system and method based on memory database |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109241111B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110111458A (en) * | 2019-04-16 | 2019-08-09 | 深圳市多度科技有限公司 | Realize method, system and door access machine that face is deployed to ensure effective monitoring and control of illegal activities |
| CN110674757B (en) * | 2019-09-25 | 2022-07-19 | 北京旷视科技有限公司 | Deployment control method, system and computer readable storage medium |
| CN111431699A (en) * | 2019-09-29 | 2020-07-17 | 杭州海康威视数字技术股份有限公司 | Method, device and system for quickly validating face authentication function |
| CN111159136B (en) * | 2019-12-27 | 2023-10-17 | 中山大学 | Face picture management and synchronization safety management system and method |
| CN113051045B (en) * | 2019-12-27 | 2022-08-16 | 南京甄视智能科技有限公司 | Method and device for dynamically balancing equipment end load under IOT platform |
| CN111291628B (en) * | 2020-01-17 | 2024-02-06 | 黄芸芸 | Face data distributed identification and storage architecture based on block chain technology |
| CN111209119A (en) * | 2020-01-21 | 2020-05-29 | 成都国翼电子技术有限公司 | Load balancing method for face snapshot rifle bolt |
| CN111314350A (en) * | 2020-02-19 | 2020-06-19 | 深圳中兴网信科技有限公司 | Image storage system, storage method, calling system and calling method |
| CN111753743B (en) * | 2020-06-28 | 2023-05-19 | 武汉虹信技术服务有限责任公司 | Face recognition method and system based on gatekeeper |
| CN111950347A (en) * | 2020-06-29 | 2020-11-17 | 武汉烽火众智数字技术有限责任公司 | Control personnel early warning system and method based on face recognition |
| CN112700214A (en) * | 2020-12-29 | 2021-04-23 | 创优数字科技(广东)有限公司 | Face attendance data processing method and device, server and storage medium |
| CN112966642A (en) * | 2021-03-23 | 2021-06-15 | 云账户技术(天津)有限公司 | Face recognition method, system and stream processing platform |
| CN113254686B (en) * | 2021-04-02 | 2023-08-01 | 青岛以萨数据技术有限公司 | Personnel behavior detection method, device and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10191768B2 (en) * | 2015-09-16 | 2019-01-29 | Salesforce.Com, Inc. | Providing strong ordering in multi-stage streaming processing |
| US10467457B2 (en) * | 2015-12-03 | 2019-11-05 | Nec Corporation Of America | System and method for capturing images used in facial recognition through effective use of exposure management |
| CN107104961B (en) * | 2017-04-21 | 2019-12-10 | 中国电子科技集团公司第二十八研究所 | distributed real-time video monitoring processing system based on ZooKeeper |
| CN108390881B (en) * | 2018-02-27 | 2021-06-15 | 北京焦点新干线信息技术有限公司 | Distributed high-concurrency real-time message pushing method and system |
-
2018
- 2018-08-27 CN CN201810981766.0A patent/CN109241111B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109241111A (en) | 2019-01-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109241111B (en) | Distributed face recognition system and method based on memory database | |
| US11983909B2 (en) | Responding to machine learning requests from multiple clients | |
| Ali et al. | RES: Real-time video stream analytics using edge enhanced clouds | |
| US20220053071A1 (en) | Preprocessing sensor data for machine learning | |
| CN101895727B (en) | Monitoring system, image capturing apparatus, analysis apparatus, and monitoring method | |
| US20180007115A1 (en) | Fog enabled telemetry embedded in real time multimedia applications | |
| US20140043480A1 (en) | Video monitoring system and method | |
| US20150242444A1 (en) | Coded image sharing system (ciss) | |
| Zhang et al. | Demo abstract: Evaps: Edge video analysis for public safety | |
| TW201516939A (en) | Method and device for inquiring user identity, method and device for acquiring user identity, and method and device for adding friend in instant messaging | |
| CN113228586A (en) | System and method for secure access to camera system | |
| CN112883011B (en) | Real-time data processing method and device | |
| US11443613B2 (en) | Real-time crime center solution with text-based tips and panic alerts | |
| US20150035934A1 (en) | Video message record terminal, video message transmitting server and method for leaving video message | |
| CN113891114B (en) | Transcoding task scheduling method and device | |
| CN113573150A (en) | Video stream processing method and device, electronic equipment and storage medium | |
| CN104170375A (en) | Architecture and system for group video distribution | |
| CN113329139B (en) | Video stream processing method, device and computer readable storage medium | |
| US10861306B2 (en) | Method and apparatus for video surveillance | |
| CN112926513A (en) | Conference sign-in method and device, electronic equipment and storage medium | |
| CN112039936B (en) | Data transmission method, first data processing equipment and monitoring system | |
| CN111212043A (en) | Multimedia file generation method and device | |
| CN108965634B (en) | Offline cross-device image printing method and system and shooting device | |
| TW201508651A (en) | Cloud-based smart monitoring system | |
| KR102505909B1 (en) | Apparatus for Detecting Object Real Time of Multi Channel Video Stream |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |