CN117522454A - Staff identification method and system - Google Patents
Staff identification method and system Download PDFInfo
- Publication number
- CN117522454A CN117522454A CN202410018772.1A CN202410018772A CN117522454A CN 117522454 A CN117522454 A CN 117522454A CN 202410018772 A CN202410018772 A CN 202410018772A CN 117522454 A CN117522454 A CN 117522454A
- Authority
- CN
- China
- Prior art keywords
- staff
- reid
- reid feature
- feature vector
- worker
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 239000013598 vector Substances 0.000 claims abstract description 231
- 238000012544 monitoring process Methods 0.000 claims abstract description 44
- 238000003062 neural network model Methods 0.000 claims abstract description 12
- 230000007935 neutral effect Effects 0.000 claims description 14
- 230000008569 process Effects 0.000 description 16
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005206 flow analysis Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a staff identification method and a system, wherein the method comprises the following steps: determining personnel in each monitoring image based on monitoring images of a plurality of monitoring points in a monitoring area, extracting corresponding ReID feature vectors by taking a personnel human body external frame area as input by utilizing a ReID model, clustering the ReID feature vectors, and clustering the ReID feature vectors of the same personnel into a class set; calculating the similarity between each class set and each ReID characteristic vector group of each worker in the worker library, wherein the ReID characteristic vector groups of the workers are stored in the worker library; determining whether the corresponding person is a worker based on the similarity between each category set and each worker's ReID feature vector set; and acquiring class sets of people which are not staff based on similarity judgment, constructing space-time vectors based on monitoring images corresponding to each class set, and judging whether the staff corresponding to the space-time vectors is staff based on a pre-trained first neural network model.
Description
Technical Field
The invention relates to the technical field of visual analysis, in particular to a staff identification method and a staff identification system.
Background
Passenger flow analysis is a method for knowing the behavior, preference and demand of customers by collecting and analyzing passenger flow data of places such as shops, scenic spots, stations and the like. Such analysis may help businesses better understand market trends and optimize products and services. In order to count effective passenger flow information, it is necessary to count and analyze the walking track of customers, the hot areas of interest, and the heat of goods, and identify staff therein to help the retail store optimize display and goods layout, providing good service.
However, since the personnel in the store are mixed, the customers are more complicated to go and get, the staff often mix between the customers, and accurate identification is difficult.
Disclosure of Invention
In view of the above, embodiments of the present invention provide a worker identification method to obviate or ameliorate one or more of the disadvantages of the prior art.
One aspect of the present invention provides a method of staff identification, the method comprising the steps of:
determining ReID feature vectors of people in each monitoring image based on monitoring images of a plurality of monitoring points in a monitoring area, clustering the ReID feature vectors, and clustering the ReID feature vectors of the same person into a class set;
calculating the similarity between each category set and each ReID feature vector group of each worker in a pre-constructed worker library, wherein each ReID feature vector group corresponding to each worker is stored in the worker library, and a plurality of ReID feature vectors corresponding to each worker are stored in each ReID feature vector group;
determining whether the person corresponding to each category set is a worker or not based on the similarity between each category set and the ReID feature vector group of each worker;
and acquiring class sets of people which are not staff based on similarity judgment, constructing space-time vectors based on monitoring images corresponding to each class set, and judging whether the staff corresponding to the space-time vectors is staff based on a pre-trained first neural network model.
By adopting the scheme, whether the personnel is a worker is firstly judged through the category set of the personnel and the pre-constructed worker library, and because the pre-constructed worker library has certain limitations, the scheme further constructs space-time vectors for the category set of the personnel which is judged not to be the worker, and whether the personnel is the worker is further judged through the space-time distinction between the worker and the customer, so that the worker mixed in the customer can be accurately judged.
In some embodiments of the present invention, in the step of calculating the similarity between each of the class sets and each of the ReID feature vector groups of the staff members in the pre-constructed staff member library, the similarity between each of the ReID feature vectors in the class sets and each of the ReID feature vector groups of the staff members is calculated.
In some embodiments of the present invention, in the step of determining whether the person corresponding to each category set is a worker based on the similarity between the person corresponding to the category set and the ReID feature vector group of each worker, it is determined whether the person corresponding to the category set corresponds to the same worker as the ReID feature vector group based on the comparison between the obtained similarities and the preset first determination threshold and second determination threshold, and if not, the ReID feature vector group of the next worker is further determined by adopting the same step.
In some embodiments of the present invention, in the step of separately calculating the similarity of each ReID feature vector in the category set to each ReID feature vector in the operator's ReID feature vector set, the similarity is calculated based on the following formula:
；
wherein,Sthe degree of similarity is indicated and,representing the transpose of the vector of any one of the ReID feature vectors in the set of classes, +.>A vector representing the composition of any one ReID feature vector of the ReID feature vector group of the staff, +.>And the inner product of the vector formed by the ReID characteristic vector in the category set and the vector formed by any one ReID characteristic vector in the ReID characteristic vector group of the staff is represented.
In some embodiments of the present invention, in the step of determining whether the person corresponding to the class set corresponds to the same worker with the ReID feature vector group based on comparing the obtained plurality of similarities with a preset first determination threshold and a preset second determination threshold, if the similarities are greater than or equal to the preset first determination threshold, an approval ticket is counted; if the similarity is smaller than or equal to a preset second judging threshold value, counting an objection ticket; if the similarity is smaller than a preset first judgment threshold value and larger than a second judgment threshold value, counting a neutral ticket; and if the proportion of the endorsed ticket in the sum of the ticket numbers of the endorsed ticket, the anti-endorsed ticket and the neutral ticket is larger than or equal to a preset first proportion threshold value and the proportion of the anti-endorsed ticket in the sum of the ticket numbers of the endorsed ticket, the anti-endorsed ticket and the neutral ticket is smaller than or equal to a preset second proportion threshold value, judging that the personnel corresponding to the class set corresponds to the same worker with the ReID feature vector group.
In some embodiments of the present invention, the step of obtaining a set of categories of persons not being staff based on similarity determination, constructing a space-time vector based on a monitored image corresponding to each set of categories, and determining whether the person corresponding to the space-time vector is a staff based on a pre-trained first neural network model includes:
judging whether personnel appear in the set time and space position based on the monitoring image, if so, counting 1 in the dimension corresponding to the time and space position, if not, counting 0 in the dimension corresponding to the time and space position, and combining the numerical values of all the dimensions to construct a space-time vector;
and inputting the space-time vector into a pre-trained first neural network model, and judging whether the person is a working person or not.
In some embodiments of the present invention, if it is determined that the person corresponding to the category set is a worker based on the space-time vector, the steps of the method include:
judging whether the ReID feature vector group stored in the current staff library is smaller than a preset number or not;
if yes, the ReID feature vector in the class set of the staff determined to be the space-time vector is used as a new ReID feature vector group and is added into a staff library;
if not, sorting based on the confidence level of the ReID feature vector group stored in the staff library, and taking the ReID feature vector in the class set of the staff determined to be the staff through the space-time vector as a new ReID feature vector group to replace the ReID feature vector group with the lowest confidence level in the staff library.
In some embodiments of the present invention, in the step of adding the ReID feature vectors in the class set of the person determined to be the staff by the space-time vector to the staff library as a new ReID feature vector group, calculating the similarity of the ReID feature vectors in the class set of the person determined to be the staff by the space-time vector in pairs, counting the similarity of each ReID feature vector to other ReID feature vectors, sorting the ReID feature vectors from high to low based on the number of times that the calculated similarity of each ReID feature vector is greater than a preset third determination threshold, combining the ReID feature vectors with a preset number of ReID feature vectors with a higher sorting rank into a new ReID feature vector group, and adding the new ReID feature vector group to the staff library.
In some embodiments of the present invention, the step of sorting based on the confidence level of the ReID feature vector sets stored in the staff member library is to count the number of times each ReID feature vector set stored in the staff member library is matched in a preset time period before counting as the confidence level.
The second aspect of the present invention also provides a staff identification system comprising a computer device comprising a processor and a memory, the memory having stored therein computer instructions for executing the computer instructions stored in the memory, the system implementing the steps implemented by the method as described above when the computer instructions are executed by the processor.
The third aspect of the present invention also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps performed by the aforementioned staff identification method.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and drawings.
It will be appreciated by those skilled in the art that the objects and advantages that can be achieved with the present invention are not limited to the above-described specific ones, and that the above and other objects that can be achieved with the present invention will be more clearly understood from the following detailed description.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate and together with the description serve to explain the invention.
FIG. 1 is a schematic diagram of one embodiment of a worker identification method of the present invention;
FIG. 2 is a schematic diagram showing steps for determining staff members through a staff member library according to the staff member identification method of the present invention;
FIG. 3 is a schematic diagram of another embodiment of the worker identification method of this invention.
Detailed Description
The present invention will be described in further detail with reference to the following embodiments and the accompanying drawings, in order to make the objects, technical solutions and advantages of the present invention more apparent. The exemplary embodiments of the present invention and the descriptions thereof are used herein to explain the present invention, but are not intended to limit the invention.
It should be noted here that, in order to avoid obscuring the present invention due to unnecessary details, only structures and/or processing steps closely related to the solution according to the present invention are shown in the drawings, while other details not greatly related to the present invention are omitted.
The method specifically comprises the following steps:
as shown in fig. 1, the present invention proposes a staff identification method, which includes the steps of:
step S100, determining ReID feature vectors of persons in each monitoring image based on monitoring images of a plurality of monitoring points in a monitoring area, clustering the ReID feature vectors, and clustering the ReID feature vectors of the same person into a class set;
in the implementation process, the monitoring area can be areas such as a mall, a store or a hotel, and each monitoring point can acquire a monitoring image through a monitoring camera.
In a specific implementation process, reID (pedestrian re-recognition) is a computer vision technology, which aims to recognize the same pedestrian under different camera angles. The essence of the method is that an algorithm is utilized to find a target in an image library, and feature extraction and matching are carried out through clothing, accessories, body states and the like of a pedestrian target, so that the method belongs to the image retrieval problem. The ReID technology is widely applied to the fields of video monitoring, intelligent transportation, face recognition and the like, and can effectively improve the intelligent level of a monitoring system and improve public safety and traffic efficiency.
In an implementation, the ReID feature vector may be calculated in a manner that includes a deep learning model.
In the implementation process, the step of clustering the ReID feature vectors of the same person into a class set can use a traditional clustering algorithm, or can assist the neural network model secondary feature optimization and then cluster in combination with the traditional clustering algorithm.
In the specific implementation process, a complete track and image snapshot of a target object under a single camera are generated through a vision-based target detection tracking technology in passenger flow analysis, and ReID feature vectors are extracted by utilizing object image snapshot for clustering, so that a more complete portrait of the target object is generated.
In the specific implementation process, a camera generates a track and an object feature vector of an object in a current video, integrates target object information of all terminal devices in the whole day, and performs clustering processing to obtain all snapshot information clustering sets of the same object.
Step S200, calculating the similarity between each category set and each ReID feature vector group of each worker in a pre-constructed worker library, wherein the ReID feature vector group corresponding to each worker is stored in the worker library, and a plurality of ReID feature vectors corresponding to each worker are stored in each ReID feature vector group;
in the specific implementation process, the characteristic that the staff has the standard clothing is utilized to establish a staff library, specifically, the previous clustering results can be manually classified, meanwhile, the staff library is recommended to be established by utilizing an algorithm, and the staff library is established to identify which of the clustering results are staff in a mode of calculating similarity.
Step S300, determining whether the person corresponding to each category set is a worker or not based on the similarity between each category set and the ReID feature vector group of each worker;
step S400, a class set of people which are not staff and are judged based on the similarity is obtained, a space-time vector is constructed based on a monitoring image corresponding to each class set, and whether the staff corresponding to the space-time vector is staff is judged based on a pre-trained first neural network model.
In a specific implementation process, in the step of constructing a space-time vector based on the monitored image corresponding to each category set, each dimension of the space-time vector corresponds to a position of time and space, the monitoring points are provided with different classification attributes, specifically, each classification attribute corresponds to at least one monitoring point, the classification attributes correspond to a gate, a cashier desk (service desk), a working area and the like, if the monitoring points comprise A, B and C, the monitoring time range is 1:00 to 24:00 per day, the time of each 1 hour is divided into 1 time slice, the whole monitoring time range is divided into 24 time slices, the space-time vector comprises 72 dimensions, each dimension corresponds to 1 time slice of one monitoring point, and the specific 1 st dimension corresponds to the 1 st time slice of the A monitoring point.
By adopting the scheme, whether the personnel is a worker is firstly judged through the category set of the personnel and the pre-constructed worker library, and because the pre-constructed worker library has certain limitations, the scheme further constructs space-time vectors for the category set of the personnel which is judged not to be the worker, and whether the personnel is the worker is further judged through the space-time distinction between the worker and the customer, so that the worker mixed in the customer can be accurately judged.
In some embodiments of the present invention, in the step of calculating the similarity between each of the class sets and each of the ReID feature vector groups of the staff members in the pre-constructed staff member library, the similarity between each of the ReID feature vectors in the class sets and each of the ReID feature vector groups of the staff members is calculated.
In some embodiments of the present invention, in the step of determining whether the person corresponding to each category set is a worker based on the similarity between the person corresponding to the category set and the ReID feature vector group of each worker, it is determined whether the person corresponding to the category set corresponds to the same worker as the ReID feature vector group based on the comparison between the obtained similarities and the preset first determination threshold and second determination threshold, and if not, the ReID feature vector group of the next worker is further determined by adopting the same step.
In the implementation process, a similarity can be calculated between each ReID feature vector in the class set and each ReID feature vector in the ReID feature vector group of the staff, and whether the staff corresponding to the class set is matched with the staff is judged based on all the calculated similarities.
In the implementation process, if the person corresponding to the category set is successfully matched with any worker, the calculation of the ReID feature vector group of other workers is stopped, and the confidence coefficient of the ReID feature vector group corresponding to the successfully matched worker is increased by 1.
In some embodiments of the present invention, in the step of separately calculating the similarity of each ReID feature vector in the category set to each ReID feature vector in the operator's ReID feature vector set, the similarity is calculated based on the following formula:
；
wherein,Sthe degree of similarity is indicated and,representing the transpose of the vector of any one of the ReID feature vectors in the set of classes, +.>A vector representing the composition of any one ReID feature vector of the ReID feature vector group of the staff, +.>And the inner product of the vector formed by the ReID characteristic vector in the category set and the vector formed by any one ReID characteristic vector in the ReID characteristic vector group of the staff is represented.
As shown in fig. 2, in some embodiments of the present invention, in the step of determining whether the person corresponding to the class set corresponds to the same worker with the ReID feature vector group based on comparing the obtained multiple similarities with the preset first determination threshold and the second determination threshold, step S311 includes counting an approval ticket if the similarities are greater than or equal to the preset first determination threshold; step S312, counting an objection ticket if the similarity is smaller than or equal to a preset second judgment threshold value; step S313, counting a neutral ticket if the similarity is smaller than a preset first judgment threshold value and larger than a second judgment threshold value; step S320, if the ratio of the sum of the numbers of the endorsement ticket, the anti-endorsement ticket, and the neutral ticket is greater than or equal to a preset first ratio threshold, and the ratio of the sum of the numbers of the anti-endorsement ticket, and the neutral ticket is less than or equal to a preset second ratio threshold, it is determined that the person corresponding to the category set corresponds to the same person as the ReID feature vector set, and step S330, the ReID feature vector set of the next person is further determined by adopting the same step.
Specifically, the first decision threshold may be 0.81, the second decision threshold may be 0.74, the first ratio threshold may be 0.5, and the second ratio threshold may be 0.2.
In the implementation process, if the calculated similarity of the class set does not meet the requirement that the proportion of the sum of the ticket numbers of the approving ticket, the disapproved ticket and the neutral ticket is larger than or equal to a preset first proportion threshold value, and if the proportion of the sum of the ticket numbers of the disapproved ticket, the disapproved ticket and the neutral ticket is smaller than or equal to a preset second proportion threshold value, judging that the person corresponding to the class set and the ReID feature vector group do not correspond to the same worker, further judging that the ReID feature vector group of the next worker is not obtained after the ReID feature vector group of all the workers is calculated, and if the matching fails.
As shown in fig. 2, in the implementation process, a similarity can be calculated between each ReID feature vector in the class set and each ReID feature vector in the ReID feature vector set of the staff, each calculated similarity is subjected to the determination in steps S311-S313 to determine the approval ticket, the disapproval ticket or the neutral ticket corresponding to the similarity, approval tickets, disapproval tickets and neutral tickets corresponding to all the similarities are counted, and step S320 is performed to determine whether the staff corresponding to the class set corresponds to the same staff as the ReID feature vector set.
By adopting the scheme, a plurality of similarities are calculated through each ReID feature vector in the class set and each ReID feature vector in the ReID feature vector group of the staff, whether the staff corresponding to the class set corresponds to the same staff corresponding to the ReID feature vector group is judged through the numerical values of the similarities, comprehensive judgment is carried out through approval ticket, counterticket or neutral ticket corresponding to each similarity, judgment can be carried out by combining all ReID feature vectors in the class set and the ReID feature vector group, and judgment accuracy is improved.
In some embodiments of the present invention, the step of obtaining a set of categories of persons not being staff based on similarity determination, constructing a space-time vector based on a monitored image corresponding to each set of categories, and determining whether the person corresponding to the space-time vector is a staff based on a pre-trained first neural network model includes:
judging whether personnel appear in the set time and space position based on the monitoring image, if so, counting 1 in the dimension corresponding to the time and space position, if not, counting 0 in the dimension corresponding to the time and space position, and combining the numerical values of all the dimensions to construct a space-time vector;
and inputting the space-time vector into a pre-trained first neural network model, and judging whether the person is a working person or not.
In the implementation process, in the step of determining whether personnel appear in a set time and space position based on the monitoring image, if so, a dimension meter corresponding to the time and space position is 1, if not, a dimension meter corresponding to the time and space position is 0, and in the step of combining numerical values of all the dimensions to construct a space-time vector, if the monitoring point comprises A, B and C, the monitoring time range is 1:00 to 24:00 per day, the time of each 1 hour is divided into 1 time slices, the whole monitoring time range is divided into 24 time slices, the space-time vector comprises 72 dimensions, each dimension corresponds to 1 time slice of one monitoring point, the 1 st dimension specifically corresponds to the 1 time slice of an A monitoring point, and if the personnel appear in the 1 st time slice of the A monitoring point, the numerical value of the 1 st dimension of the constructed space-time vector is 1; if the person is not present in the 1 st time slice of the A monitoring point, the 1 st dimension of the constructed space-time vector has a value of 0.
With the adoption of the scheme, as the active areas of the staff and the customers often differ, for example, the staff can appear in the counter, but the customers cannot; the staff often more appear in the area near the storehouse, but the customer can not, therefore, through space-time vector statistics staff's activity area, this scheme is based on the difference in the activity area of customer and staff to judge whether this staff is the staff according to the first neural network model of training in advance, can automatic discernment not count the staff in the staff storehouse of pre-constructing, reduce to staff storehouse dependence, and guarantee the accurate discernment to the staff.
As shown in fig. 3, in some embodiments of the present invention, if it is determined that the person corresponding to the category set is a worker based on the space-time vector, the steps of the method include:
step S510, judging whether the ReID feature vector group stored in the current staff library is smaller than the preset number;
step S520, if yes, taking the ReID characteristic vector in the class set of the staff member judged by the space-time vector as a new ReID characteristic vector group, and adding the ReID characteristic vector into a staff member library;
step S530, if not, sorting based on the confidence level of the ReID feature vector group stored in the staff member library, and using the ReID feature vector in the class set of the staff member determined by the space-time vector as a new ReID feature vector group to replace the ReID feature vector group with the lowest confidence level in the staff member library.
By adopting the scheme, the ReID feature vector group with the lowest confidence level is replaced through the confidence level of the ReID feature vector group stored in the staff library, and new staff can be automatically identified and replaced due to the fact that staff leave the staff in an actual scene, so that timeliness of the staff library is guaranteed, automatic identification capability of the staff library is improved, and identification accuracy is improved.
In some embodiments of the present invention, in the step of adding the ReID feature vectors in the class set of the person determined to be the staff by the space-time vector to the staff library as a new ReID feature vector group, calculating the similarity of the ReID feature vectors in the class set of the person determined to be the staff by the space-time vector in pairs, counting the similarity of each ReID feature vector to other ReID feature vectors, sorting the ReID feature vectors from high to low based on the number of times that the calculated similarity of each ReID feature vector is greater than a preset third determination threshold, combining the ReID feature vectors with a preset number of ReID feature vectors with a higher sorting rank into a new ReID feature vector group, and adding the new ReID feature vector group to the staff library.
In the implementation process, the similarity of ReID feature vectors in a class set of staff determined by the space-time vectors is calculated in pairs, the similarity between ReID feature vectors of the class set is counted, wherein each ReID feature vector calculates the similarity with each other ReID feature vector, the number of the ReID feature vectors, which is larger than a preset third judgment threshold value in the similarity of each other ReID feature vector, is counted for one ReID feature vector, the number of the ReID feature vectors, which is larger than the preset third judgment threshold value, in the calculated similarity of each ReID feature vector of the class set is counted in the same way, each ReID feature vector is ranked from high to low, the preset number of the ReID feature vectors with higher ranking rank are combined into a new ReID feature vector group, and the new ReID feature vector group is added into a staff library.
In a specific implementation process, the preset number may be 4, 5 or 6.
By adopting the scheme, the ReID feature vector in the class set corresponding to one person is found out by calculating the similarity, and the ReID feature vector which is the most representative for the person is constructed as the ReID feature vector group, so that the storage pressure of a staff library is reduced, and meanwhile, the accuracy of the ReID feature vector group is improved.
In some embodiments of the present invention, the step of sorting based on the confidence level of the ReID feature vector sets stored in the staff member library is to count the number of times each ReID feature vector set stored in the staff member library is matched in a preset time period before counting as the confidence level.
In the implementation process, if the person corresponding to the category set is successfully matched with any worker, the confidence coefficient of the ReID feature vector group corresponding to the successfully matched worker is increased by 1.
The embodiment of the invention also provides a staff identification system, which comprises computer equipment, wherein the computer equipment comprises a processor and a memory, the memory is stored with computer instructions, the processor is used for executing the computer instructions stored in the memory, and the system realizes the steps realized by the method when the computer instructions are executed by the processor.
The embodiment of the invention also provides a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, realizes the steps realized by the staff identification method. The computer readable storage medium may be a tangible storage medium such as Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disk, a removable memory disk, a CD-ROM, or any other form of storage medium known in the art.
Those of ordinary skill in the art will appreciate that the various illustrative components, systems, and methods described in connection with the embodiments disclosed herein can be implemented as hardware, software, or a combination of both. The particular implementation is hardware or software dependent on the specific application of the solution and the design constraints. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention. When implemented in hardware, it may be, for example, an electronic circuit, an Application Specific Integrated Circuit (ASIC), suitable firmware, a plug-in, a function card, or the like. When implemented in software, the elements of the invention are the programs or code segments used to perform the required tasks. The program or code segments may be stored in a machine readable medium or transmitted over transmission media or communication links by a data signal carried in a carrier wave.
It should be understood that the invention is not limited to the particular arrangements and instrumentality described above and shown in the drawings. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present invention are not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between steps, after appreciating the spirit of the present invention.
In this disclosure, features that are described and/or illustrated with respect to one embodiment may be used in the same way or in a similar way in one or more other embodiments and/or in combination with or instead of the features of the other embodiments.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, and various modifications and variations can be made to the embodiments of the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method of staff identification, the method comprising the steps of:
determining ReID feature vectors of people in each monitoring image based on monitoring images of a plurality of monitoring points in a monitoring area, clustering the ReID feature vectors, and clustering the ReID feature vectors of the same person into a class set;
calculating the similarity between each category set and each ReID feature vector group of each worker in a pre-constructed worker library, wherein each ReID feature vector group corresponding to each worker is stored in the worker library, and a plurality of ReID feature vectors corresponding to each worker are stored in each ReID feature vector group;
determining whether the person corresponding to each category set is a worker or not based on the similarity between each category set and the ReID feature vector group of each worker;
and acquiring class sets of people which are not staff based on similarity judgment, constructing space-time vectors based on monitoring images corresponding to each class set, and judging whether the staff corresponding to the space-time vectors is staff based on a pre-trained first neural network model.
2. The staff identification method according to claim 1, wherein in the step of calculating the similarity between each of the class sets and each of the ReID feature vector groups of the staff members in the pre-constructed staff member library, the similarity between each of the ReID feature vectors in the class sets and each of the ReID feature vector groups of the staff members is calculated.
3. The method according to claim 2, wherein in the step of determining whether the person corresponding to each of the class sets is a worker based on the similarity between the class set and the ReID feature vector group of the worker, it is determined whether the person corresponding to the class set corresponds to the same worker as the ReID feature vector group based on the comparison between the obtained similarities and the preset first determination threshold and second determination threshold, and if not, it is further determined that the ReID feature vector group of the next worker is the same step.
4. A worker recognition method according to claim 2 or 3, wherein in the step of calculating the similarity of each ReID feature vector in the category set and each ReID feature vector in the ReID feature vector group of the worker, respectively, the similarity is calculated based on the following formula:
；
wherein,Sthe degree of similarity is indicated and,represents a transpose of the vector made up of any of the ReID feature vectors in the set of classes,a vector formed by any one of the ReID feature vectors in the ReID feature vector group representing the worker.
5. The staff member identification method according to claim 3, wherein in the step of determining whether the staff member corresponding to the class set corresponds to the same staff member as the ReID feature vector group based on the plurality of calculated similarities being compared with a preset first determination threshold and a second determination threshold, if the similarities are greater than or equal to the preset first determination threshold, an approval ticket is counted; if the similarity is smaller than or equal to a preset second judging threshold value, counting an objection ticket; if the similarity is smaller than a preset first judgment threshold value and larger than a second judgment threshold value, counting a neutral ticket; and if the proportion of the endorsed ticket in the sum of the ticket numbers of the endorsed ticket, the anti-endorsed ticket and the neutral ticket is larger than or equal to a preset first proportion threshold value and the proportion of the anti-endorsed ticket in the sum of the ticket numbers of the endorsed ticket, the anti-endorsed ticket and the neutral ticket is smaller than or equal to a preset second proportion threshold value, judging that the personnel corresponding to the class set corresponds to the same worker with the ReID feature vector group.
6. The method of claim 1, wherein the step of obtaining a set of classes of people not being staff based on the similarity determination, constructing a spatio-temporal vector based on the monitored image corresponding to each set of classes, and determining whether the person corresponding to the spatio-temporal vector is a staff based on the pre-trained first neural network model comprises:
judging whether personnel appear in the set time and space position based on the monitoring image, if so, counting 1 in the dimension corresponding to the time and space position, if not, counting 0 in the dimension corresponding to the time and space position, and combining the numerical values of all the dimensions to construct a space-time vector;
and inputting the space-time vector into a pre-trained first neural network model, and judging whether the person is a working person or not.
7. The method for identifying staff members according to claim 1, wherein if it is determined that the person corresponding to the category set is a staff member based on the space-time vector, the method comprises the steps of:
judging whether the ReID feature vector group stored in the current staff library is smaller than a preset number or not;
if so, the ReID feature vector in the class set of the staff member determined by the space-time vector is used as a new ReID feature vector group and is added into the staff member library.
8. The staff identification method according to claim 7, wherein in the step of adding the ReID feature vectors in the class set of the staff member determined as the staff member by the space-time vector as a new ReID feature vector group to the staff member library, the similarity of the ReID feature vectors in the class set of the staff member determined as the staff member by the space-time vector is calculated two by two, the similarity of each ReID feature vector to other ReID feature vectors is counted, and the respective ReID feature vectors are sorted from high to low based on the number of times greater than a preset third determination threshold value in the calculated similarity of each ReID feature vector, and the newly ReID feature vector group is combined with the ReID feature vectors with higher sorting rank, and added to the staff member library.
9. The staff identification method as claimed in claim 7, wherein the step of sorting based on the confidence level of the ReID feature vector group stored in the staff library is to count the number of times each ReID feature vector group stored in the staff library is matched in a preset time period before counting as the confidence level.
10. A staff identification system, characterized in that it comprises a computer device comprising a processor and a memory, said memory having stored therein computer instructions for executing the computer instructions stored in said memory, which system, when executed by the processor, realizes the steps realized by the method according to any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410018772.1A CN117522454B (en) | 2024-01-05 | 2024-01-05 | Staff identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410018772.1A CN117522454B (en) | 2024-01-05 | 2024-01-05 | Staff identification method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117522454A true CN117522454A (en) | 2024-02-06 |
| CN117522454B CN117522454B (en) | 2024-04-16 |
Family
ID=89755369
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410018772.1A Active CN117522454B (en) | 2024-01-05 | 2024-01-05 | Staff identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117522454B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506700A (en) * | 2017-08-07 | 2017-12-22 | 苏州经贸职业技术学院 | Pedestrian's recognition methods again based on the study of broad sense similarity measurement |
| CN110941978A (en) * | 2019-05-23 | 2020-03-31 | 罗普特科技集团股份有限公司 | Face clustering method and device for unidentified personnel and storage medium |
| US20200134321A1 (en) * | 2018-02-12 | 2020-04-30 | Beijing Sensetime Technology Development Co., Ltd. | Pedestrian re-identification methods and apparatuses, electronic devices, and storage media |
| WO2020098158A1 (en) * | 2018-11-14 | 2020-05-22 | 平安科技（深圳）有限公司 | Pedestrian re-recognition method and apparatus, and computer readable storage medium |
| CN111860471A (en) * | 2020-09-21 | 2020-10-30 | 之江实验室 | Work clothes wearing identification method and system based on feature retrieval |
| CN112257615A (en) * | 2020-10-26 | 2021-01-22 | 上海数川数据科技有限公司 | Clustering-based customer number statistical method |
| CN112613496A (en) * | 2021-03-08 | 2021-04-06 | 深圳市安软科技股份有限公司 | Pedestrian re-identification method and device, electronic equipment and storage medium |
| CN114491148A (en) * | 2022-04-14 | 2022-05-13 | 武汉中科通达高新技术股份有限公司 | Target person searching method and device, computer equipment and storage medium |
| CN114565894A (en) * | 2022-03-03 | 2022-05-31 | 成都佳华物链云科技有限公司 | Work garment identification method and device, electronic equipment and storage medium |
| CN114639165A (en) * | 2022-03-16 | 2022-06-17 | 平安科技（深圳）有限公司 | Pedestrian re-identification method, device, equipment and storage medium based on artificial intelligence |
| CN116894962A (en) * | 2023-07-27 | 2023-10-17 | 云从科技集团股份有限公司 | Clustering method, clustering system, control device and readable storage medium |
-
2024
- 2024-01-05 CN CN202410018772.1A patent/CN117522454B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506700A (en) * | 2017-08-07 | 2017-12-22 | 苏州经贸职业技术学院 | Pedestrian's recognition methods again based on the study of broad sense similarity measurement |
| US20200134321A1 (en) * | 2018-02-12 | 2020-04-30 | Beijing Sensetime Technology Development Co., Ltd. | Pedestrian re-identification methods and apparatuses, electronic devices, and storage media |
| WO2020098158A1 (en) * | 2018-11-14 | 2020-05-22 | 平安科技（深圳）有限公司 | Pedestrian re-recognition method and apparatus, and computer readable storage medium |
| CN110941978A (en) * | 2019-05-23 | 2020-03-31 | 罗普特科技集团股份有限公司 | Face clustering method and device for unidentified personnel and storage medium |
| CN111860471A (en) * | 2020-09-21 | 2020-10-30 | 之江实验室 | Work clothes wearing identification method and system based on feature retrieval |
| CN112257615A (en) * | 2020-10-26 | 2021-01-22 | 上海数川数据科技有限公司 | Clustering-based customer number statistical method |
| CN112613496A (en) * | 2021-03-08 | 2021-04-06 | 深圳市安软科技股份有限公司 | Pedestrian re-identification method and device, electronic equipment and storage medium |
| CN114565894A (en) * | 2022-03-03 | 2022-05-31 | 成都佳华物链云科技有限公司 | Work garment identification method and device, electronic equipment and storage medium |
| CN114639165A (en) * | 2022-03-16 | 2022-06-17 | 平安科技（深圳）有限公司 | Pedestrian re-identification method, device, equipment and storage medium based on artificial intelligence |
| CN114491148A (en) * | 2022-04-14 | 2022-05-13 | 武汉中科通达高新技术股份有限公司 | Target person searching method and device, computer equipment and storage medium |
| CN116894962A (en) * | 2023-07-27 | 2023-10-17 | 云从科技集团股份有限公司 | Clustering method, clustering system, control device and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117522454B (en) | 2024-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11275932B2 (en) | Human body attribute recognition method, apparatus, and device and medium | |
| CN106776619B (en) | Method and device for determining attribute information of target object | |
| US7136524B1 (en) | Robust perceptual color identification | |
| CN108304816B (en) | Identity recognition method and device, storage medium and electronic equipment | |
| CN109977132B (en) | Student abnormal behavior pattern analysis method based on unsupervised clustering pattern | |
| CN113674037B (en) | Data acquisition and recommendation method based on shopping behaviors | |
| CN109902644A (en) | Face identification method, device, equipment and computer-readable medium | |
| US20220084102A1 (en) | Commodity recommendation method, server, shopping cart and shopping system | |
| CN102831615A (en) | Object monitoring method and device as well as monitoring system operating method | |
| CN111696080B (en) | Face fraud detection method, system and storage medium based on static texture | |
| US11023714B2 (en) | Suspiciousness degree estimation model generation device | |
| US20190213610A1 (en) | Evaluation device and evaluation method | |
| CN115861940B (en) | Work scene behavior evaluation method and system based on human body tracking and recognition technology | |
| US20170053172A1 (en) | Image processing apparatus, and image processing method | |
| CN112700307A (en) | Commodity recommendation method based on data analysis | |
| AU2017231602A1 (en) | Method and system for visitor tracking at a POS area | |
| CN110348516B (en) | Data processing method, data processing device, storage medium and electronic equipment | |
| CN112131477A (en) | Library book recommendation system and method based on user portrait | |
| CN117522454B (en) | Staff identification method and system | |
| EP4066197A1 (en) | System and method for interactive perception and content presentation | |
| CN112258243A (en) | Accurate marketing evaluation system based on visual perception and operation data quantification | |
| CN112580397A (en) | Data processing method, device and storage medium | |
| CN115392351A (en) | Risk user identification method and device, electronic equipment and storage medium | |
| CN114694008A (en) | Remote face recognition system | |
| CN114491156A (en) | Method for intelligently pushing approximate object based on video image |
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 | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 1602, 16th Floor, Building 4, Zone 4, No. 81 Beiqing Road, Haidian District, Beijing, 100094 Patentee after: BEIJING VION INTELLIGENT TECHNOLOGY Co.,Ltd. Country or region after: China Address before: Unit 801, Unit 5, No. 2 Building, 9 Yuan, Fenghao East Road, Haidian District, Beijing 100094 Patentee before: BEIJING VION INTELLIGENT TECHNOLOGY Co.,Ltd. Country or region before: China |