Intelligent visitor management system for AI intelligent community

The AI-powered smart community intelligent visitor management system utilizes facial recognition and distributed access control modules to solve the problems of personal information leakage and access trajectory monitoring in community access control systems, achieving accurate visitor management and security protection.

CN119049166APending Publication Date: 2024-11-29ANHUI HUOJIXUNZONG TECHNOLOGY CO LTD

Patent Information

Application Number
CN202411318504.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-20
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

In existing community access control systems, facial recognition is prone to personal information leakage, cannot effectively monitor the actual visitor's access trajectory, and the problem of inconsistency between the registered access information and the information of the buildings visited is difficult to solve.

Method used

The AI-powered smart community intelligent visitor management system uses a facial recognition module to identify visitor facial images, generate recommended access routes, and utilizes a distributed access control module and a trajectory vector generation module to identify abnormal trajectories, monitor visitor access trajectories in real time, protect personal information security, and ensure the consistency of access information.

Benefits of technology

It achieves accurate identification and information protection of visitor facial images, ensures consistency between access information and building information, effectively monitors visitor trajectories, prevents information leakage, and promptly reports abnormal situations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119049166A_ABST
    Figure CN119049166A_ABST
Patent Text Reader

Abstract

The invention relates to visitor management, in particular to an AI smart community intelligent visitor management system, which is characterized in that if a visitor is an unregistered person, an access information acquisition module acquires access information registered by the visitor and sends the access information to a database through a server; the access path generation module generates a corresponding recommended access path according to the access information and sends the recommended access path to the database by the server, meanwhile, the server sends the recommended access path to the visitor terminal through the wireless communication module, and a manager issues a visiting card to the visitor and inputs the information of the visiting card to the database; the database associates and stores the face recognition result, the access information, the recommended access path and the access card information of the visitor, and an entrance access control module opens an entrance access; according to the technical scheme provided by the invention, the defects that personal information leakage is easily caused by face recognition, real access tracks of visitors cannot be supervised, and inconsistent conditions of registered access information and visited building information cannot be effectively managed can be overcome.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to visitor management, and in particular to an AI smart community intelligent visitor management system. Background Art

[0002] The safety of the living environment has gradually become the primary choice of the owners, and the access control system in the community plays a very important role in the community, which is related to the personal and property safety of the residents of the entire community. In order to manage the access control of the community more efficiently and reliably, the access control systems of many communities use the face recognition mode to replace the traditional access control card mode. This change greatly facilitates the management of residents and visitors entering and leaving the community, and brings a certain degree of protection to the community security.

[0003] However, although the above-mentioned changes in recognition mode have facilitated the management of residents and visitors entering and leaving the community to a certain extent, face recognition can easily lead to personal information leakage, and it is difficult to effectively solve the problems such as the inability to monitor the actual visit trajectory of visitors, and the inconsistency between the registered visit information and the visited building information. Therefore, the existing community access control system still has a lot of room for improvement. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the above-mentioned shortcomings of the prior art, the present invention provides an AI smart community intelligent visitor management system, which can effectively overcome the defects of the prior art that face recognition can easily lead to personal information leakage, cannot monitor the actual visit trajectory of visitors, and cannot effectively manage the inconsistency between registered visit information and visited building information.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] An AI smart community intelligent visitor management system includes a server, wherein the server collects visitor face images at the entrance through an entrance face image collection module, and uses a face recognition module to perform face recognition on the visitor face images at the entrance based on a database to determine whether the visitor is a registered person;

[0009] If the visitor is a registered person, the entrance access control module opens the entrance access; if the visitor is an unregistered person, the access information collection module collects the visitor's registered access information and sends it to the database by the server, the access path generation module generates a corresponding recommended access path according to the access information and sends it to the database by the server, and at the same time, the server sends the recommended access path to the visitor terminal through the wireless communication module, and the management personnel issues a visit card to the visitor and inputs the visit card information into the database, the database associates and stores the visitor's face recognition result, access information, recommended access path and visit card information, and the entrance access control module opens the entrance access;

[0010] The server collects visitor images at each building through a distributed face image collection module, and uses a face recognition module to perform face recognition on visitor face images at each building based on a database. The distributed access control module determines whether to open the building access control according to the face recognition result;

[0011] The server reads the location information of the visitor card in real time through the distributed location information reading module, and uses the access trajectory generation module to generate a corresponding access trajectory according to the real-time location information of the visitor card. The trajectory vector generation module generates an access trajectory trajectory vector and a recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path. The abnormal trajectory judgment module performs abnormal trajectory judgment on the access trajectory according to the access trajectory trajectory vector and the recommended access path trajectory vector.

[0012] Preferably, the face recognition module performs face recognition on the visitor's face image based on the database, including:

[0013] S11, extracting facial feature information from the visitor's facial image;

[0014] S12, encoding the facial feature information into a fixed-length feature digital sequence capable of representing the main features of the visitor's facial image;

[0015] S13, comparing the characteristic digital sequence of the visitor's facial image with the characteristic digital sequence of the facial image of the registered person pre-stored in the database to obtain a comparison result;

[0016] S14. Perform facial recognition on the visitor's facial image according to the comparison result, and delete the visitor's facial image.

[0017] Preferably, in S13, the characteristic digital sequence of the visitor's facial image is compared with the characteristic digital sequence of the facial image of the registered person pre-stored in the database to obtain the comparison result, including:

[0018] S131, feature digital sequence s of visitor face image a aThe characteristic digital sequence S of the registered person's face image i pre-stored in the database i Compare and get the comparison result C a,i :

[0019]

[0020] Among them, s a,j is the characteristic digital sequence s of visitor face image a a The jth position, S i,j is the characteristic digital sequence S of the registered person's face image i i The jth position of , ⊙ is the XOR operator, m is the number of digits in the feature digital sequence, i∈{1,2,...,n}, n is the number of registered person face images pre-stored in the database;

[0021] S132, repeat S131 until the characteristic digital sequence s of the visitor's face image a is a The feature digital sequence comparison is completed with all pre-stored facial images of registered persons in the database, and multiple comparison results are obtained.

[0022] Preferably, performing face recognition on the visitor's face image according to the comparison result in S14 includes:

[0023] If there is a comparison result greater than a first preset threshold value among the multiple comparison results, the visitor is determined to be a registered person, otherwise the visitor is determined to be an unregistered person.

[0024] Preferably, the distributed access control module determines whether to open the building access control according to the face recognition result, including:

[0025] If the visitor is a registered person, the distributed access control module opens the building access control. Otherwise, the corresponding access information is obtained from the database based on the face recognition result of the visitor to determine whether the access information is consistent with the building visited;

[0026] If the access information is consistent with the visited building, the distributed access control module opens the building access and records the visitor's arrival and departure time. Otherwise, the server sends the visitor's face recognition result and visited building information to the management center through the wireless communication module.

[0027] Preferably, the system also includes a building access security judgment module, which calculates the access security index of each building based on the visitor access situation of a single building and the visitor access situation of all buildings, and judges the access security situation of each building based on the access security index.

[0028] Preferably, the building access security judgment module calculates the access security index of each building according to the visitor access situation of a single building and the visitor access situation of all buildings, and judges the access security situation of each building according to the access security index, including:

[0029] S21. Calculate the access security index S of building b based on the visitor access situation of a single building and the visitor access situation of all buildings. b :

[0030]

[0031] Among them, T b,d is the d-th visit residence time data of building b, T b,d =t' b,d -t b,d , t' b,d is the visitor's departure time, t b,d is the visitor arrival time, P b is the number of visitors to building b, d∈{1,2,...,D b}, D b is the amount of visit and residence time data of building b, b∈{1,2,...,B}, B is the number of buildings;

[0032] S22, repeat S21 until the access security indexes of all buildings are calculated;

[0033] S23. If the access security index of the building is less than the second preset threshold, it is determined that the building is currently in an access security state; otherwise, it is determined that the building is currently in a frequent access state, and the server sends the building information to the management center through the wireless communication module.

[0034] Preferably, the trajectory vector generation module generates an access trajectory vector and a recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path, including:

[0035] S31, collecting the original data of the access trajectory and the corresponding recommended access path, performing data cleaning on the original data, and performing segmentation processing on the access trajectory and the corresponding recommended access path;

[0036] S32, extracting the spatial features, temporal features and correlation features of each sub-trajectory in the access trajectory to obtain trajectory feature information of each sub-trajectory;

[0037] S33, inputting the trajectory feature information of each sub-trajectory into the trained trajectory vector generation model to obtain a numerical vector of a fixed length corresponding to each sub-trajectory, i.e., a trajectory vector;

[0038] S34, performing dimension reduction and normalization processing on the trajectory vectors corresponding to each sub-trajectory, and combining the processed trajectory vectors corresponding to each sub-trajectory to form an access trajectory vector;

[0039] S35, repeat S32 to S34 to obtain a recommended access path trajectory vector;

[0040] The original data includes position coordinates and timestamps, the spatial features include the starting position coordinates, end position coordinates, length and direction of the sub-trajectory, the temporal features include the starting timestamp, end timestamp and time interval of the sub-trajectory, and the associated features include the speed change, acceleration and turning angle of the sub-trajectory.

[0041] Preferably, before inputting the trajectory feature information of each sub-trajectory into the trained trajectory vector generation model in S33, the following steps are included:

[0042] A trajectory vector generation model is constructed based on the recurrent neural network (RNN), and the trajectory vector generation model is trained:

[0043] S331, constructing a data set, and dividing the data set into a training set, a validation set, and a test set according to a preset ratio;

[0044] S332, setting the loss function and optimizer of the trajectory vector generation model, and inputting the training set into the trajectory vector generation model for model training;

[0045] S333, calculating the loss value based on the loss function, and the optimizer updates the model parameters according to the loss value and the network gradient information;

[0046] S334, if the loss value is less than the third preset threshold, the model training ends, and the current trajectory vector generation model is the trained trajectory vector generation model, otherwise, the process returns to S332, and the model training continues using the training set;

[0047] S335, inputting the validation set into the trained trajectory vector generation model, evaluating the generalization ability of the model by observing the performance on the validation set, and tuning the hyperparameters and structure of the model;

[0048] S336. Input the test set into the optimized trajectory vector generation model to evaluate the model performance.

[0049] Preferably, the abnormal trajectory judgment module performs abnormal trajectory judgment on the access trajectory according to the access trajectory vector and the recommended access path trajectory vector, including:

[0050] S41, calculating the similarity between the trajectory vectors of the same sub-trajectory in the access trajectory trajectory vector and the recommended access path trajectory vector, and screening out the target sub-trajectory whose similarity is less than a fourth preset threshold;

[0051] S42, matching the access trajectory and the target sub-trajectory in the corresponding recommended access path, and counting the distances of unmatched sections in the access trajectory;

[0052] S43, calculating the abnormality judgment index A of the access trajectory according to the unmatched road section distances and their corresponding travel times in the access trajectory:

[0053]

[0054] Among them, l x is the distance of the unmatched road section in the x-th target sub-trajectory, L x is the length of the x-th target sub-trajectory, t x is the moving time corresponding to the mismatched road section distance in the x-th target sub-trajectory, T x is the time interval of the x-th target sub-trajectory, T x =T x2 -T x1 , T x2 is the endpoint timestamp of the x-th target sub-trajectory, T x1 is the starting timestamp of the x-th target sub-trajectory, and N is the number of target sub-trajectories;

[0055] S44: If the abnormality judgment index of the access trajectory is less than the fifth preset threshold, the access trajectory is judged to be an abnormal trajectory, and the server sends the access trajectory and its corresponding face recognition result and access information to the management center through the wireless communication module.

[0056] (III) Beneficial effects

[0057] Compared with the prior art, the AI ​​smart community intelligent visitor management system provided by the present invention has the following beneficial effects:

[0058] 1) When performing face recognition on visitor face images at the entrance and visitor personnel images at each building, extract facial feature information from the visitor face images, encode the facial feature information into a fixed-length feature digital sequence that can characterize the main features of the visitor face image, quantize the visitor face image into a feature digital sequence, and compare the feature digital sequence to achieve accurate face recognition. At the same time, delete the visitor face image in a timely manner after face recognition to prevent personal information leakage and effectively protect personal information security;

[0059] 2) After the distributed face image acquisition module collects the visitor's image at each building, if the visitor is an unregistered person, the distributed access control module obtains the corresponding access information from the database according to the face recognition result of the visitor, and determines whether the access information is consistent with the visited building. The building access control will be opened only when the access information is consistent with the visited building, so that the inconsistency between the registered access information and the visited building information can be effectively managed;

[0060] 3) The distributed location information reading module reads the location information of the visitor card in real time. The access trajectory generation module generates the corresponding access trajectory according to the real-time location information of the visitor card. The trajectory vector generation module generates the access trajectory trajectory vector and the recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path. The abnormal trajectory judgment module makes an abnormal trajectory judgment on the access trajectory according to the access trajectory trajectory vector and the recommended access path trajectory vector. The server sends the abnormal trajectory judgment result to the management center through the wireless communication module. The trajectory vector can be used to make an accurate abnormal trajectory judgment on the access trajectory. When the access trajectory is an abnormal trajectory, it will be reported to the management center in time, so as to realize the effective supervision of the visitor's real access trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0062] Figure 1 It is a system schematic diagram of the present invention;

[0063] Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] An AI smart community intelligent visitor management system, such as Figure 1 and Figure 2As shown, it includes a server, which collects the facial images of visitors at the entrance through the entrance facial image collection module, and uses the facial recognition module to perform facial recognition on the facial images of visitors at the entrance based on the database to determine whether the visitor is a registered person;

[0066] If the visitor is a registered person, the entrance access control module opens the entrance access; if the visitor is an unregistered person, the access information collection module collects the visitor's registered access information and sends it to the database by the server, the access path generation module generates a corresponding recommended access path based on the access information and sends it to the database by the server, and the server sends the recommended access path to the visitor terminal through the wireless communication module, and the manager issues a visit card to the visitor and inputs the visit card information into the database, the database associates and stores the visitor's face recognition result, access information, recommended access path and visit card information, and the entrance access control module opens the entrance access;

[0067] The server collects images of visitors in each building through the distributed face image acquisition module, and uses the face recognition module to perform face recognition on the face images of visitors in each building based on the database. The distributed access control module determines whether to open the building access according to the face recognition results.

[0068] ① The face recognition module performs face recognition on visitor face images based on the database, including:

[0069] S11, extracting facial feature information from the visitor's facial image;

[0070] S12, encoding the facial feature information into a fixed-length feature digital sequence capable of representing the main features of the visitor's facial image;

[0071] S13, comparing the characteristic digital sequence of the visitor's facial image with the characteristic digital sequence of the facial image of the registered person pre-stored in the database to obtain a comparison result;

[0072] S14. Perform facial recognition on the visitor's facial image according to the comparison result, and delete the visitor's facial image.

[0073] 1) In S13, the characteristic digital sequence of the visitor's facial image is compared with the characteristic digital sequence of the facial image of the registered person pre-stored in the database to obtain the comparison result, including:

[0074] S131, feature digital sequence s of visitor face image a a The characteristic digital sequence S of the registered person's face image i pre-stored in the database i Compare and get the comparison result C a,i :

[0075]

[0076] Among them, s a,j is the characteristic digital sequence s of visitor face image a a The jth position, S i,j is the characteristic digital sequence S of the registered person's face image i i The jth position of , ⊙ is the XOR operator, m is the number of digits in the feature digital sequence, i∈{1,2,...,n}, n is the number of registered person face images pre-stored in the database;

[0077] S132, repeat S131 until the characteristic digital sequence s of the visitor's face image a is a The feature digital sequence comparison is completed with all pre-stored facial images of registered persons in the database, and multiple comparison results are obtained.

[0078] 2) In S14, facial recognition is performed on the visitor's facial image according to the comparison result, including:

[0079] If there is a comparison result greater than a first preset threshold value among the multiple comparison results, the visitor is determined to be a registered person, otherwise the visitor is determined to be an unregistered person.

[0080] ②The distributed access control module determines whether to open the building access control based on the face recognition results, including:

[0081] If the visitor is a registered person, the distributed access control module opens the building access control. Otherwise, the corresponding access information is obtained from the database based on the face recognition result of the visitor to determine whether the access information is consistent with the building visited;

[0082] If the access information is consistent with the visited building, the distributed access control module opens the building access and records the visitor's arrival and departure time. Otherwise, the server sends the visitor's face recognition result and visited building information to the management center through the wireless communication module.

[0083] The above technical scheme, when performing face recognition on the visitor face images at the entrance and the visitor personnel images at each building, extracts facial feature information from the visitor face images, encodes the facial feature information into a feature digital sequence of a fixed length that can characterize the main features of the visitor face image, quantizes the visitor face image into a feature digital sequence, and achieves accurate face recognition by comparing the feature digital sequence. At the same time, the visitor face image is deleted in time after face recognition to prevent the leakage of personal information, thereby effectively protecting the security of personal information.

[0084] In addition, after the distributed face image acquisition module collects the images of visitors in each building, if the visitor is an unregistered person, the distributed access control module obtains the corresponding access information from the database based on the face recognition result of the visitor, and determines whether the access information is consistent with the visited building. The building access control will only be opened when the access information is consistent with the visited building, thereby effectively managing the inconsistency between the registered access information and the visited building information.

[0085] In the technical solution of this application, on the basis of the distributed access control module recording the arrival time and departure time of visitors, it is also possible to accurately judge the access security situation of each building. The specific technical solution is as follows:

[0086] The system also includes a building access security judgment module, which calculates the access security index of each building based on the visitor access situation of a single building and the visitor access situation of all buildings, and judges the access security situation of each building based on the access security index, specifically including:

[0087] S21. Calculate the access security index S of building b based on the visitor access situation of a single building and the visitor access situation of all buildings. b :

[0088]

[0089] Among them, T b,d is the d-th visit residence time data of building b, T b,d =t' b,d -t b,d , t' b,d is the visitor's departure time, t b,d is the visitor arrival time, P b is the number of visitors to building b, d∈{1,2,...,D b}, D b is the amount of visit and residence time data of building b, b∈{1,2,...,B}, B is the number of buildings;

[0090] S22, repeat S21 until the access security indexes of all buildings are calculated;

[0091] S23. If the access security index of the building is less than the second preset threshold, it is determined that the building is currently in an access security state; otherwise, it is determined that the building is currently in a frequent access state, and the server sends the building information to the management center through the wireless communication module.

[0092] like Figure 1 and Figure 2As shown, the server reads the location information of the visitor card in real time through the distributed location information reading module, and uses the access trajectory generation module to generate the corresponding access trajectory according to the real-time location information of the visitor card. The trajectory vector generation module generates an access trajectory trajectory vector and a recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path. The abnormal trajectory judgment module performs abnormal trajectory judgment on the access trajectory according to the access trajectory trajectory vector and the recommended access path trajectory vector.

[0093] ① The trajectory vector generation module generates access trajectory vectors and recommended access path trajectory vectors according to the access trajectory and its corresponding recommended access path, including:

[0094] S31, collecting the original data of the access trajectory and the corresponding recommended access path, performing data cleaning on the original data, and performing segmentation processing on the access trajectory and the corresponding recommended access path;

[0095] S32, extracting the spatial features, temporal features and correlation features of each sub-trajectory in the access trajectory to obtain trajectory feature information of each sub-trajectory;

[0096] S33, inputting the trajectory feature information of each sub-trajectory into the trained trajectory vector generation model to obtain a numerical vector of a fixed length corresponding to each sub-trajectory, i.e., a trajectory vector;

[0097] S34, performing dimension reduction and normalization processing on the trajectory vectors corresponding to each sub-trajectory, and combining the processed trajectory vectors corresponding to each sub-trajectory to form an access trajectory vector;

[0098] S35, repeat S32 to S34 to obtain a recommended access path trajectory vector;

[0099] The original data includes position coordinates and timestamps, the spatial features include the starting position coordinates, end position coordinates, length and direction of the sub-trajectory, the temporal features include the starting timestamp, end timestamp and time interval of the sub-trajectory, and the associated features include the speed change, acceleration and turning angle of the sub-trajectory.

[0100] Before inputting the trajectory feature information of each sub-trajectory into the trained trajectory vector generation model in S33, the following steps are included:

[0101] A trajectory vector generation model is constructed based on the recurrent neural network (RNN), and the trajectory vector generation model is trained:

[0102] S331, constructing a data set, and dividing the data set into a training set, a validation set, and a test set according to a preset ratio;

[0103] S332, setting the loss function and optimizer of the trajectory vector generation model, and inputting the training set into the trajectory vector generation model for model training;

[0104] S333, calculating the loss value based on the loss function, and the optimizer updates the model parameters according to the loss value and the network gradient information;

[0105] S334, if the loss value is less than the third preset threshold, the model training ends, and the current trajectory vector generation model is the trained trajectory vector generation model, otherwise, the process returns to S332, and the model training continues using the training set;

[0106] S335, inputting the validation set into the trained trajectory vector generation model, evaluating the generalization ability of the model by observing the performance on the validation set, and tuning the hyperparameters and structure of the model;

[0107] S336. Input the test set into the optimized trajectory vector generation model to evaluate the model performance.

[0108] ② The abnormal trajectory judgment module judges the abnormal trajectory of the access trajectory according to the access trajectory vector and the recommended access path trajectory vector, including:

[0109] S41, calculating the similarity between the trajectory vectors of the same sub-trajectory in the access trajectory trajectory vector and the recommended access path trajectory vector, and screening out the target sub-trajectory whose similarity is less than a fourth preset threshold;

[0110] S42, matching the access trajectory and the target sub-trajectory in the corresponding recommended access path, and counting the distances of unmatched sections in the access trajectory;

[0111] S43, calculating the abnormality judgment index A of the access trajectory according to the unmatched road section distances and their corresponding travel times in the access trajectory:

[0112]

[0113] Among them, l x is the distance of the unmatched road section in the x-th target sub-trajectory, L x is the length of the x-th target sub-trajectory, t x is the moving time corresponding to the mismatched road section distance in the x-th target sub-trajectory, T x is the time interval of the x-th target sub-trajectory, T x =T x2 -T x1 , T x2 is the endpoint timestamp of the x-th target sub-trajectory, T x1 is the starting timestamp of the x-th target sub-trajectory, and N is the number of target sub-trajectories;

[0114] S44: If the abnormality judgment index of the access trajectory is less than the fifth preset threshold, the access trajectory is judged to be an abnormal trajectory, and the server sends the access trajectory and its corresponding face recognition result and access information to the management center through the wireless communication module.

[0115] In the above technical solution, the distributed location information reading module reads the location information of the visitor card in real time, the access trajectory generation module generates the corresponding access trajectory according to the real-time location information of the visitor card, the trajectory vector generation module generates the access trajectory trajectory vector and the recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path, the abnormal trajectory judgment module performs abnormal trajectory judgment on the access trajectory according to the access trajectory trajectory vector and the recommended access path trajectory vector, the server sends the abnormal trajectory judgment result to the management center through the wireless communication module, and the trajectory vector can be used to accurately judge the abnormal trajectory of the access trajectory. When the access trajectory is an abnormal trajectory, it is reported to the management center in time, thereby realizing effective supervision of the visitor's real access trajectory.

[0116] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AI smart community intelligent visitor management system, characterized by: The server includes a server, which collects facial images of visitors at the entrance through an entrance facial image collection module, and uses a facial recognition module to perform facial recognition on the facial images of visitors at the entrance based on a database to determine whether the visitor is a registered person; If the visitor is a registered person, the entrance access control module opens the entrance door; If the visitor is an unregistered person, the access information collection module collects the visitor's registered access information and sends it to the database by the server. The access path generation module generates a corresponding recommended access path based on the access information and sends it to the database by the server. At the same time, the server sends the recommended access path to the visitor terminal through the wireless communication module, and the manager issues a visitor card to the visitor and inputs the visitor card information into the database. The database associates and stores the visitor's face recognition result, access information, recommended access path and visitor card information, and the entrance access control module opens the entrance access. The server collects visitor images at each building through a distributed face image collection module, and uses a face recognition module to perform face recognition on visitor face images at each building based on a database. The distributed access control module determines whether to open the building access control according to the face recognition result; The server reads the location information of the visitor card in real time through the distributed location information reading module, and uses the access trajectory generation module to generate a corresponding access trajectory according to the real-time location information of the visitor card. The trajectory vector generation module generates an access trajectory trajectory vector and a recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path. The abnormal trajectory judgment module performs abnormal trajectory judgment on the access trajectory according to the access trajectory trajectory vector and the recommended access path trajectory vector.

2. The AI ​​smart community intelligent visitor management system according to claim 1 is characterized by: The face recognition module performs face recognition on the visitor's face image based on the database, including: S11, extracting facial feature information from the visitor's facial image; S12, encoding the facial feature information into a fixed-length feature digital sequence capable of representing the main features of the visitor's facial image; S13, comparing the characteristic digital sequence of the visitor's facial image with the characteristic digital sequence of the facial image of the registered person pre-stored in the database to obtain a comparison result; S14. Perform facial recognition on the visitor's facial image according to the comparison result, and delete the visitor's facial image.

3. The AI ​​smart community intelligent visitor management system according to claim 2 is characterized by: In S13, the characteristic digital sequence of the visitor's facial image is compared with the characteristic digital sequence of the facial image of the registered person pre-stored in the database to obtain the comparison result, including: S131, feature digital sequence s of visitor face image a a The characteristic digital sequence S of the registered person's face image i pre-stored in the database i Compare and get the comparison result C a,i : Among them, s a,j is the characteristic digital sequence s of visitor face image a a The jth position, S i,j is the characteristic digital sequence S of the registered person's face image i i The jth position of , ⊙ is the XOR operator, m is the number of digits in the feature digital sequence, i∈{1,2,...,n}, n is the number of registered person face images pre-stored in the database; S132, repeat S131 until the characteristic digital sequence s of the visitor's face image a is a The feature digital sequence comparison is completed with all pre-stored facial images of registered persons in the database, and multiple comparison results are obtained.

4. The AI ​​smart community intelligent visitor management system according to claim 3 is characterized by: In S14, face recognition is performed on the visitor's face image according to the comparison result, including: If there is a comparison result greater than a first preset threshold value among the multiple comparison results, the visitor is determined to be a registered person, otherwise the visitor is determined to be an unregistered person.

5. The AI ​​smart community intelligent visitor management system according to claim 2 is characterized by: The distributed access control module determines whether to open the building access control according to the face recognition result, including: If the visitor is a registered person, the distributed access control module opens the building access control. Otherwise, the corresponding access information is obtained from the database based on the face recognition result of the visitor to determine whether the access information is consistent with the building visited; If the access information is consistent with the visited building, the distributed access control module opens the building access and records the visitor's arrival and departure time. Otherwise, the server sends the visitor's face recognition result and visited building information to the management center through the wireless communication module.

6. The AI ​​smart community intelligent visitor management system according to claim 5 is characterized by: The system also includes a building access security judgment module, which calculates the access security index of each building based on the visitor access situation of a single building and the visitor access situation of all buildings, and judges the access security situation of each building based on the access security index.

7. The AI ​​smart community intelligent visitor management system according to claim 6 is characterized by: The building access security judgment module calculates the access security index of each building according to the visitor access situation of a single building and the visitor access situation of all buildings, and judges the access security situation of each building according to the access security index, including: S21. Calculate the access security index S of building b based on the visitor access situation of a single building and the visitor access situation of all buildings. b : Among them, T b,d is the d-th visit residence time data of building b, T b,d =t' b,d -t b,d , t' b,d is the visitor's departure time, t b,d is the visitor arrival time, P b is the number of visitors to building b, d∈{1,2,...,D b }, D b is the amount of visit and residence time data of building b, b∈{1,2,...,B}, B is the number of buildings; S22, repeat S21 until the access security indexes of all buildings are calculated; S23. If the access security index of the building is less than the second preset threshold, it is determined that the building is currently in an access security state; otherwise, it is determined that the building is currently in a frequent access state, and the server sends the building information to the management center through the wireless communication module.

8. The AI ​​smart community intelligent visitor management system according to claim 1 is characterized by: The trajectory vector generation module generates an access trajectory vector and a recommended access path trajectory vector according to the access trajectory and its corresponding recommended access path, including: S31, collecting the original data of the access trajectory and the corresponding recommended access path, performing data cleaning on the original data, and performing segmentation processing on the access trajectory and the corresponding recommended access path; S32, extracting the spatial features, temporal features and correlation features of each sub-trajectory in the access trajectory to obtain trajectory feature information of each sub-trajectory; S33, inputting the trajectory feature information of each sub-trajectory into the trained trajectory vector generation model to obtain a numerical vector of a fixed length corresponding to each sub-trajectory, i.e., a trajectory vector; S34, performing dimension reduction and normalization processing on the trajectory vectors corresponding to each sub-trajectory, and combining the processed trajectory vectors corresponding to each sub-trajectory to form an access trajectory vector; S35, repeat S32 to S34 to obtain a recommended access path trajectory vector; The original data includes position coordinates and timestamps, the spatial features include the starting position coordinates, end position coordinates, length and direction of the sub-trajectory, the temporal features include the starting timestamp, end timestamp and time interval of the sub-trajectory, and the associated features include the speed change, acceleration and turning angle of the sub-trajectory.

9. The AI ​​smart community intelligent visitor management system according to claim 8, characterized in that: Before inputting the trajectory feature information of each sub-trajectory into the trained trajectory vector generation model in S33, the following steps are included: A trajectory vector generation model is constructed based on the recurrent neural network (RNN), and the trajectory vector generation model is trained: S331, constructing a data set, and dividing the data set into a training set, a validation set, and a test set according to a preset ratio; S332, setting the loss function and optimizer of the trajectory vector generation model, and inputting the training set into the trajectory vector generation model for model training; S333, calculating the loss value based on the loss function, and the optimizer updates the model parameters according to the loss value and the network gradient information; S334, if the loss value is less than the third preset threshold, the model training ends, and the current trajectory vector generation model is the trained trajectory vector generation model, otherwise, the process returns to S332, and the model training continues using the training set; S335, inputting the validation set into the trained trajectory vector generation model, evaluating the generalization ability of the model by observing the performance on the validation set, and tuning the hyperparameters and structure of the model; S336. Input the test set into the optimized trajectory vector generation model to evaluate the model performance.

10. The AI ​​smart community intelligent visitor management system according to claim 8, characterized in that: The abnormal trajectory judgment module performs abnormal trajectory judgment on the access trajectory according to the access trajectory trajectory vector and the recommended access path trajectory vector, including: S41, calculating the similarity between the trajectory vectors of the same sub-trajectory in the access trajectory trajectory vector and the recommended access path trajectory vector, and screening out the target sub-trajectory whose similarity is less than a fourth preset threshold; S42, matching the access trajectory and the target sub-trajectory in the corresponding recommended access path, and counting the distances of unmatched sections in the access trajectory; S43, calculating the abnormality judgment index A of the access trajectory according to the unmatched road section distances and their corresponding travel times in the access trajectory: Among them, l x is the distance of the unmatched road section in the x-th target sub-trajectory, L x is the length of the x-th target sub-trajectory, t x is the moving time corresponding to the mismatched road section distance in the x-th target sub-trajectory, T x is the time interval of the x-th target sub-trajectory, T x =T x2 -T x1 , T x2 is the endpoint timestamp of the x-th target sub-trajectory, T x1 is the starting timestamp of the x-th target sub-trajectory, and N is the number of target sub-trajectories; S44: If the abnormality judgment index of the access trajectory is less than the fifth preset threshold, the access trajectory is judged to be an abnormal trajectory, and the server sends the access trajectory and its corresponding face recognition result and access information to the management center through the wireless communication module.

Citation Information

Patent Citations

  • Face image biological key generating method

    CN103886235A

  • Appointment access method and system, computer equipment and storage medium

    CN109360314A

  • Intelligent passage management method and system for smart park, terminal and storage medium

    CN113284282A

  • Control method and device of face recognition system, access control equipment and storage medium

    CN117711102A

  • Intelligent building safety control method and system, terminal and storage medium

    CN118644922A

Cited By

  • Real-time visitor tracking method and control system based on artificial intelligence

    CN119721471A

  • An artificial intelligence-based visitor real-time tracking method and control system

    CN119721471B

  • Digital intelligent building maintenance method and system

    CN121073711A