Intelligent online patrol comprehensive management platform

Abstract

The invention relates to an intelligent online patrol comprehensive management platform, and belongs to the technical field of intelligent monitoring and management. The platform adopts a multi-layer distributed architecture and comprises a data acquisition layer, a transmission layer, a platform layer and an application layer. The platform layer integrates a video resource unified access module, an artificial intelligence behavior analysis module, a patrol task scheduling engine, a multi-stage patrol management module and an early warning disposal closed-loop management module. Multi-source video streams are integrated through a standard protocol, video streams of key areas such as examination rooms and test paper storage rooms are analyzed in real time by using a deep learning algorithm, and illegal behaviors such as abnormal number of people and going out with articles are automatically identified and intelligently early warned. Meanwhile, the system supports multi-stage collaborative patrol and disposal process closed-loop management, full-process automation and intelligentization from problem discovery, early warning pushing, task disposal to result feedback are achieved, the supervision efficiency, accuracy and response speed of examination patrol are remarkably improved, and manual dependence is reduced.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring and information management technology, and in particular, it is an intelligent online patrol and comprehensive management platform. Background Technology

[0002] With the popularization of video surveillance technology, online patrol systems have been widely used in fields such as examination supervision and security management. However, existing technologies mainly have the following shortcomings: System isolation and information barriers: Each functional module (such as video surveillance, behavior analysis, and alarm) is often deployed independently, and data cannot be shared, forming information silos and making it difficult to achieve unified supervision across regions and levels.

[0003] Reliance on manual labor and inefficiency: It mainly relies on real-time monitoring or post-event playback by humans. In scenarios with many monitoring points and large amounts of data (such as national examinations), fatigue can easily lead to oversights in supervision, resulting in inefficiency.

[0004] Insufficient early warning capabilities: Lacking intelligent analysis capabilities, it is unable to automatically identify abnormal behaviors (such as cheating in the examination room or violations of test paper storage regulations), and the response relies on subjective human judgment, resulting in delayed early warnings.

[0005] Disconnected process management: After problems are discovered during inspections, the handling process (recording, reporting, and feedback) is not well connected, making it difficult to form an effective management loop.

[0006] Limited coverage: It is difficult to achieve comprehensive and real-time remote video supervision of scattered lower-level test centers or special areas (such as test paper storage rooms).

[0007] Therefore, there is an urgent need for a comprehensive inspection and management platform that can integrate multi-source data, achieve intelligent analysis, support collaborative management, and form a closed-loop disposal mechanism. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the existing technology and propose an intelligent online inspection and comprehensive management platform. The platform integrates architecture and artificial intelligence technology to realize intelligent and automated inspection and closed-loop management of multiple scenarios such as examination rooms and test paper storage rooms.

[0009] The technical problem solved by this invention is achieved through the following technical solution: An intelligent online patrol and management platform adopts a multi-layered distributed architecture, including a data acquisition layer, a transmission layer, a platform layer, and an application layer, which are connected sequentially. The data acquisition layer is used to access front-end monitoring equipment through standard protocols to collect video data from various areas of the examination room, lower-level examination sites, and the test paper storage room. The transmission layer is used to encode, compress, and encrypt the video data for transmission. The platform layer serves as the core of the system, and the application layer provides a web management terminal and mobile application for human-computer interaction, enabling real-time preview, alarm information push, patrol task configuration, and handling feedback.

[0010] Furthermore, the platform layer includes: a unified video resource access module, an artificial intelligence behavior analysis module, a patrol task scheduling engine, a multi-level patrol management module, and an early warning and response closed-loop management module. The unified video resource access module integrates and manages multi-source video streams from the data acquisition layer; the artificial intelligence behavior analysis module performs real-time analysis of the video streams based on deep learning algorithms, automatically identifies preset abnormal behaviors, and generates alarm events; the abnormal behaviors include at least abnormal numbers of people in the exam paper storage room, carrying items outside the room, and using unauthorized tools; the patrol task scheduling engine configures and manages automated patrol plans and schedules the artificial intelligence behavior analysis module to execute analysis tasks; the multi-level patrol management module supports collaborative patrols and video previews between upper and lower-level patrol centers; and the early warning and response closed-loop management module tracks and manages the alarm events throughout the entire process, achieving a closed loop from early warning generation and task allocation to result feedback.

[0011] Furthermore, the AI ​​behavior analysis module includes: an abnormal number of people identification unit, a carried-out-of-area detection unit, an unauthorized tool usage identification unit, and an abnormal on-duty personnel alarm unit. The abnormal number of people identification unit is used to count the number of people in a specific area through face detection and multi-target tracking algorithms, and compare it with a preset threshold to determine whether the number of people is abnormal. The carried-out-of-area detection unit is used to identify whether people are carrying unauthorized items out of a preset area using target detection and instance segmentation technology. The unauthorized tool usage identification unit is used to identify preset unauthorized tools based on image classification algorithms. The abnormal on-duty personnel alarm unit is used to verify the identity and number of on-duty personnel through face recognition technology.

[0012] Moreover, the abnormal number of people identification unit uses an improved YOLOv5s model for face detection and combines it with the DeepSORT algorithm to achieve multi-target tracking. The formula for counting the number of people is: N_t = Σ_{i=1}^{M} I(s_i>τ), where N_t is the number of people counted at time t, M is the total number of detection boxes, s_i is the confidence score of the i-th face, and τ is a preset threshold.

[0013] Moreover, the patrol task scheduling engine is based on a lightweight timed task thread pool, which can parse the patrol plan configured by the administrator into timed tasks and automatically schedule the artificial intelligence behavior analysis module to analyze the specified video stream at a specified time.

[0014] Furthermore, the multi-level patrol management module enables interconnection and interoperability between different levels of systems based on the session initiation protocol, and supports dynamic display of patrol trajectories based on the geographic information system platform.

[0015] Furthermore, the early warning and handling closed-loop management module has a built-in workflow engine for defining and managing the handling process of alarm events. The process includes at least event reporting, task allocation, on-site handling, and result feedback, and supports the inspected unit to submit rectification measures online.

[0016] Furthermore, the data acquisition layer supports access to monitoring devices of different brands and models via ONVIF, RTSP protocols, and device SDKs.

[0017] The advantages and positive effects of this invention are: 1. This invention achieves closed-loop management of the entire process from video acquisition, intelligent analysis, early warning push to handling feedback, which greatly improves regulatory efficiency and problem-solving speed.

[0018] 2. This invention achieves proactive intelligent early warning through artificial intelligence algorithms, which greatly reduces the reliance on manual monitoring and lowers the risk of missed reports due to negligence.

[0019] 3. This invention breaks down the information silos of traditional systems and achieves deep integration and collaboration of subsystems such as video surveillance, behavior analysis, and process management.

[0020] 4. This invention adopts a modular design, which can be flexibly configured with functions as needed. It is not only suitable for examination inspection, but can also be extended to multiple fields such as urban management and production safety, making it highly adaptable.

[0021] 5. This invention utilizes a deep learning model trained for specific scenarios, which has a high accuracy rate in identifying abnormal behavior and effectively reduces false positives and false negatives. Detailed Implementation

[0022] A smart online patrol and management platform, employing a multi-layered distributed architecture, includes: The data acquisition layer is used to connect to front-end devices such as examination room cameras, infrared PTZ cameras in the test paper storage room, and network video recorders at lower-level examination sites via standard protocols such as ONVIF and RTSP or device SDKs.

[0023] The transport layer, based on the TCP / IP network, uses encoding technologies such as H.265 to compress video streams and transmits control signaling through encryption protocols such as HTTPS to ensure data security and real-time performance.

[0024] The platform layer, the core of the system, contains multiple functional modules: Unified video resource access module: Integrates video streams from different sources, providing unified real-time preview, PTZ control, video playback, and group permission management functions.

[0025] Artificial intelligence behavior analysis module: It uses deep learning algorithms (such as YOLOv5, Mask R-CNN) to perform real-time analysis of video streams, specifically targeting scenarios such as exam paper counting, to identify violations such as abnormal number of people, carrying items out of the office, and using mobile phones, and automatically trigger alarms.

[0026] Patrol task scheduling engine: Based on a lightweight timed task framework (such as XXL-JOB), it allows administrators to configure patrol plans (time, location, analysis rules) and automatically schedule the execution of analysis tasks.

[0027] Multi-level inspection management module: Supports multi-level inspection systems at the provincial, municipal, and examination site levels. Superiors can preview videos from subordinate levels in real time, display inspection trajectories based on GIS, and achieve cross-system communication via SIP protocol.

[0028] Early warning and response closed-loop management module: When an anomaly is detected, an early warning event is automatically generated and notified through multiple channels such as the platform and SMS. The built-in workflow engine tracks the entire process of an event from reporting, assignment, handling to feedback, and supports rectification data entry, forming a closed-loop management system.

[0029] Application layer: Provides a web-based management backend and mobile APP for patrol and management personnel to perform real-time monitoring, receive alarms, process tasks, and view reports.

[0030] The working method of this invention is as follows: Step 1: System Deployment and Device Access A platform server cluster, including application servers, AI analysis servers, and database servers, will be deployed in the examination center's computer room. Existing surveillance cameras at each examination site, as well as newly added high-definition infrared cameras in the test paper storage room, will be integrated into the platform's unified video resource module via the ONVIF / RTSP protocol or the manufacturer's SDK.

[0031] Step 2: Patrol Plan Configuration Administrators log in to the system via a web interface. For example, for the "exam paper counting" process, a patrol plan can be configured: starting one hour before the exam begins, the "abnormal number of people detection" (threshold set to 2-4 people), "detection of items taken out of the room," and "identification of unauthorized tool use" functions can be enabled for "camera #1 in the exam paper storage room." The patrol task scheduling engine will convert this plan into a scheduled task.

[0032] Step 3: Real-time monitoring and intelligent analysis During system operation, the AI ​​behavior analysis module performs real-time analysis on the specified video stream.

[0033] If a fifth person appears in the count area, the abnormal number identification unit will immediately calculate and determine it as abnormal, triggering a "number of people exceeds the limit" alarm.

[0034] If the system detects through Mask R-CNN that someone is carrying an unauthorized document bag and moving towards the door, the item carrying detection unit will trigger an alarm for "suspected carrying of exam papers out of the house".

[0035] If a staff member is detected holding a mobile phone, the unauthorized tool use identification unit will generate an alarm for "using unauthorized communication tools".

[0036] Step 4: Early Warning Response and Closed-Loop Management Once an alarm is triggered, the early warning response and closed-loop management module will perform the following operations: The abnormal target box is highlighted on the monitoring screen.

[0037] The system automatically generates early warning work orders and notifies the examination affairs supervisor and on-site inspectors via platform messages and SMS.

[0038] The work order is assigned to the designated personnel. The personnel receive the task through a mobile app, go to the site to verify, and report the handling results through the app (such as "the situation is true and has been stopped" or "false report, the situation is normal").

[0039] The entire process was recorded. If the problem is verified, the relevant unit can be required to enter corrective measures into the system to complete the closed loop.

[0040] Step 5: Data Statistics and Report Generation The system backend automatically generates inspection statistical analysis reports on a regular basis (such as daily or after each exam), including statistics on alarm event types, timeliness of handling at each test center, and completion status of rectification, providing data support for management decisions.

[0041] It should be emphasized that the embodiments described in this invention are illustrative rather than limiting. Therefore, this invention includes, but is not limited to, the embodiments described in the specific implementation. Any other implementations derived by those skilled in the art based on the technical solutions of this invention are also within the scope of protection of this invention.

Claims

1. An intelligent online patrol and comprehensive management platform, characterized in that, The system adopts a multi-layered distributed architecture, including a data acquisition layer, a transmission layer, a platform layer, and an application layer. These layers are connected sequentially. The data acquisition layer is used to access front-end monitoring equipment via standard protocols to collect video data from various areas of the examination room, lower-level examination sites, and the test paper storage room. The transmission layer is used to encode, compress, and encrypt the video data before transmission. The platform layer serves as the core of the system, while the application layer provides a web management interface and mobile applications for human-computer interaction, enabling real-time preview, alarm information push, patrol task configuration, and handling feedback.

2. The intelligent online patrol and integrated management platform according to claim 1, characterized in that, The platform layer includes: a unified video resource access module, an artificial intelligence behavior analysis module, a patrol task scheduling engine, a multi-level patrol management module, and an early warning and response closed-loop management module. The unified video resource access module integrates and manages multi-source video streams from the data acquisition layer. The artificial intelligence behavior analysis module performs real-time analysis of the video streams based on deep learning algorithms, automatically identifies preset abnormal behaviors, and generates alarm events. These abnormal behaviors include at least abnormal numbers of people in the test paper storage room, carrying items outside the room, and using unauthorized tools. The patrol task scheduling engine configures and manages automated patrol plans and schedules the artificial intelligence behavior analysis module to execute analysis tasks. The multi-level patrol management module supports collaborative patrols and video previews between upper and lower-level patrol centers. The early warning and response closed-loop management module tracks and manages the alarm events throughout the entire process, achieving a closed loop from early warning generation and task allocation to result feedback.

3. The intelligent online patrol and integrated management platform according to claim 2, characterized in that, The AI ​​behavior analysis module includes: an abnormal number of people identification unit, a carried-out-of-area detection unit, an unauthorized tool usage identification unit, and an abnormal on-duty personnel alarm unit. The abnormal number of people identification unit is used to count the number of people in a specific area through face detection and multi-object tracking algorithms, and compare it with a preset threshold to determine whether the number of people is abnormal. The carried-out-of-area detection unit is used to identify whether people are carrying unauthorized items out of a preset area using object detection and instance segmentation technology. The unauthorized tool usage identification unit is used to identify preset unauthorized tools based on image classification algorithms. The abnormal on-duty personnel alarm unit is used to verify the identity and number of on-duty personnel through face recognition technology.

4. The intelligent online patrol and integrated management platform according to claim 2, characterized in that, The abnormal number of people identification unit uses an improved YOLOv5s model for face detection and combines it with the DeepSORT algorithm to achieve multi-target tracking. The formula for counting the number of people is: N_t = Σ_{i=1}^{M} I(s_i > τ), where N_t is the number of people counted at time t, M is the total number of detection boxes, s_i is the confidence score of the i-th face, and τ is a preset threshold.

5. The intelligent online patrol and integrated management platform according to claim 1, characterized in that, The patrol task scheduling engine is implemented based on a lightweight timed task thread pool, which can parse the patrol plan configured by the administrator into timed tasks and automatically schedule the artificial intelligence behavior analysis module to analyze the specified video stream at a specified time.

6. The intelligent online patrol and integrated management platform according to claim 1, characterized in that, The multi-level patrol management module enables interconnection and interoperability between different levels of systems based on the session initiation protocol, and supports dynamic display of patrol trajectories based on the geographic information system platform.

7. The intelligent online patrol and integrated management platform according to claim 1, characterized in that, The early warning and response closed-loop management module has a built-in workflow engine for defining and managing the handling process of alarm events. The process includes at least event reporting, task allocation, on-site handling, and result feedback, and supports the inspected unit to submit rectification measures online.

8. The intelligent online patrol and integrated management platform according to any one of claims 1-6, characterized in that, The data acquisition layer supports access to monitoring devices of different brands and models via ONVIF, RTSP protocols, and device SDKs.