A smart community traffic record matching and recognition system and method based on image label enhancement

Abstract

The present invention is applicable to the field of community smart management, and provides a smart community access record matching and identification system and method based on image tag enhancement, including the following steps: first shooting an image, transmitting it to a system terminal to perform face recognition and confirmation on individual personnel; confirming the label code and label description through feature item information and corresponding to the individual user, giving the label code access rights within a time limit; shooting an image a second time, transmitting it to a system terminal, regenerating the label code and label description, querying and matching the regenerated label code among users with time limit rights, and performing actions on the gate and prompt system according to the matching situation. In this way, the present invention can improve the efficiency of social security management and enhance the monitoring and early warning capabilities of strangers entering and leaving.

Description

Technical Field

[0001] The present invention is applicable to the field of community smart management, and provides a smart community traffic record matching and recognition system and method based on image tag enhancement. Background Art

[0002] Smart community is a new community model that uses information technology, Internet of Things, artificial intelligence and other advanced technologies to comprehensively manage and serve the community. It aims to improve the management efficiency, convenience and quality of life of the community through intelligent means. Smart communities usually include applications in smart security, smart home, smart medical care, smart transportation, environmental monitoring and other aspects. In smart communities, accurately recording and identifying people's passage records is of great significance for social governance, public safety, crowd analysis and other aspects.

[0003] Existing technologies mainly rely on facial feature recognition systems to match captured photos with registered face libraries. However, due to factors such as camera angle, lighting conditions and some unregistered persons, the recognition process takes a long time, causing people to stay at the gate for a long time. In some scenarios with large population flow and the need to accurately identify people entering and leaving, congestion and long queues often occur at the entrance and exit gates, which greatly affects the user experience and the control of passenger flow. Summary of the invention

[0004] In view of the above-mentioned defects, the purpose of the present invention is to provide a smart community traffic record matching and recognition system and method based on image tag enhancement, the purpose of which is to solve the problems raised in the background technology. The system includes a terminal recording system, a monitoring and recognition system, a prompt system and a gate system;

[0005] The terminal recording system includes a terminal computer, the terminal computer includes a storage device, and the storage device stores an operation database, a feature item database and a data storage library; the operation database stores the necessary programs for the computer operation; the feature item database contains feature items indexed by the tag name; the data storage library stores the facial features of all registered users and all the entry and exit data of all registered users and unregistered users within a period of time;

[0006] The monitoring and recognition system includes a monitoring system and a recognition system; the monitoring system includes a number of high-definition monitoring cameras distributed at the building entrances and exits and gate channels; the recognition system includes a face recognition system and a feature recognition system; the face recognition system and the feature recognition system respectively recognize, extract, and compare individual faces and feature items, and generate corresponding label descriptions and label codes;

[0007] The prompt system is a voice prompt device or display device installed near the gate system; the gate system is a gate device installed at the entrance and exit of the community.

[0008] Furthermore, the monitoring camera equipment is distributed at the entrances and exits of each unit in the community and at the entrance and exit gates of the community.

[0009] Furthermore, the face recognition system and feature recognition system use models trained by Haar cascade detector or Transformer and CLIP technology; the feature recognition system includes height estimation, and the height estimation uses an algorithm based on 3D posture estimation.

[0010] Furthermore, the types of feature items mainly include clothing, hairstyle, body shape, carried items and accessories, which correspond to the clothing database, hairstyle database, body shape database, carried items database and accessories database stored in the feature item database respectively.

[0011] A method for matching and identifying traffic records in a smart community based on image label enhancement is based on the above-mentioned smart community traffic record matching and identifying system based on image label enhancement, and includes the following steps:

[0012] S1.1 The surveillance equipment on the road sections in the direction of the community’s internal exit and the road sections in the direction of the community’s external entrance takes high-definition images for the first time, captures images of individuals appearing in the images, and transmits the videos and images to the storage device and recognition system in the system terminal;

[0013] S1.2 The face recognition system and feature recognition system detect, identify, and match face information and feature item information respectively, confirm the user individual through face information, and confirm the label code and label description through feature item information;

[0014] S1.3 The tag code and tag description are mapped to the user individual, and the system terminal grants the tag code access rights within a certain period of time;

[0015] S1.4 The monitoring equipment of the entrance and exit gates takes a high-definition image for the second time, captures the images of the individuals appearing in the image, and transmits the video and images to the storage device and recognition system in the system terminal;

[0016] S1.5 The feature recognition system detects and recognizes feature item information, regenerates label codes and label descriptions, and searches and matches the regenerated label codes among users with time-limited permissions;

[0017] If S1.6 matches, the gate will be controlled to open in advance when the individual approaches the gate, and the prompt system will prompt the person to enter the designated gate.

[0018] Furthermore, in step S1.1 and step S1.4, the monitoring device records the time information while taking the high-definition images for the first and second time and transmitting them to the storage device in the system terminal.

[0019] Furthermore, the generation process of the label encoding and label description in step S1.2 includes the following steps:

[0020] S2.1 The monitoring system transmits the captured high-definition images to the feature recognition system;

[0021] S2.2 The feature recognition system uses the Haar level detector to locate and select each region, each region represents a different feature item type;

[0022] S2.3 extracts the feature value in each feature item type, and compares the feature value with the features in the feature item database to find matching feature items;

[0023] S2.4 combines the numbers of several feature items and the descriptions corresponding to each feature item number to generate a label code and a label description;

[0024] S2.5 assigns the label code and label description to the data of the current individual and stores it in the data repository.

[0025] Furthermore, in step S1.3, the system terminal grants the tag code access rights within a certain period of time, including the following steps:

[0026] S3.1 storing the recognition results of the road section feature recognition in the direction of the community internal exit / the road section feature recognition in the direction of the community external entrance into the database;

[0027] S3.2 According to whether the person is a registered user or a permitted outsider, the gate opening time limit corresponding to the characteristic code of the person is assigned;

[0028] S3.3 The result of import gate feature recognition / export gate feature recognition is matched with the valid feature code, and an open command is sent to the gate after a successful match.

[0029] Furthermore, multiple label descriptions are input through the terminal recording system, and the system retrieves personnel information, entry and exit event records, face recognition records, and image information data that match the label descriptions by querying historical records in the data storage library.

[0030] Therefore, the functions and beneficial effects that can be achieved by the present invention are as follows:

[0031] 1. Image data collection and preprocessing: Install high-definition cameras at the entrance and exit of the community and inside and outside the community to capture and identify images of people. Achieving large-scale monitoring of people entering and leaving the community and passers-by is conducive to the reliability of community security, improving the efficiency of social security management, and improving the monitoring and early warning capabilities of strangers entering and leaving the community.

[0032] 2. Face recognition and label generation: Use a high-precision face recognition algorithm to identify faces in images. When identifying users in the community, they will correspond to the label code, and when entering the gate, only two label codes need to be detected and matched to complete the recognition. This can prevent users from repeatedly verifying their faces in front of the gate, delaying the entry and exit of other people.

[0033] 3. Control of outsiders: Face recognition and feature recognition work together to complete the automated permission, monitoring, and prompting of outsiders. Feature tags are assigned to outsiders, and facial recognition is prompted before permission is given, so that all people entering the community are verified by face and monitored on file, further improving the reliability of community security and the efficiency of social security management.

[0034] 4. Time validity of permission: The system assigns time-limited permission for entry and exit to the label code corresponding to the individual person. During the valid period, the individual user can be identified and enter and exit repeatedly. In situations where some users temporarily enter and exit the community, it can effectively shorten the verification time, further ensure the smooth passage of the gate machine, and avoid repeated verification.

[0035] 5. Quick search and filtering through labels The system provides quick search and filtering functions. Through image label descriptions, people with specific characteristics can be quickly located to improve screening and positioning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Arrange concept map for feature recognition points;

[0037] Figure 2 This is a flowchart of the process of identifying the characteristics of the road sections in the directions of the inner / outer / entrance of the community;

[0038] Figure 3 A schematic diagram of the process flow of generating label encoding and label description in feature recognition system;

[0039] Figure 4 A flowchart of granting the tag encoding access rights within a certain period of time to the system terminal;

[0040] Figure 5 This is a schematic diagram of the face recognition processing flow for captured images. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0042] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0043] See also Figure 1-Figure 5 The purpose of the present invention is to provide a smart community traffic record matching and recognition system and method based on image label enhancement. The smart community traffic record matching and recognition system based on image label enhancement includes a monitoring and recognition system, a terminal recording system, a prompt system and a gate system.

[0044] The terminal recording system is used to run programs, store data, and record the historical data of personnel entering and leaving. The historical data of personnel entering and leaving includes registered user information, registered user feature tags, unregistered personnel registration information, unregistered personnel feature tags and other data. The terminal recording system includes a terminal computer, which serves as the operating host of the terminal recording system and includes a processor, a display, a storage device and other necessary hardware devices that support the operation of the computer. The storage device stores the operation database, the feature item database and the data storage library.

[0045] The operation database stores the necessary programs for computer operation.

[0046] The feature item database is stored inside the storage device, which contains a large number of feature items indexed by tag names. The types of feature items mainly include clothing, hairstyle, body shape, carried items and accessories, which correspond to the clothing database, hairstyle database, body shape database, carried items database and accessories database stored inside the feature item database. Specifically, the feature item is an external feature that can be recognized by a computer with the human body as the main body, that is, the feature item is mainly a description of the external features related to the user, such as the feature items possessed by an individual: blue jeans, white short-sleeved shirt, short hair, height 175cm, black shoulder bag, and duckbill cap. That is, taking a single target subject as an individual, a single individual can have several feature items at the same time, and each feature item in the feature item database corresponds to a specific tag description and tag number.

[0047] The data repository is a large-capacity data repository, which stores the facial features of all registered users and all the input and output data of all registered users and unregistered users within a period of time.

[0048] The monitoring and identification system includes a monitoring system and an identification system. The monitoring system includes a number of high-definition monitoring cameras distributed at the entrances and exits of buildings and gates. The monitoring cameras can be distributed at the entrances and exits of each unit in the community, the necessary roads for the entrances and exits of buildings in the community, and the gates for the entrances and exits of the community to record and capture people passing by. At the same time, the distance and height of the camera are calibrated so that the height information of the captured person can be estimated in combination with the calibration parameters.

[0049] The recognition system includes a face recognition system and a feature recognition system. The high-definition surveillance camera in the monitoring system captures the overall image of a person, and then the face recognition system and feature recognition system extract the facial features and whole body features of the person respectively. Both the face recognition system and the feature recognition system use models trained by Haar cascade detectors (cascade classifiers) or Transformer and CLIP technologies. The feature recognition system includes height estimation, which uses an algorithm based on 3D posture estimation.

[0050] In actual application scenarios, the Cascade Classifier is an object detection algorithm based on Haar features, which is mainly used for face detection and identification of other objects. The Cascade Classifier has the characteristics of high detection accuracy and fast detection speed.

[0051] The detection process of the Haar-based cascade classifier can be simply described by the following steps. The first is image preprocessing, which involves normalizing and histogram equalization after image conversion. Next, the cascade classifier will calculate the Haar feature for each window at different positions and windows of different sizes in the image. The Haar feature is a method for calculating the difference between black and white rectangular boxes in an image. The calculated Haar feature value will be sent to the AdaBoost classifier for classification. If the output value of the classifier is greater than the preset threshold, it is considered that there is a target object in the current window. If all weak classifiers in the cascade classifier pass the verification of the current window, it is considered that the entire cascade classifier has detected the target object. Since the operating principle of the Haar cascade detector (cascade classifier) ​​is a prior art, it will not be described here.

[0052] The face recognition system uses the face recognition devices commonly used in the market today. Its principle is as follows: First, the camera captures a picture of the face or a real-time video stream. The system uses the Haar cascade detector to identify and locate the face area, and detect the face, eyes, and mouth. Finally, these features are compared with the facial features of all registered users in the data repository to finally identify the individual's identity.

[0053] The feature recognition system also uses the Haar cascade detector to identify and extract individual feature items. It uses the Haar cascade detector to identify and locate areas such as the head, upper body, lower body, and feet of a person, detects each area and compares the extracted data with the features in the feature item database, generates corresponding label descriptions and label numbers, and integrates several label descriptions and label numbers to form a collection of label descriptions and label codes to facilitate subsequent related operations.

[0054] Label encoding uses a language model based on Text Embedding technology to vectorize and then match. The Text Embedding language model uses the semantic information of the text. Texts with the same semantics but different expressions can be mapped to the same position, while texts with different semantics maintain corresponding distances in the vector space. Compared with traditional technologies, the label encoding generated by this method is more structured, easier to store, reduces dimensions, and improves matching and search efficiency; it has excellent semantic understanding, can better handle fuzzy descriptions and synonym matching, and improves matching accuracy.

[0055] The gate system is interconnected with the terminal recording system and the monitoring identification system through the network. The gate is a gate device installed at the entrance and exit of the community. The gate device is used to control the entry and exit of people and prevent illegal elements from entering and exiting at will.

[0056] The prompt system is a voice prompt device or display device installed near the gate system, which can provide operation prompts to people entering and exiting the gate channel through voice or text display.

[0057] See also Figure 1 The feature recognition points are arranged at the entrance and exit gates, the road sections in the direction of the community's internal exits, and the road sections in the direction of the community's external entrances. In the overall design of the distribution of monitoring equipment, a number of high-definition monitoring cameras corresponding to the gates are suspended above the entrance and exit gates of the community through poles; at the same time, high-definition monitoring cameras are installed on poles on each section of the community's internal exit direction and the section of the community's external entrance direction to shoot and record each individual person walking towards the gate.

[0058] See attached Figure 2 ,The smart community traffic record matching and recognition method based on image label enhancement includes the following steps:

[0059] S1.1 The surveillance equipment on the road sections in the direction of the internal exit of the community and the road sections in the direction of the external entrance of the community will capture high-definition images for the first time, capture images of individual people appearing in the images, transmit the videos and images to the storage device and recognition system in the system terminal, and record the time information at the same time.

[0060] S1.2 The face recognition system and feature recognition system detect, identify, and match face information and feature item information respectively, confirm the user individual through the face information, and confirm the label code and label description through the feature item information.

[0061] S1.3 The tag code and tag description are mapped to the user individual, and the system terminal grants the tag code access rights within a certain period of time.

[0062] The monitoring equipment of S1.4 entry and exit gates takes high-definition images for the second time, captures images of individuals appearing in the images, transmits the videos and images to the storage device and recognition system in the system terminal, and records the time information at the same time.

[0063] S1.5 The feature recognition system detects and recognizes feature item information, regenerates label codes and label descriptions, and searches and matches the regenerated label codes among users with time-limited permissions.

[0064] S1.6 If there is a match, the gate will be controlled to open in advance when the individual approaches the gate, and the prompt system will prompt the person to enter the designated gate; if there is no match, the regenerated tag code and tag description will be queried for a match in the user individual, and the person will be recorded at the same time, and then the gate will be opened.

[0065] Among them, see Appendix Figure 3 , the generation process of label encoding and label description in the feature recognition system in step S1.2 includes the following steps:

[0066] S2.1 The monitoring system transmits the captured high-definition images to the feature recognition system;

[0067] S2.2 The feature recognition system uses the Haar level detector to locate and select each region, each region represents a different feature item type;

[0068] S2.3 extracts the feature value in each feature item type, and compares the feature value with the features in the feature item database to find matching feature items;

[0069] S2.4 combines the numbers of several feature items and the descriptions corresponding to each feature item number to generate a label code and a label description;

[0070] S2.5 assigns the label code and label description to the data of the current individual and stores it in the data repository.

[0071] Among them, see Appendix Figure 4 The system terminal encodes the access rights within a certain period of time and the identification of the rights includes the following steps:

[0072] S3.1 storing the recognition results of the road section feature recognition in the direction of the community internal exit / the road section feature recognition in the direction of the community external entrance into the database;

[0073] S3.2 According to whether the person is a registered user or a permitted outsider, the gate opening time limit corresponding to the characteristic code of the person is assigned;

[0074] S3.3 The result of import gate feature recognition / export gate feature recognition is matched with the valid feature code, and an open command is sent to the gate after a successful match.

[0075] Among them, see Appendix Figure 2 The comparison process of face recognition in the direction of the community's internal exit and / or the direction of the community's external entrance includes the following steps:

[0076] S4.1 After facial recognition on the road section in the direction of the community's internal exit and / or the direction of the community's external entrance, the extracted facial information is compared with the facial features of registered users in the data repository; if the user is matched, a time-sensitive label code is assigned to the user and the steps continue; if the user is not matched, it is determined whether the face is obscured.

[0077] S4.2 If the face is blocked, the feature recognition result will be associated with the person; then when the person passes through the gate, the system will prompt to remove the blockage.

[0078] S4.3 If the face is not covered, the person is recorded as a foreigner, and the facial features are recorded and stored; the label description and label code of the feature items are recorded; and a time-limited entry and exit permit is granted.

[0079] See also Figure 5 After the monitoring device takes a photo of a single person, it first passes through the system's image quality filtering, then extracts the person's label features and encodes them, including height recognition, and stores the labels. Feature extraction also performs face recognition. If facial features are captured, the facial features are compared with the facial features of the people already entered in the database; if the comparison is successful, the travel record is associated with the user. If there is no result, it is recorded as a stranger and the travel record is created as a stranger. The stranger is classified and clustered according to facial features. If the comparison is unsuccessful, the feature label or the generated label code is matched with the label of the registered user. If the match is successful, it is recorded as the travel record of the registered person queried. If the match is unsuccessful, it is compared with the features in the stranger database. If successful, the travel record is matched to the travel record of the stranger. If unsuccessful, a new stranger record is created for this purpose and stored according to the features.

[0080] Example 1

[0081] In actual use scenarios, if a person approaches the community entrance along the main road opposite the community entrance, the monitoring device located on the main road far away from the gate takes an image and extracts feature items. The label description is: height 181cm, short hair, black long-sleeved jacket, white round-neck shirt, no glasses, jeans, sneakers, unrecognized watch, shoulder bag; the label code is "0181.0001.0024.2350.0258.0364.0451.2459.1255". The face recognition is the resident of 101 in the first unit of the first building. The system grants this label code the right to enter and exit within 20 minutes, stores this label code in the authorization list, and records the label information in the system. When the person approaches the gate, the monitoring device in front of the gate takes an image, extracts feature items, generates a second label code, compares the second label code with the label code in the authorization list, and matches successfully. The gate is pre-opened to allow the person to pass, and the entry and exit information is recorded in the system.

[0082] Example 2

[0083] In actual usage scenarios, if a person approaches the community entrance along the main road opposite the community exit, the surveillance equipment located on the main road in the community, far away from the gate, takes an image and extracts feature items. The label description is: height 181cm, short hair, black long-sleeved jacket, white round-neck shirt, no glasses, jeans, sneakers, unrecognized watch, shoulder bag; the label code is "0181.0001.0024.2350.0258.0364.0451.2459.1255". Face recognition does not find a matching user, and the detection does not cover the face, so it is recorded as an outsider. At the same time, the label information is recorded in the system, and the facial feature information is recorded and stored. The label description and the label code with a time limit are assigned to the person. The system assigns this label code entry and exit rights within 20 minutes and stores this label code in the authorization list. When a person approaches the gate, the monitoring equipment in front of the gate captures the image, extracts the feature items, generates a second tag code, compares the second tag code with the tag codes in the authorization list, and if a match is successful, the gate is pre-opened to allow the person to pass, and the entry and exit information is recorded in the system.

[0084] Example 3

[0085] In actual usage scenarios, if a person approaches the community entrance along the main road opposite the community exit, the surveillance equipment located on the main road in the community, far away from the gate, takes an image and extracts feature items. The label description is: height 181cm, short hair, black long-sleeved jacket, white round-neck shirt, no glasses, jeans, sneakers, unrecognized watch, shoulder bag; the label code is "0181.0001.0024.2350.0258.0364.0451.2459.1255". Face recognition does not find a matching user. After detection, the person covers his face and is recorded as a suspicious person. The label description and label code are assigned to the person, and the label information is recorded in the system. The system does not grant entry and exit permissions. When a person approaches the gate, the monitoring device in front of the gate captures the image, extracts the feature items, generates a second tag code, and compares the second tag code with the tag code in the authorization list. If the match is successful, the gate prompts the user to remove the occlusion through the display device. After the occlusion is removed, a second face recognition is performed (including extracting the recognized face information and comparing it with the facial features of registered users in the data storage database). If the user is matched, the gate is opened to allow the person to pass through, and the entry and exit information is recorded in the system. If the user is not matched, it is recorded as an outsider, the gate is opened to allow the person to pass through, and the entry and exit information is recorded in the system.

[0086] Example 4

[0087] In actual use scenarios, the in-and-out records of suspicious persons in the community are investigated. The administrator enters multiple label descriptions through the terminal recording system, such as the label description is: short hair, black long-sleeved jacket, white round-neck shirt, no glasses, jeans. The system queries the historical records in the data storage database and retrieves the matching personnel information, in-and-out event records, face recognition records, image information and other data for investigation.

[0088] Therefore, the functions and beneficial effects that can be achieved by the present invention are as follows:

[0089] 1. Image data collection and preprocessing: Install high-definition cameras at the entrance and exit of the community and inside and outside the community to capture and identify images of people. Achieving large-scale monitoring of people entering and leaving the community and passers-by is conducive to the reliability of community security, improving the efficiency of social security management, and improving the monitoring and early warning capabilities of strangers entering and leaving the community.

[0090] 2. Face recognition and label generation: Use a high-precision face recognition algorithm to identify faces in images. When identifying users in the community, they will correspond to the label code, and when entering the gate, only two label codes need to be detected and matched to complete the recognition. This can prevent users from repeatedly verifying their faces in front of the gate, delaying the entry and exit of other people.

[0091] 3. Control of outsiders: Face recognition and feature recognition work together to complete the automated permission, monitoring, and prompting of outsiders. Feature tags are assigned to outsiders, and facial recognition is prompted before permission is given, so that all people entering the community are verified by face and monitored on file, further improving the reliability of community security and the efficiency of social security management.

[0092] 4. Time validity of permission: The system assigns time-limited permission for entry and exit to the label code corresponding to the individual person. During the valid period, the individual user can be identified and enter and exit repeatedly. In situations where some users temporarily enter and exit the community, it can effectively shorten the verification time, further ensure the smooth passage of the gate machine, and avoid repeated verification.

[0093] 5. Quick search and filtering through labels The system provides quick search and filtering functions. Through image label descriptions, people with specific characteristics can be quickly located to improve screening and positioning efficiency.

[0094] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, technicians familiar with the field may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A smart community traffic record matching and recognition method based on image label enhancement, characterized in that: The following steps are involved: S1.1 The surveillance equipment on the road sections in the direction of the community’s internal exit and the road sections in the direction of the community’s external entrance takes high-definition images for the first time, captures images of individuals appearing in the images, and transmits the videos and images to the storage device and recognition system in the system terminal; S1.2 The face recognition system and feature recognition system detect, identify, and match face information and feature item information respectively, confirm the user individual through face information, and confirm the label code and label description through feature item information; S1.3 The tag code and tag description are mapped to the user individual, and the system terminal grants the tag code access rights within a certain period of time; S1.4 The monitoring equipment of the entrance and exit gates takes a high-definition image for the second time, captures the images of the individuals appearing in the image, and transmits the video and images to the storage device and recognition system in the system terminal; S1.5 The feature recognition system detects and recognizes feature item information, regenerates label codes and label descriptions, and searches and matches the regenerated label codes among users with time-limited permissions; S1.6 If there is a match, the gate will be controlled to open in advance when the individual approaches the gate, and the prompt system will prompt the person to enter the designated gate; if there is no match, the regenerated tag code and tag description will be queried for a match in the user individual, and the person will be recorded and the gate will be opened later.

2. According to the method for matching and identifying smart community traffic records based on image label enhancement according to claim 1, it is characterized in that: In the step S1.1 and step S1.4, the monitoring device records the time information while taking the high-definition images for the first and second time and transmitting them to the storage device in the system terminal.

3. According to the method for matching and identifying smart community traffic records based on image label enhancement in claim 1, it is characterized in that: The generation process of the label encoding and label description in step S1.2 includes the following steps: S2.1 The monitoring system transmits the captured high-definition images to the feature recognition system; S2.2 The feature recognition system uses the Haar level detector to locate and select each region, each region represents a different feature item type; S2.3 extracts the feature value in each feature item type, and compares the feature value with the features in the feature item database to find matching feature items; S2.4 combines the numbers of several feature items and the descriptions corresponding to each feature item number to generate a label code and a label description; S2.5 assigns the label code and label description to the data of the current individual and stores it in the data repository.

4. According to the method for matching and identifying smart community traffic records based on image label enhancement according to claim 1, it is characterized in that: In step S1.3, the system terminal grants the tag code access rights within a certain period of time, including the following steps: S3.1 storing the recognition results of the road section feature recognition in the direction of the community internal exit / the road section feature recognition in the direction of the community external entrance into the database; S3.2 According to whether the person is a registered user or a permitted outsider, the gate opening time limit corresponding to the characteristic code of the person is assigned; S3.3 The result of import gate feature recognition / export gate feature recognition is matched with the valid feature code, and an open command is sent to the gate after a successful match.

5. According to the method for matching and identifying smart community traffic records based on image label enhancement in claim 1, it is characterized in that: Multiple label descriptions are input through the terminal recording system, and the system retrieves personnel information, entry and exit event records, face recognition records, and image information data that match the label descriptions by querying historical records in a data storage database.

6. The method for matching and identifying traffic records in a smart community based on image label enhancement according to claim 1 is characterized in that: The method uses a smart community traffic record matching and recognition system based on image label enhancement, which includes a terminal recording system, a monitoring and recognition system, a prompt system, and a gate system; The terminal recording system includes a terminal computer, the terminal computer includes a storage device, and the storage device stores an operation database, a feature item database and a data storage library; the operation database stores the necessary programs for the computer operation; the feature item database contains feature items indexed by the tag name; the data storage library stores the facial features of all registered users and all the entry and exit data of all registered users and unregistered users within a period of time; The monitoring and recognition system includes a monitoring system and a recognition system; the monitoring system includes a number of high-definition monitoring cameras distributed at the building entrances and exits and gate channels; the recognition system includes a face recognition system and a feature recognition system; the face recognition system and the feature recognition system respectively recognize, extract, and compare individual faces and feature items, and generate corresponding label descriptions and label codes; The prompt system is a voice prompt device or display device installed near the gate system; the gate system is a gate device installed at the entrance and exit of the community.

7. The smart community traffic record matching and recognition system based on image tag enhancement according to claim 6 is characterized in that: The monitoring camera equipment is distributed at the entrances and exits of each unit in the community and at the community entrance and exit gates.

8. The smart community traffic record matching and recognition system based on image tag enhancement according to claim 6 is characterized in that: The face recognition system and feature recognition system adopt Haar cascade detector or Transformer, CLIP technology trained model; the feature recognition system includes height estimation, and the height estimation adopts an algorithm based on 3D posture estimation.

9. The smart community traffic record matching and recognition system based on image tag enhancement according to claim 6 is characterized in that: The types of feature items mainly include clothing, hairstyle, body shape, carried items and accessories, which correspond to the clothing database, hairstyle database, body shape database, carried items database and accessories database stored in the feature item database respectively.

Images