CN-121982598-A - Personnel digital comprehensive management method based on camera and related equipment

CN121982598ACN 121982598 ACN121982598 ACN 121982598ACN-121982598-A

Abstract

The invention discloses a camera-based personnel digital comprehensive management method and related equipment, and relates to the technical field of personnel management based on artificial intelligence, wherein the method comprises the following steps: collecting video stream data containing personnel through a plurality of cameras deployed in a closed management area, preprocessing the video stream data, and then extracting various biological characteristics of target personnel; the method comprises the steps of obtaining a person identity identifier of a target person, performing behavior semantezation on current position information, current wearing information and current posture information corresponding to the person identity identifier of the target person according to a pre-stored knowledge graph to generate a behavior event, automatically triggering a pre-defined response strategy according to the behavior event, and displaying on a visual map in real time. The invention breaks the information island among the management systems, realizes the integrated operation of personnel identity management, safety management and work efficiency management, and provides a complete digital management solution for the closed places.

Inventors

LI XUEJING
ZHAO YUAN
GENG YIDE
LIU TONG
WANG HONGLI

Assignees

太原理工大学

Dates

Publication Date: 20260505
Application Date: 20251201

Claims (10)

1. The method for the digital comprehensive management of the personnel based on the camera is characterized by comprising the following steps of: collecting video stream data containing personnel through a plurality of cameras deployed in a closed management area, and preprocessing the video stream data; extracting various biological characteristics of a target person from the preprocessed video stream data; Based on a pre-established mapping relation among the personnel identity identifier, the face image and the human body image of each personnel, carrying out weighted fusion and joint reasoning on various biological characteristics of the target personnel, and outputting the personnel identity identifier of the target personnel; Performing behavior semanteme on current position information, current wearing information and current posture information corresponding to the personnel identity identifier of the target personnel according to a pre-stored position knowledge graph, wearing knowledge graph and posture knowledge graph to generate a behavior event; And automatically triggering a predefined response strategy according to the behavior event, and displaying the personnel identity, the position track, the wearing and the gesture actions of the target personnel on a visual map in real time.
2. The method for digital integrated management of camera-based personnel of claim 1, further comprising tracking a movement track of the target personnel in a closed management area by utilizing a spatial relationship among multiple cameras and a target re-recognition technology based on a personnel identity identifier of the target personnel, and dynamically updating personnel states of the target personnel.
3. The method for digital integrated management of camera-based personnel of claim 1, wherein the plurality of biological features comprises coarse-grained features and fine-grained features, wherein the extracting the plurality of biological features of the target personnel from the preprocessed video stream data comprises: Extracting coarse granularity characteristics of a target person from the preprocessed video stream data by using a human body pre-training big model, and extracting fine granularity characteristics of the target person from the preprocessed video stream data by using a special model trained based on the human body pre-training big model, wherein the coarse granularity characteristics comprise apparent colors, body types, sexes and whether big articles are carried or not, and the fine granularity characteristics comprise five sense organs, ages, textures, clothes patterns, accessories and gestures.
4. The method for digitized integrated management of personnel based on cameras of claim 3 wherein the behavior events comprise a position-related behavior event, a wearing-related behavior event and a gesture-related behavior event, wherein the behavior semantically comprises the steps of performing behavior semantication on current position information, current wearing information and current gesture information corresponding to a personnel identity identifier of the target personnel according to a pre-stored position knowledge graph, a wearing knowledge graph and a gesture knowledge graph, and generating the behavior events, wherein the behavior events comprise: Inquiring from the position knowledge graph based on the personnel identity identifier, the current time and the current position information of the target personnel to obtain a position-related behavior event, wherein the position-related behavior event comprises an area invasion state, an on-duty state or an off-duty state, inquiring from the wearing knowledge graph based on the personnel identity identifier, the current time, the current position information and the current wearing information of the target personnel to obtain a wearing-related behavior event, wherein the wearing-related behavior event comprises whether to wear a specified garment or not, and inquiring from the gesture knowledge graph based on the personnel identity identifier, the current time, the current position information and the current gesture information of the target personnel to obtain a gesture-related behavior event, wherein the gesture-related behavior event comprises a normal gesture, an abnormal gesture, standing, walking, running, falling, loitering, gathering, taking a frame, smoking or sleeping.
5. The system is characterized by comprising a data acquisition preprocessing module, a biological characteristic extraction module, a personnel identity acquisition module, a behavior event acquisition module and a real-time display module; The data acquisition preprocessing module is used for acquiring video stream data containing personnel through a plurality of cameras deployed in a closed management area and preprocessing the video stream data; the biological feature extraction module is used for extracting various biological features of a target person from the preprocessed video stream data; The personnel identity acquisition module is used for carrying out weighted fusion and joint reasoning on various biological characteristics of a target personnel based on a pre-established mapping relation among personnel identity identifiers, face images and body images of each personnel and outputting the personnel identity identifiers of the target personnel; the behavior event acquisition module is used for performing behavior semantezation on current position information, current wearing information and current posture information corresponding to the personnel identity identifier of the target personnel according to a pre-stored position knowledge graph, wearing knowledge graph and posture knowledge graph to generate a behavior event; The real-time display module is used for automatically triggering a predefined response strategy according to the behavior event and displaying the personnel identity, the position track, the wearing and the gesture actions of the target personnel on the visual map in real time.
6. The system for digitized integrated management of camera-based personnel of claim 5 further comprising a trajectory tracking module for tracking a trajectory of movement of said target personnel within a closed management area using spatial relationships between multiple cameras and target re-recognition techniques based on a personnel identity of said target personnel and dynamically updating a personnel status of said target personnel.
7. The system for digital integrated management of camera-based personnel of claim 5, wherein the plurality of biological features comprises coarse-grained features and fine-grained features, wherein the biological feature extraction module is specifically configured to extract the coarse-grained features of the target personnel from the preprocessed video stream data using a human pre-training large model, and extract the fine-grained features of the target personnel from the preprocessed video stream data using a dedicated model trained based on the human pre-training large model, wherein the coarse-grained features comprise apparent color, body shape, gender, and whether large items are carried, and wherein the fine-grained features comprise five sense organs, age, texture, clothing pattern, accessories, and gestures.
8. The system for digital integrated management of personnel based on cameras according to claim 7, wherein the behavior events comprise a position-related behavior event, a wearing-related behavior event and an attitude-related behavior event, the behavior event acquisition module is specifically configured to query from the position knowledge graph based on a personnel identity identifier, a current time and current position information of the target personnel, obtain the position-related behavior event, the position-related behavior event comprises an area invasion state, an on-duty state or an off-duty state, query from the wearing knowledge graph based on the personnel identity identifier, the current time, the current position information and the current wearing information of the target personnel, obtain the wearing-related behavior event, wherein the wearing-related behavior event comprises whether to wear a prescribed garment or not, whether to wear a prescribed device, and query from the attitude knowledge graph based on the personnel identity identifier, the current time, the current position information and the current attitude information of the target personnel, obtain the attitude-related behavior event, wherein the attitude-related behavior event comprises a normal attitude, an abnormal attitude, a walking, a running, a falling, a sleeping, a leaning, a standing, a leaning, or a standing up.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing a method of camera-based personnel digital integrated management as claimed in any one of claims 1 to 4 when the computer program is executed.
10. A computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and when executed by a processor, the computer program implements a method for digitally and comprehensively managing personnel based on a camera according to any one of claims 1 to 4.

Description

Personnel digital comprehensive management method based on camera and related equipment Technical Field The invention relates to the technical field of personnel management based on artificial intelligence, in particular to a camera-based personnel digital comprehensive management method and related equipment. Background Management of personnel in relatively closed sites such as parks, factories, buildings, businesses, and the like presents multiple challenges. Personnel in these venues are engaged in various behavioral activities while facing various security risks. In order to realize effective management, all-round monitoring of identity management, safety management and work efficiency management is needed for the personnel, including accurately identifying personnel identity, accurately positioning the area, checking wearing condition in detail, comprehensively analyzing behavioral activities, thereby completely grasping which personnel are where, whether safe and what to do. Specific regulatory requirements include, but are not limited to, determining whether a person is in a work area, unauthorized access to an area or dangerous area, confirming whether a person is wearing prescribed safety equipment, identifying whether a person is working or resting, detecting whether a person has abnormal behavior, and the like. There are various technical means to try to solve the above management requirements, and separate technical systems are formed. In the field of identity recognition, a traditional feature recognition technology based on card swiping recognition and fingerprint recognition is mainly adopted, in the field of safety monitoring, a technical scheme of combining manual supervision with a video monitoring system is commonly applied, and in the field of behavior analysis, a personnel detection and gesture recognition technology based on a deep learning algorithm is deployed. The technical means are respectively embodied in a plurality of independent systems such as an access control system, an attendance system, a video monitoring system, a visitor management system and the like, and each system provides personnel management functions from a specific angle. However, in-depth analysis found that the prior art solution has three significant drawbacks. Firstly, the accuracy and stability of the identity recognition method are obviously insufficient, and the method is characterized in that security holes such as card punching and trailing exist in card punching recognition and fingerprint recognition, and the recognition rate is obviously reduced when the face recognition technology is simply relied on to face an application scene with a mask or poor illumination condition, and meanwhile, the generalization capability of the deep learning model in practical application is directly insufficient due to obvious differences of data characteristics of different places. And secondly, each management system runs independently to form a serious information island phenomenon, so that data sharing and business linkage cannot be realized, such as the fact that a security monitoring system lacks the identity recognition capability of illegal personnel, the fact that an access control attendance system cannot effectively sense the specific behaviors of the personnel in an authorized area, and the like. In addition, the existing method also lacks of deep mining and analysis of data such as personnel movement tracks, aggregation behaviors and residence time, and advanced management and prediction functions such as hot spot area analysis and personnel density early warning are difficult to provide. In summary, the urgent need for integrated management of personnel identity, safety and work efficiency in a closed place is difficult to be met by the prior art system, and development of a comprehensive solution capable of effectively integrating various personnel data, realizing accurate identity recognition, having intelligent analysis capability and supporting an active management mode is needed. The innovative solution needs to break through the limitation of the prior art, realize advanced functions such as multi-feature collaborative identification, multi-system deep fusion, intelligent early warning and the like, and therefore improve the overall management efficiency. Disclosure of Invention The invention aims to solve the technical problems of the prior art, and particularly provides a camera-based personnel digital integrated management method and related equipment, wherein the method comprises the following steps of: 1) The invention provides a camera-based personnel digital comprehensive management method, which comprises the steps of collecting video stream data containing personnel through a plurality of cameras deployed in a closed management area, preprocessing the video stream data, extracting various biological characteristics of target personnel from the preprocessed video stream data, carrying out wei