Campus safety informatization management method based on artificial intelligence

Abstract

The invention discloses a campus security informatization management method based on artificial intelligence, and relates to the technical field of campus security informatization management, and the method comprises the steps: collecting the multi-modal identity features of passing personnel in real time, combining with a pre-stored authorized personnel database, employing a deep learning-based convolutional neural network CNN and a spatial-temporal feature extraction model, and obtaining the multi-modal identity features of the passing personnel; the method comprises the following steps: performing high-precision comparison on human face and gait features, firstly matching human face features, triggering gait feature recognition after failure, and generating suspicious person marks or check notifications in a classified manner according to the matching failure condition in combination with a dynamic comparison decision-making mechanism, so that the check process is efficient and accurate; a people flow density monitoring model and a congestion prediction model are integrated, a long short-term memory (LSTM) network is utilized to carry out trend prediction on data acquired in real time, a density growth rate and congestion state marks are calculated, peak congestion risks are identified in advance, and a passing mode is automatically adjusted; and an additional channel is started or a rapid verification mode is switched, so that the passing efficiency in peak periods is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of campus safety information management, and in particular to an artificial intelligence-based campus safety information management method. Background Art

[0002] With the continuous expansion of campus scale and the increasing frequency of personnel flow, information management has gradually become an important support means for campus security management. Schools generally deploy surveillance cameras, access control systems and visitor registration systems to provide basic guarantees for campus safety.

[0003] During daily peak hours, such as when students go to and from school or when parents visit, school entrances and exits are often faced with heavy traffic. Although traditional access cards or manual verification methods can complete basic identity verification, and visitor registration systems can also record information on outsiders, such processes are cumbersome and inefficient, and can easily lead to queues and congestion at entrances and exits. In a chaotic environment, unauthorized personnel can easily take the opportunity to sneak into the campus, posing a potential safety hazard.

[0004] In addition, the role of traditional monitoring equipment is more limited to post-event tracing, and its ability to prevent real-time risks is limited. Faced with scenes with rapid personnel flow, relying solely on manual review of monitoring footage or verification of visitor information is not only difficult to detect problems in a timely manner, but is also prone to management loopholes due to human negligence, affecting the daily travel efficiency of teachers and students. Therefore, there is an urgent need for an artificial intelligence-based campus security information management method to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides an artificial intelligence-based campus security information management method to solve the problem that a large number of students and visitors pass through the school gate at the same time during peak hours, morning and evening school entry periods, or school opening and holidays, and traditional access card or manual verification methods are inefficient and easily lead to queue congestion.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The embodiment of the present invention provides a campus safety information management method based on artificial intelligence, which includes:

[0009] Step S1, deploying smart cameras that support face recognition and gait recognition at the school gate to collect identity features of people passing by in real time;

[0010] Step S2, building a database for pre-storing authorized personnel data, and using an artificial intelligence algorithm to compare the identity features collected in step S1 with the data in the database to confirm the identity of the person;

[0011] Step S3, the intelligent camera additionally monitors the pedestrian flow density data in the school gate area, and uses the congestion prediction model to predict the collected pedestrian flow density data. When it is predicted that a peak congestion is about to occur or it is detected that congestion has actually occurred, adjustment measures are automatically executed. The adjustment measures include enabling additional access channels and switching to the fast verification mode.

[0012] As a preferred solution of the campus security informatization management method based on artificial intelligence according to the present invention, wherein: the identity feature information includes a face image and a gait feature.

[0013] As a preferred solution of the campus security informatization management method based on artificial intelligence according to the present invention, wherein: the step of constructing the database of pre-stored authorized personnel data is as follows.

[0014] Define the dataset of authorized personnel as D:

[0015]

[0016] wherein, I i is the original face image of the i-th authorized person, G i is the original gait feature data of the i-th authorized person, L i is the identity label of the i-th authorized person, and n is the total number of authorized personnel in the dataset;

[0017] Extract features from the face image I i to generate a feature vector expressed as:

[0018]

[0019] wherein, f face is a face feature extraction model, adopting a CNN model based on a convolutional neural network, is the embedded face feature vector of the i-th authorized person;

[0020] Extract features from the gait feature G i to generate a feature vector expressed as:

[0021]

[0022] wherein,

[0023] g gait is a gait feature extraction model, implemented based on a spatio-temporal feature extraction method, is the embedded gait feature vector of the i-th authorized person;

[0024] Store the extracted features in the database to construct a feature set, expressed as:

[0025]

[0026] Among them, DB is the finally constructed database, are the feature vectors of face and gait respectively, and L i is the identity label.

[0027] As a preferred solution of the method for campus security informatization management based on artificial intelligence according to the present invention, wherein: the comparison process in step S2 includes face feature comparison and gait feature comparison. The face feature comparison is performed by using a convolutional neural network CNN framework based on deep learning combined with an angular loss function, and the gait feature comparison is performed by using a deep learning model for spatio-temporal feature extraction.

[0028] As a preferred solution of the method for campus security informatization management based on artificial intelligence according to the present invention, wherein: during the comparison process in step S2,

[0029] first match the collected face features. If the face features match successfully, further perform gait feature recognition. When the gait feature recognition is successful, automatically release;

[0030] If the face feature matching fails, also perform gait feature recognition. If the gait feature recognition fails, mark the person as a suspicious person and notify the doorman for verification. If the gait feature recognition is successful, only notify the doorman for verification without marking the suspicious status.

[0031] As a preferred solution of the method for campus security informatization management based on artificial intelligence according to the present invention, wherein: the authorized personnel data includes the face images and gait features of students and faculty members;

[0032] In the database, add temporary reservation data as needed, including the face images of temporary personnel.

[0033] As a preferred solution of the method for campus security informatization management based on artificial intelligence according to the present invention, wherein: the steps of using a convolutional neural network CNN framework based on deep learning combined with an angular loss function for face feature comparison and using a deep learning model for spatio-temporal feature extraction for gait feature comparison are as follows,

[0034] Define the input feature and the reference feature respectively as:

[0035] X f = f face (I), X g = g gait (G),

[0036] Among them, X fX is the face feature vector of the person collected in real time. g I is the gait feature vector of the person collected in real time. G is the face image and gait feature data collected in real time. f face and g gait are the face and gait feature extraction models respectively.

[0037] Based on the convolutional neural network CNN, face feature comparison is carried out, and the angular loss function is combined to optimize the similarity. The formula of the angular loss function is:

[0038]

[0039] Among them, L face is the angular loss function for face feature comparison. θ j is the included angle between the j-th collected feature and the feature vector in the database. θ k is the included angle between the k-th collected feature in the database and the feature vector in the database. m is the number of face features collected in real time, and n is the number of pre-stored face features in the database.

[0040] For spatio-temporal feature comparison of gait features, the similarity calculation formula is:

[0041]

[0042] Among them, S gait is the gait feature similarity score. α t is the gait time series weighting coefficient. is the time series angle of the gait feature vector. T is the number of time series frames of the gait data.

[0043] Calculate the comprehensive comparison score. The calculation formula is:

[0044] S total =λ f ·S face +λ g ·S gait ,

[0045] Among them, S total is the comprehensive similarity score. S face is the face feature similarity score.

[0046] λ f and λ g are weight parameters.

[0047] Make a decision on the comparison result:

[0048] If S total ≥τ, then Decision = Accept.

[0049] If S total< τ, then Decision = Reject,

[0050] where Decision is the comparison result, Accept indicates an authorized identity, Reject indicates an unauthorized identity and further determination is required, τ is the similarity threshold, which is set as a predefined constant.

[0051] As a preferred solution of the campus security informatization management method based on artificial intelligence according to the present invention, wherein: the comparison process in step S2 includes:

[0052] Input the real-time collected features and the features pre-stored in the database for comparison, and calculate the similarity. The calculation formula is:

[0053]

[0054] where is the face feature X collected in real time f and the face feature of the i-th person in the database similarity, is the gait feature X collected in real time g and the gait feature of the i-th person in the database similarity, cos is the cosine similarity for calculating between two feature vectors;

[0055] Determine whether the face feature comparison is successful. The determination formula is:

[0056] If then FaceMatch i = 1,

[0057] If then FaceMatch i = 0,

[0058] where FaceMatch i is the face comparison result flag, 1 is success, 0 is failure, τf is the face feature comparison similarity threshold;

[0059] When the face feature comparison fails, trigger the gait feature comparison. The determination formula is:

[0060] If then GaitMatch i = 1,

[0061] If then GaitMatch i = 0,

[0062] where GaitMatch i is the gait comparison result flag, 1 is success, 0 is failure, τg is the similarity threshold for gait feature comparison;

[0063] Generate an audit status flag based on the comparison result, which is expressed as:

[0064] If FaceMatch i = 1 and GaitMatch i = 1, then AuditState = Pass,

[0065] If FaceMatch i = 0 and GaitMatch i = 1, then AuditState = Warning,

[0066] If FaceMatch i = 0 and GaitMatch i = 0, then AuditState = Alert,

[0067] Among them, AuditState is the audit status, Pass means successful comparison and personnel confirmation, Warning means successful gait match but failed face match, and Alert means both face and gait do not match successfully;

[0068] Record the marking information, including:

[0069] M = {(X f , X g , AuditState)},

[0070] Among them, M is the set of marking information, and AuditState is the audit status corresponding to the current comparison.

[0071] As a preferred solution of the method for campus security informatization management based on artificial intelligence described in the present invention, wherein: the comparison process of the quick verification mode is as follows: after successful face feature matching, directly release without performing gait feature recognition; if face feature matching fails, then perform gait feature recognition, and automatically mark and notify the doorman to record the identity of the personnel with failed gait feature recognition.

[0072] As a preferred solution of the method for campus security informatization management based on artificial intelligence described in the present invention, wherein: the step of automatically executing adjustment measures when predicting upcoming peak congestion or detecting actual congestion by using the congestion prediction model for predicting the collected crowd density data is,

[0073] Define the sequence of real-time collected crowd density data as D t , which is expressed as:

[0074] Dt = {d t-k , d t-k+1 , …, d t},

[0075] where D t is the sequence of pedestrian flow density data collected at the k + 1 moments before time t, and d t-k , d t-k+1 , …, d t are the local pedestrian flow density data collected between time t - k and t, and k is the time series length of the prediction window;

[0076] Input the pedestrian flow density data into the time series prediction model to generate the future density prediction value, which is expressed as:

[0077]

[0078] where is the predicted pedestrian flow density value at time t + 1, and f lstm is the density prediction model based on the long short-term memory network LSTM;

[0079] Calculate the density growth rate and compare it with the threshold. The formula for calculating the density growth rate is:

[0080]

[0081] where g t+1 is the density growth rate at time t + 1, τ g is the density growth rate threshold, and Δ is the state indicating that the growth rate exceeds the threshold;

[0082] Judge the congestion state. The judgment formula is:

[0083] If g t+1 < τ g , then CongestionState = Normal,

[0084] If g t+1 ≥ τ g , then CongestionState = Warning,

[0085] where CongestionState is the congestion state flag, Normal means the current predicted density is within the normal range, and Warning means the current predicted density exceeds the threshold range and congestion may occur;

[0086] Adjust the traffic mode according to the congestion state. The adjustment method is:

[0087] If CongestionState = Normal, then Action = None,

[0088] If CongestionState = Warning, then Action = OpenExtraChannels,

[0089] where Action is the response action, and OpenExtraChannels is to enable additional access channels,

[0090] Adjust the feature comparison process in the fast verification mode, and the adjustment method is as follows:

[0091] If CongestionState = Warning, then Verification = FaceOnly,

[0092] If CongestionState = Normal, then Verification = FaceAndGait,

[0093] where Verification is the feature comparison mode, FaceOnly is to perform only face feature comparison, and FaceAndGait is to perform comprehensive comparison of face and gait features.

[0094] The beneficial effects of the present invention are as follows: The present invention collects the multi-modal identity features of passing personnel in real time, combines with the pre-stored authorized personnel database, and uses the convolutional neural network CNN and spatio-temporal feature extraction model based on deep learning to perform high-precision comparison of face and gait features; in the verification process, first match the face features, and then trigger the gait feature recognition after failure, and combine with the dynamic comparison decision mechanism to generate suspicious person marks or verification notices for the cases of matching failure, and the verification process is efficient and accurate;

[0095] The present invention integrates a crowd density monitoring and congestion prediction model, uses a long short-term memory network LSTM to predict the trend of real-time collected data, calculates the density growth rate and congestion status mark, identifies the peak congestion risk in advance, and automatically adjusts the access mode; enables additional channels or switches to the fast verification mode to ensure the access efficiency and safety during peak hours; the fast verification mode simplifies the process, directly releases after successful face feature matching, avoids unnecessary gait comparison, and retains basic monitoring for unmatched personnel, further balancing efficiency and safety.

[0096] The present invention uses an angular loss function and a gait spatio-temporal feature weighted similarity calculation method in the feature comparison process, which enhances the robustness and generality of the comparison, and effectively reduces the false alarm rate and missed alarm rate in complex scenarios. Description of the Drawings

[0097] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0098] Figure 1 It is a schematic flow chart of the campus security informatization management method based on artificial intelligence of the present invention. Specific embodiments

[0099] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings of the specification.

[0100] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0101] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an individual or alternative embodiment that is mutually exclusive with other embodiments.

[0102] Embodiment 1, refer to Figure 1 , this embodiment provides a campus security informatization management method based on artificial intelligence, including:

[0103] Step S1, deploy intelligent cameras supporting face recognition and gait recognition at the school gate to collect the identity characteristics of passing personnel in real time;

[0104] The identity characteristic information includes face images and gait characteristics;

[0105] Step S2, construct a database storing authorized personnel data, and compare the identity characteristics collected in step S1 with the data in the database through artificial intelligence algorithms to confirm the identity of the personnel;

[0106] The step of constructing the database storing authorized personnel data is

[0107] Define the dataset of authorized personnel as D:

[0108]

[0109] Wherein, I iis the original face image of the i-th authorized person, G i is the original gait feature data of the i-th authorized person, L i is the identity label of the i-th authorized person, and n is the total number of authorized persons in the dataset;

[0110] Extract features from the face image I i to generate a feature vector which is expressed as:

[0111]

[0112] where f face is a face feature extraction model, using a CNN model based on a convolutional neural network, is the embedded face feature vector of the i-th authorized person;

[0113] Extract features from the gait feature G i to generate a feature vector which is expressed as:

[0114]

[0115] where

[0116] g gait is a gait feature extraction model, implemented based on a spatio-temporal feature extraction method, is the embedded gait feature vector of the i-th authorized person;

[0117] Store the extracted features in a database to construct a feature set, which is expressed as:

[0118]

[0119] where DB is the finally constructed database, are the feature vectors of face and gait respectively, L i is the identity label;

[0120] Specifically, this step defines the original dataset, extracts face and gait features, stores the feature vectors in the database for subsequent comparison; by using a convolutional neural network and a spatio-temporal feature extraction model, the data storage process and accuracy are improved, and the feature extraction deeply encodes the input data of different dimensions, strengthening the generality of the database.

[0121] The comparison process in step S2 includes face feature comparison and gait feature comparison. The convolutional neural network CNN framework based on deep learning combined with an angular loss function is used for face feature comparison, and a deep learning model based on spatio-temporal feature extraction is used for gait feature comparison;

[0122] During the comparison process of step S2,

[0123] First, match the collected face features. If the face feature matching is successful, further perform gait feature recognition. When the gait feature recognition is successful, automatically release.

[0124] If the face feature matching fails, also perform gait feature recognition. If the gait feature recognition fails, mark the person as a suspicious person and notify the security guard for verification. If the gait feature recognition is successful, only notify the security guard for verification without marking the suspicious status.

[0125] The authorized personnel data includes the face images and gait features of students and faculty members.

[0126] In the database, add temporary reservation data as needed, including the face images of temporary personnel.

[0127] The steps of using a convolutional neural network CNN framework based on deep learning combined with an angular loss function for face feature comparison and using a deep learning model for spatio-temporal feature extraction for gait feature comparison are as follows:

[0128] Define the input feature and the reference feature as:

[0129] X f = f face (I), X g = g gait (G),

[0130] where X f is the face feature vector of the person collected in real time, X g is the gait feature vector of the person collected in real time, I and G are the face image and gait feature data collected in real time, f face and g gait are the face and gait feature extraction models respectively.

[0131] Based on the convolutional neural network CNN, perform face feature comparison and optimize the similarity by combining the angular loss function. The formula of the angular loss function is:

[0132]

[0133] where L face is the angular loss function for face feature comparison, θ j is the included angle between the j-th collected feature and the feature vector in the database, θ k is the included angle between the k-th collected feature in the database and the feature vector in the database, m is the number of face features collected in real time, and n is the number of pre-stored face features in the database.

[0134] Perform spatio-temporal feature comparison on the gait features. The similarity calculation formula is:

[0135]

[0136] Among them, S gait is the gait feature similarity score, and α t is the gait time series weighting coefficient, is the time series angle of the gait feature vector, T is the number of time series frames of the gait data,

[0137] Calculate the comprehensive comparison score, and the calculation formula is:

[0138] S total = λ f ·S face + λ g ·S gait ,

[0139] Among them, S total is the comprehensive similarity score, S face is the face feature similarity score,

[0140] λ f and λ g are weight parameters,

[0141] Make a judgment on the comparison result:

[0142] If S total ≥ τ, then Decision = Accept,

[0143] If S total < τ, then Decision = Reject,

[0144] Among them, Decision is the comparison result, Accept indicates an authorized identity, Reject indicates an unauthorized identity and needs further judgment, and T is the similarity threshold, which is set as a predefined constant;

[0145] Specifically, in this step, multi-level feature comparison is performed, and an angular loss function and spatio-temporal weighted similarity calculation are used to improve the robustness of the comparison; in addition, the comprehensive similarity calculation combines weight adjustment to dynamically balance the face and gait features.

[0146] The comparison process in step S2 includes:

[0147] Input the features collected in real time and the features pre-stored in the database for comparison, and calculate the similarity. The calculation formula is:

[0148]

[0149] Among them, is the face feature X collected in real time fThe similarity with the facial features of the i-th person in the database is the similarity between the gait feature X collected in real time g and the gait features of the i-th person in the database , where cos is the cosine similarity calculated between two feature vectors;

[0150] To determine whether the facial feature comparison is successful, the determination formula is:

[0151] If then FaceMatch i = 1,

[0152] If then FaceMatch i = 0,

[0153] where FaceMatch i is the facial comparison result flag, 1 means success, 0 means failure, and τ f is the facial feature comparison similarity threshold;

[0154] When the facial feature comparison fails, trigger the gait feature comparison. The determination formula is:

[0155] If then GaitMatch i = 1,

[0156] If then GaitMatch i = 0,

[0157] where GaitMatch i is the gait comparison result flag, 1 means success, 0 means failure, and τ g is the gait feature comparison similarity threshold;

[0158] Generate an audit status flag based on the comparison results, which is expressed as:

[0159] If FaceMatch i = 1 and GaitMatch i = 1, then AuditState = Pass,

[0160] If FaceMatch i = 0 and GaitMatch i = 1, then AuditState = Warning,

[0161] If FaceMatch i = 0 and GaitMatch iIf it is equal to 0, then AuditState = Alert.

[0162] Among them, AuditState is the audit status, Pass means successful comparison and personnel confirmation, Warning means successful gait matching but failed face matching, and Alert means both face and gait matching failed.

[0163] Record the marking information, including:

[0164] M = {(X f , X g , AuditState)},

[0165] Among them, M is the set of marking information, and AuditState is the audit status corresponding to the current comparison.

[0166] Specifically, through a multi-level comparison decision tree, the identity accuracy is improved. For the case of failed face feature comparison, gait comparison is started in a timely manner to minimize the false alarm rate to the greatest extent; at the same time, based on the hierarchical management mechanism of the audit status, normal, suspicious, and high-risk personnel are clearly distinguished.

[0167] Step S3, the intelligent camera additionally monitors the pedestrian flow density data in the school gate area, uses the congestion prediction model to predict the collected pedestrian flow density data, and when it is predicted that peak congestion is about to occur or it is detected that congestion has actually occurred, adjustment measures are automatically executed. The adjustment measures include enabling additional access channels and switching to the fast verification mode;

[0168] The comparison process of the fast verification mode is as follows: directly release after successful face feature matching without performing gait feature recognition; if face feature matching fails, perform gait feature recognition, and automatically mark and notify the doorman to record the identity of the personnel with failed gait feature recognition;

[0169] The step of using the congestion prediction model to predict the collected pedestrian flow density data and automatically executing adjustment measures when it is predicted that peak congestion is about to occur or it is detected that congestion has actually occurred is,

[0170] Define the real-time collected pedestrian flow density data sequence as D t , expressed as:

[0171] D t = {d t-k , d t-k+1 , …, d t},

[0172] Among them, D t is the pedestrian flow density data sequence collected at the previous k + 1 moments of time t, d t-k , d t-k+1 ,...,, dt is the local pedestrian flow density data collected between time t - k and t, where k is the time series length of the prediction window;

[0173] Input the pedestrian flow density data into the time series prediction model to generate the future density prediction value, expressed as:

[0174]

[0175] where is the predicted pedestrian flow density value at time t + 1, and f lstm is the density prediction model based on the long short - term memory network (LSTM);

[0176] Calculate the density growth rate and compare it with the threshold. The formula for the density growth rate is:

[0177]

[0178] where g t+1 is the density growth rate at time t + 1, τ g is the density growth rate threshold, and Δ is the state flag indicating that the growth rate exceeds the threshold;

[0179] Judge the congestion state. The judgment formula is:

[0180] If g t+1 < τ g , then CongestionState = Normal,

[0181] If g t+1 ≥ τ g , then CongestionState = Warning,

[0182] where CongestionState is the congestion state flag, Normal means the current predicted density is within the normal range, and Warning means the current predicted density exceeds the threshold range and congestion may occur;

[0183] Adjust the traffic mode according to the congestion state. The adjustment method is:

[0184] If CongestionState = Normal, then Action = None,

[0185] If CongestionState = Warning, then Action = OpenExtraChannels,

[0186] where Action is the response action, and OpenExtraChannels is to enable additional traffic channels,

[0187] Adjust the feature comparison process in the fast verification mode, and the adjustment method is as follows:

[0188] If CongestionState = Warning, then Verification = FaceOnly,

[0189] If CongestionState = Normal, then Verification = FaceAndGait,

[0190] Among them, Verification is the feature comparison mode, FaceOnly is to perform only face feature comparison, and FaceAndGait is to perform comprehensive comparison of face and gait features;

[0191] Specifically, the crowd density is predicted through a time series prediction model, and early warning of potential congestion is achieved by combining the growth rate threshold; based on the dynamic adjustment mechanism of the congestion state, the verification load during peak periods is effectively reduced while ensuring the passing efficiency and safety; and the fast verification mode provides a balance solution of real-time performance and safety, further improving the adaptability of the present invention.

[0192] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A campus safety information management method based on artificial intelligence, characterized in that: include, Step S1, deploying smart cameras that support face recognition and gait recognition at the school gate to collect identity features of people passing by in real time; Step S2, building a database for pre-storing authorized personnel data, and using an artificial intelligence algorithm to compare the identity features collected in step S1 with the data in the database to confirm the identity of the person; Step S3, the smart camera additionally monitors the crowd density data in the school gate area, and uses the congestion prediction model to predict the collected crowd density data. When peak congestion is predicted to occur or congestion is detected to have actually occurred, adjustment measures are automatically executed, including enabling additional passages and switching to quick verification mode.

2. A campus safety information management method based on artificial intelligence as claimed in claim 1, characterized in that: The identity feature information includes a face image and gait features.

3. A campus safety information management method based on artificial intelligence as claimed in claim 2, characterized in that: The step of constructing a database for pre-storing authorized personnel data is: Define the dataset of authorized personnel as D: Among them, I i is the original face image of the i-th authorized person, G i is the original gait feature data of the i-th authorized person, L i is the identity label of the i-th authorized person, and n is the total number of authorized persons in the dataset; For face image I i Perform feature extraction and generate feature vectors It is expressed as: Among them, f face As a face feature extraction model, a CNN model based on convolutional neural network is used. is the embedded facial feature vector of the i-th authorized person; For gait characteristics G i Perform feature extraction and generate feature vectors It is expressed as: in, g gait It is a gait feature extraction model, which is implemented based on the spatiotemporal feature extraction method. is the embedded gait feature vector of the i-th authorized person; The extracted features are stored in the database and a feature set is constructed, which is expressed as: Among them, DB is the final constructed database, are the feature vectors of face and gait respectively, L i For identity tags.

4. A campus safety information management method based on artificial intelligence as claimed in claim 3, characterized in that: The comparison process of step S2 includes facial feature comparison and gait feature comparison. The facial feature comparison is performed using a convolutional neural network (CNN) framework based on deep learning combined with an angle loss function, and the gait feature comparison is performed using a deep learning model based on spatiotemporal feature extraction.

5. A campus safety information management method based on artificial intelligence as claimed in claim 4, characterized in that: During the comparison process of step S2, First, the collected facial features are matched. If the facial features are matched successfully, the gait features are further recognized. When the gait features are recognized successfully, the vehicle is automatically released. If the facial feature matching fails, gait feature recognition is also performed. If gait feature recognition fails, the person is marked as a suspicious person and the security guard is notified for verification. If gait feature recognition is successful, only the security guard is notified for verification and the suspicious status is not marked.

6. A campus safety information management method based on artificial intelligence as claimed in claim 5, characterized in that: The authorized personnel data includes facial images and gait characteristics of students and faculty members; In the database, temporary reservation data is added as needed, including facial images of temporary personnel.

7. The campus safety information management method based on artificial intelligence as claimed in claim 6, characterized in that: The steps of using a convolutional neural network (CNN) framework based on deep learning combined with an angle loss function to compare facial features and using a deep learning model based on spatiotemporal feature extraction to compare gait features are as follows: The input features and reference features are defined as follows: X f =f face (I),X g =g gait (G), Among them, X f is the facial feature vector collected in real time, X g is the gait feature vector collected in real time, I and G are the face image and gait feature data collected in real time, and f face , g gait They are face and gait feature extraction models respectively. Based on the convolutional neural network CNN, facial feature comparison is performed and the similarity is optimized by combining the angle loss function. The angle loss function formula is: Among them, L face is the angle loss function for facial feature comparison, θ j is the angle between the jth collected feature and the feature vector in the database, θ k is the angle between the kth collected feature in the database and the feature vector in the database, m is the number of face features collected in real time, and n is the number of face features pre-stored in the database; The temporal and spatial features of the gait features are compared, and the similarity calculation formula is: Among them, S gait is the gait feature similarity score, α t is the gait timing weighting coefficient, is the temporal angle of the gait feature vector, T is the temporal frame number of the gait data, Calculate the comprehensive comparison score using the following formula: S total =λ f ·S face +λ g ·S gait , Among them, S total is the comprehensive similarity score, S face is the facial feature similarity score, λ f , g is the weight parameter, Compare the results and determine: If S total ≥τ, then Decision=Accept, If S total <τ, then Decision=Reject, Among them, Decision is the comparison result, Accept indicates authorized identity, Reject indicates unauthorized identity and needs further judgment, and τ is the similarity threshold, which is set to a predefined constant.

8. The campus safety information management method based on artificial intelligence as claimed in claim 7, characterized in that: The comparison process of step S2 includes: Input the real-time collected features and compare them with the pre-stored features in the database to calculate the similarity. The calculation formula is: in, The face feature X collected in real time f and the facial features of the i-th person in the database The similarity of is the gait feature X collected in real time g and the gait characteristics of the i-th person in the database Similarity, cos is the cosine similarity between two feature vectors; To determine whether the facial feature comparison is successful, the judgment formula is: like FaceMatch i =1, like FaceMatch i =0, Among them, FaceMatch i is the face comparison result flag, 1 for success, 0 for failure, τ f is the similarity threshold for facial feature comparison; When the facial feature comparison fails, the gait feature comparison is triggered, and the judgment formula is: like GaitMatch i =1, like GaitMatch i =0, Among them, GaitMatch i is the gait comparison result flag, 1 for success, 0 for failure, τ g is the similarity threshold for gait feature comparison; The audit status flag is generated based on the comparison results, expressed as: If FaceMatch i =1 and GaitMatch i =1, then AuditState = Pass, If FaceMatch i =0 and GaitMatch i =1, then AuditState=Warning, If FaceMatch i =0 and GaitMatch i = 0, then AuditState = Alert, Among them, AuditState is the audit status, Pass is the matching success and personnel confirmation, Warning is the gait matching success but face matching failure, and Alert is the face and gait matching failure; Record tag information, including: M={(X f ,X g ,AuditState)}, Among them, M is a set of marking information, and AuditState is the audit status corresponding to the current comparison.

9. The campus safety information management method based on artificial intelligence as claimed in claim 8, characterized in that: The comparison process of the quick verification mode is as follows: if the facial feature match is successful, the person is released directly without the need for gait feature recognition; if the facial feature match fails, gait feature recognition is performed, and the person for whom gait feature recognition fails is automatically marked and the guard is notified to record his or her identity.

10. The campus safety information management method based on artificial intelligence as claimed in claim 9, characterized in that: The step of using the congestion prediction model to predict the collected crowd density data and automatically executing adjustment measures when it is predicted that peak congestion is about to occur or congestion is detected to have actually occurred is: Define the real-time collected crowd density data sequence as D t , expressed as: D t ={d t-k ,d t-k+1 ,…,d t }, Among them, D t is the crowd density data sequence collected k+1 moments before time t, d t-k ,d t-k+1 ,…,d t is the local crowd density data collected between time tk and t, and k is the time series length of the prediction window; Input the crowd density data into the time series forecasting model to generate the future density forecast value, which is expressed as: in, is the predicted crowd density value at time t+1, f lstm It is a density prediction model based on long short-term memory network LSTM; Calculate the density growth rate and compare it with the threshold. The density growth rate calculation formula is: Among them, g t+1 is the density growth rate at time t+1, τ g is the density growth rate threshold, Δ is the state where the marking growth rate exceeds the threshold; The congestion state is determined by the following formula: If g t+1 <τ g , then CongestionState=Normal, If g t+1 ≥τ g , then CongestionState=Warning, Among them, CongestionState is the congestion state mark, Normal means the current predicted density is within the normal range, and Warning means the current predicted density exceeds the threshold range and congestion may occur; Adjust the traffic mode according to the congestion status. The adjustment method is as follows: If CongestionState = Normal, then Action = None, If CongestionState = Warning, then Action = OpenExtraChannels, Among them, Action is the response action, OpenExtraChannels is to enable additional channels, Adjust the feature comparison process in the quick verification mode by: If CongestionState = Warning, then Verification = FaceOnly, If CongestionState = Normal, then Verification = FaceAndGait, Among them, Verification is the feature comparison mode, FaceOnly is for comparing only facial features, and FaceAndGait is for a comprehensive comparison of facial and gait features.

Images