A method for intelligent event reporting and closing based on a CV model

Abstract

The application belongs to the technical field of city operation management, and discloses a method for intelligent event reporting and case closing based on a CV model. The application realizes real-time intelligent reporting by using a fine-tuned Qwen-VL large model to analyze image data in real time, realizes automatic identification and timely reporting of city events, significantly improves event response speed, and reduces manual patrol cost. Efficient intelligent case closing: by using a deep learning algorithm to compare photos before and after disposal, objectively and accurately judging the event disposal result, greatly reducing the workload of manual auditing, improving the efficiency of case closing, and ensuring the fairness and consistency of the city event management process. Cross-modal understanding enhances recognition accuracy: the cross-modal understanding ability of the Qwen-VL large model helps to accurately identify various events in complex urban environments, reduces the risk of false positives and false negatives, and improves the overall efficiency of city operation management.

Description

Technical Field

[0001] This invention belongs to the field of urban operation management technology, and in particular relates to a method for intelligent event reporting and case closure based on a CV model. Background Technology

[0002] Existing technological solutions for incident reporting, handling, and case closure in urban operation management primarily rely on manual patrols, public complaints, and operator registration for incident reporting. Through task assignment, on-site handling, information feedback, and case closure review, a closed-loop management system for urban management issues is achieved. While some steps may incorporate digital tools, the overall process remains largely manual, with relatively limited automation, particularly in incident identification and intelligent case closure, where further improvements in automation and intelligence are needed.

[0003] Based on the above analysis, the problems and shortcomings of the existing technology are as follows: (1) Event reporting relies on manual patrols and proactive reporting: The existing system has failed to achieve automatic machine identification and timely reporting of urban events, resulting in limited event resolution speed and is easily affected by factors such as manpower coverage and patrol frequency, which may lead to event omissions.

[0004] (2) Low efficiency of case closure review: Manually comparing photos before and after the handling to determine whether the incident has been properly handled is not only time-consuming and labor-intensive, but also highly subjective and easily affected by human factors, making it difficult to guarantee accuracy.

[0005] The technical problem that this application aims to solve is: This enables automatic identification and intelligent reporting of urban incidents, reducing the burden of manual patrols and improving incident response speed and coverage.

[0006] Develop an intelligent case closure method based on a large CV model to automatically compare photos before and after the handling process, objectively and efficiently determine the completion status of the case handling, reduce the workload of staff, and improve the efficiency of case closure review. Summary of the Invention

[0007] To address the problems existing in the prior art, this invention provides a method for intelligent event reporting and case closure based on a CV model.

[0008] This invention is implemented as follows: A method for intelligent event reporting and case closure based on a CV model includes: Step 1, Data Input: Real-time access to urban surveillance video and periodic capture of static images of designated areas; Step 2, event recognition; Step 3, intelligent reporting; Step 4: Upload the processing results; Step 5: Before and after comparison and analysis; Step 6: Intelligent judgment and automatic case closure.

[0009] Furthermore, the event identification: The Qwen-VL large model analyzes input video frames or images to identify whether there are management category events within the detection range. During the identification process, the model utilizes its cross-modal understanding capabilities to make judgments by combining image features and semantic information.

[0010] Furthermore, the intelligent reporting: When the model identifies an event, it immediately and automatically assembles an event report containing information such as the event type, location, time, and relevant image evidence, and automatically pushes it to the city operation management platform; the platform then assigns the corresponding staff to handle the event based on the geographical location information.

[0011] Furthermore, the processing results are uploaded: After handling the incident on the city operation management platform, staff members upload photos of the scene after the incident.

[0012] Furthermore, the before-and-after comparative analysis: The Qwen-VL large model receives photo pairs before and after processing, uses deep learning algorithms to identify objects or scenes in the two sets of photos, extracts relevant element features, analyzes and compares them, and identifies changes in the event area.

[0013] Furthermore, the intelligent judgment and automatic case closure: Based on the comparative analysis results, the model objectively judges whether the event has been effectively handled; if it is determined that the handling is completed, the platform automatically closes the event; otherwise, the staff will conduct secondary handling or submit supplementary relevant information.

[0014] Another object of the present invention is to provide a system for intelligent event reporting and closure based on a CV model, which implements the method for intelligent event reporting and closure based on a CV model as described in any one of claims 1-6, characterized in that the system for intelligent event reporting and closure based on a CV model comprises: The data input module is used to access urban surveillance video in real time and periodically capture static images of designated areas; The event recognition module is used by the Qwen-VL large model to analyze the input video frames or images and identify whether there are management category events within the detection range. During the recognition process, the model uses its cross-modal understanding capabilities to make judgments by combining image features and semantic information. The intelligent reporting module is used to automatically assemble an event report containing information such as the event type, location, time, and relevant image evidence when the model identifies an event, and automatically push it to the city operation management platform; the platform then assigns the corresponding staff to handle the event based on the geographical location information. The incident handling result upload module is used by staff to upload photos of the scene after handling the incident on the city operation management platform; The before-and-after comparison analysis module is used by the Qwen-VL large model to receive photo pairs before and after processing, and to identify objects or scenes in the two sets of photos through deep learning algorithms, extract relevant element features and perform analysis and comparison to identify changes in the event area. The case closure module is used by the model to objectively determine whether an event has been effectively handled based on the comparative analysis results. If it is determined that the event has been handled, the platform will automatically close the event; otherwise, the staff will conduct secondary processing or submit supplementary relevant information.

[0015] Another object of the present invention is to provide a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the method for intelligent event reporting and case closure based on the CV model.

[0016] Another object of the present invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method for intelligent event reporting and case closure based on the CV model.

[0017] Another objective of this invention is to provide an information data processing terminal for implementing the intelligent event reporting and case closure system based on the CV model.

[0018] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows: Real-time urban event recognition technology based on Qwen-VL large model: Utilizing the fine-tuning capability of pre-trained large model, it achieves accurate recognition and intelligent reporting of urban events in static images.

[0019] Application of cross-modal understanding in event recognition: The model integrates image features and semantic information to improve the accuracy of event recognition in complex urban scenarios.

[0020] Intelligent case closure algorithm: By comparing photos before and after the handling of a case using deep learning algorithms, the algorithm objectively judges the completion status of the case handling and achieves automatic case closure.

[0021] Points to be protected Intelligent reporting method and system architecture based on Qwen-VL large model.

[0022] The specific application steps and strategies of cross-modal understanding in urban event recognition.

[0023] A deep learning-based intelligent event recognition and discovery algorithm and its integration into an urban operation management platform.

[0024] The intelligent case-closing algorithm combining deep learning and its integration into the urban operation management platform.

[0025] What are the technical advantages of this application proposal compared to existing technologies? Real-time intelligent reporting: By using a finely tuned Qwen-VL large model to analyze image data in real time, the system can automatically identify and promptly report urban events, significantly improving the speed of event response and reducing the cost of manual inspections.

[0026] Efficient and intelligent case closure: Using deep learning algorithms to compare photos before and after the handling, the system can objectively and accurately judge the outcome of the incident, significantly reducing the workload of manual review, improving case closure efficiency, and ensuring the fairness and consistency of the city's incident management process.

[0027] Cross-modal understanding enhances recognition accuracy: The cross-modal understanding capability of the Qwen-VL large model helps to accurately identify various events in complex urban environments, reduce the risk of false alarms and missed alarms, and improve the overall efficiency of urban operation and management.

[0028] What are the technical advantages of this application proposal compared to existing technologies? Real-time intelligent reporting: Automated Identification: This invention utilizes a finely tuned Qwen-VL large model to automatically identify predetermined urban events in images, such as exposed garbage and illegally parked vehicles, without the need for manual patrols or photo reporting. This significantly reduces reliance on manual labor, greatly improves the timeliness and coverage of event detection, and ensures that urban issues receive a faster and more comprehensive response.

[0029] Enhanced real-time performance: Compared to existing technologies that can only perform post-event analysis on already taken photos, the system proposed in this application can analyze image data in real time, instantly identify and report newly occurring events, shorten the time interval from the occurrence of an event to its reporting, and help improve the emergency response speed of urban management departments.

[0030] Efficient and intelligent case closure: Objective judgment: Through deep learning algorithms, the Qwen-VL large model can accurately compare photos before and after the handling of the incident, objectively evaluate the effect of the handling of the incident, avoid subjective differences and judgment errors that may be caused by manual review, and ensure the fairness and consistency of the case closure decision.

[0031] Reduced workload: The intelligent case closure function automates the verification of event handling results, eliminating the need for manual comparison of photos one by one. This greatly reduces the review burden on staff, allowing them to devote more energy to handling complex issues or preventative work, thereby improving overall work efficiency.

[0032] Precise identification and high accuracy: Cross-modal understanding: The Qwen-VL large model has cross-modal understanding capabilities, which can integrate image features and semantic information during the recognition process. This enhances the recognition accuracy of urban events in complex, blurry, or occluded scenes, reduces false alarms and false negatives, and improves the overall quality of urban event management.

[0033] Model fine-tuning and optimization: By making targeted fine-tuning to the Qwen-VL large model, it is made more adaptable to specific urban environments and event types, further improving the model's recognition accuracy and generalization ability in practical applications. Attached Figure Description

[0034] Figure 1 This is a flowchart of the intelligent event reporting and case closure method based on the CV model provided in this embodiment of the invention.

[0035] Figure 2 This is a detailed flowchart of the intelligent event reporting and case closure method based on the CV model provided in this embodiment of the invention.

[0036] Figure 3 This is a system structure diagram of intelligent event reporting and case closure based on the CV model provided in the embodiments of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0038] like Figure 1 , 2 As shown in the figure, the intelligent event reporting and case closure method based on the CV model provided by this embodiment of the invention includes the following steps: S101, Data Input: Real-time access to urban surveillance video and periodic capture of static images of designated areas; S102, Event Recognition; S103, intelligent reporting; S104, Upload the processing results; S105, before and after comparison analysis; S106, intelligent judgment and automatic case closure.

[0039] Event recognition provided by embodiments of the present invention: The Qwen-VL large model analyzes input video frames or images to identify whether there are management category events within the detection range. During the identification process, the model utilizes its cross-modal understanding capabilities to make judgments by combining image features and semantic information.

[0040] The intelligent reporting provided in this embodiment of the invention: When the model identifies an event, it immediately and automatically assembles an event report containing information such as the event type, location, time, and relevant image evidence, and automatically pushes it to the city operation management platform; the platform then assigns the corresponding staff to handle the event based on the geographical location information.

[0041] Uploading of processing results provided in this embodiment of the invention: After handling the incident on the city operation management platform, staff members upload photos of the scene after the incident.

[0042] Before-and-after comparative analysis provided by the embodiments of the present invention: The Qwen-VL large model receives photo pairs before and after processing, uses deep learning algorithms to identify objects or scenes in the two sets of photos, extracts relevant element features, analyzes and compares them, and identifies changes in the event area.

[0043] The intelligent judgment and automatic case closure provided by the embodiments of the present invention: Based on the comparative analysis results, the model objectively judges whether the event has been effectively handled; if it is determined that the handling is completed, the platform automatically closes the event; otherwise, the staff will conduct secondary handling or submit supplementary relevant information.

[0044] like Figure 3 As shown in the figure, an intelligent event reporting and case closure system based on a CV model provided by an embodiment of the present invention includes: The data input module is used to access urban surveillance video in real time and periodically capture static images of designated areas; The event recognition module is used by the Qwen-VL large model to analyze the input video frames or images and identify whether there are management category events within the detection range. During the recognition process, the model uses its cross-modal understanding capabilities to make judgments by combining image features and semantic information. The intelligent reporting module is used to automatically assemble an event report containing information such as the event type, location, time, and relevant image evidence when the model identifies an event, and automatically push it to the city operation management platform; the platform then assigns the corresponding staff to handle the event based on the geographical location information. The incident handling result upload module is used by staff to upload photos of the scene after handling the incident on the city operation management platform; The before-and-after comparison analysis module is used by the Qwen-VL large model to receive photo pairs before and after processing, and to identify objects or scenes in the two sets of photos through deep learning algorithms, extract relevant element features and perform analysis and comparison to identify changes in the event area. The case closure module is used by the model to objectively determine whether an event has been effectively handled based on the comparative analysis results. If it is determined that the event has been handled, the platform will automatically close the event; otherwise, the staff will conduct secondary processing or submit supplementary relevant information.

[0045] Another object of the present invention is to provide a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the method for intelligent event reporting and case closure based on the CV model.

[0046] Another object of the present invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method for intelligent event reporting and case closure based on the CV model.

[0047] Another objective of this invention is to provide an information data processing terminal for implementing the intelligent event reporting and case closure system based on the CV model.

[0048] Specific implementation of the present invention: Intelligent reporting method This invention employs a Qwen-VL-based large-scale model for training and fine-tuning as the CV large-scale model for this intelligent reporting and intelligent case closure system. This model possesses powerful cross-modal understanding capabilities and can analyze static images. The specific implementation steps are as follows: Data input: Real-time access to urban surveillance video and periodic capture of static images of designated areas.

[0049] Event Recognition: The Qwen-VL large-scale model analyzes input video frames or images to identify whether there are any management-related events (such as exposed garbage, illegally parked vehicles, street vendors, road flooding, road damage, etc.) within the detection range. During the recognition process, the model utilizes its cross-modal understanding capabilities, combining image features and semantic information for accurate judgment.

[0050] Intelligent reporting: When the model identifies an event, it immediately and automatically assembles an event report containing information such as the event type, location, time, and relevant image evidence, and automatically pushes it to the city operation management platform. The platform then assigns appropriate staff to handle the event based on the geographic location information.

[0051] Intelligent case closure method During the incident handling process, this invention further utilizes the Qwen-VL large model to achieve intelligent case closure function, with the specific steps as follows: Upload of handling results: After completing the handling of the incident on the city operation management platform, the staff will upload photos of the scene after the handling.

[0052] Before-and-after comparative analysis: The Qwen-VL large model receives photo pairs before and after processing, uses deep learning algorithms to identify objects or scenes in the two sets of photos, extracts relevant element features and analyzes and compares them to identify changes in the event area.

[0053] Intelligent judgment and automatic case closure: Based on comparative analysis results, the model objectively judges whether the event has been effectively handled. If it is determined that the handling is complete, the platform automatically closes the event; otherwise, staff will conduct secondary handling or submit supplementary information.

[0054] Example 1: Automatic reporting and case closure of incidents involving street vendors occupying urban roads In this embodiment, the system accesses video surveillance streams from urban main roads and captures static images of designated road sections at set time intervals. After the images are input into a visual semantic model, the model identifies street vending events based on the characteristics of stalls, sunshades, shelves, and crowd gatherings in the images, combined with road semantic tags. The system automatically generates an event record containing the event type "street vending," the corresponding latitude and longitude location, the time of occurrence, and the original image, and pushes it to the management platform, which automatically assigns local staff to handle the situation on-site.

[0055] After the staff completes the cleanup, they upload photos of the completed process. The model compares and analyzes the number of stalls, the proportion of obstruction, and the passability of the road in the photos before and after the cleanup. It determines that the obstructing targets have disappeared and the road has been restored to normal, thus automatically determining that the cleanup is complete and executing the case closure operation.

[0056] Example 2: Identification and Closed-Loop Processing of Urban Waste Dumping Incidents The system periodically acquires images from residential area cameras. The model identifies abnormally piled-up objects on the roadside, including plastic bags, construction waste, and scattered debris, and uses semantic tags to determine that these are illegal dumping events. The system automatically generates event information and sends it to the city management platform, which then assigns sanitation workers to clean up the debris.

[0057] Once the cleanup is complete, the uploaded photos of the scene no longer contain any trash, and the road surface background has returned to its original state. The model determines that the event is complete based on the disappearance of the target and the restoration of the scene, and thus automatically closes the case.

[0058] Example 3: Illegal Parking Vehicle Identification and Automatic Transfer The system identifies stationary vehicles on both sides of the road in video surveillance deployed in the city's core business districts, and determines these vehicles as illegally parked by combining semantic information such as lane markings and no-parking signs. The system generates event information including vehicle location, captured image, and timestamp, and uploads it to the platform.

[0059] After the personnel complete the towing or persuasion to move the vehicle, they upload a new image. The model compares and analyzes the images and finds that the illegally parked vehicle has disappeared and the lane is now passable, thus automatically closing the case.

[0060] Example 4: Detection of Abnormal Dust Events During Construction The model identified numerous dust clouds and exposed soil areas from videos surrounding the construction site, and, combined with weather and construction semantic information, determined it to be a dust pollution event. The system reported the event and assigned the construction unit to carry out dust suppression measures by spraying water.

[0061] In the processed and uploaded images, the dust clouds have clearly disappeared, and the ground surface appears moist. Based on this, the model determines that the environmental condition has returned to normal and automatically closes the case.

[0062] Example 5: Handling of Public Facility Damage Incidents The system identifies abnormal structural features such as tilted streetlights and broken guardrails, and combines them with semantic tags of urban facilities to identify them as facility damage events, which are then reported and transferred to the maintenance department.

[0063] After the repair was completed, the uploaded images showed that the facility was now upright and structurally intact. The model comparison analysis confirmed the repair was complete and the case was automatically closed.

[0064] Example 6: Supervision of illegal construction incidents at night The model identifies abnormal lighting intensity and construction equipment activity at the construction site during nighttime hours, and determines these as illegal construction events based on temporal semantics. The system then reports these events and dispatches law enforcement personnel to handle them.

[0065] Once the enforcement action is completed and the uploaded image shows the equipment has stopped operating and the lights have gone out, the model determines the event has ended and the case is automatically closed.

[0066] Example 7: Monitoring of floating objects in river channels The system identifies floating objects such as plastic bottles and foam on the river surface, and determines that it is a river pollution incident based on the semantics of the water surface. It then reports the incident and assigns a cleanup unit to handle it.

[0067] After the cleanup is complete, the floating targets in the image disappear, the water surface texture returns to its natural state, and the model is automatically closed.

[0068] Example 8: Handling of flooding incidents after heavy rain The model detects water reflection features on the road surface and vehicle wading behavior from the video, identifies it as a water accumulation event, and reports it.

[0069] After drainage is completed, the uploaded images show that the road surface is dry and the water accumulation has disappeared, thus completing the case closure judgment.

[0070] Example 9: Identification of Illegal Posting and Hanging of Advertisements The model detected unusual stickers and banners on walls and utility poles, and based on urban management semantics, determined them to be illegal advertisements, and reported the incident.

[0071] After the removal is complete, the stickers and banners in the image will disappear, and the model judgment and processing will be completed.

[0072] Example 10: Multi-Event Concurrent Collaborative Processing Scenario The system simultaneously identifies garbage dumping and illegal parking incidents in the same area, generates two separate events, and distributes them to different departments for simultaneous processing.

[0073] The system independently compares and judges the two types of events before and after, achieving independent closed-loop case closure for multiple events.

[0074] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0075] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent event reporting and case closure based on a computer vision model, characterized in that, Includes the following steps: Step 1, Data Input Steps: Real-time access to city surveillance video and periodic capture of static images of designated areas; Step 2, Event Identification Step: Analyze the static image to identify whether there is a management category event; Step 3, Intelligent Reporting Step: When a management category event is identified, an event report containing event type, location, time of occurrence, and image evidence is automatically generated and uploaded to the city operation management platform; Step 4, Uploading the handling results: After completing the handling of the incident, the staff uploads the on-site images after the handling; Step 5, before-and-after comparison analysis: Compare and analyze the images before and after the treatment to identify changes in the event area; Step 6, Intelligent Judgment and Automatic Case Closure Step: Based on the changes, determine whether the event has been processed and automatically close the case or trigger secondary processing.

2. The method as described in claim 1, characterized in that, In the event recognition step, a visual semantic model with cross-modal understanding capability is used to jointly analyze image features and semantic information to complete event recognition.

3. The method as described in claim 1, characterized in that, The before-and-after comparison analysis steps include extracting features from the target category, spatial location, and semantic state in the image and calculating the changes.

4. The method as described in claim 1, characterized in that, The intelligent judgment step makes a comprehensive judgment based on whether the target in the event area has disappeared, whether the state has changed, and whether the scene has been restored.

5. An intelligent event reporting and case closure system based on a computer vision model, characterized in that, include: The data input module is used to access urban surveillance video and capture still images; The event recognition module is used to analyze static images to identify events of a management category; The intelligent reporting module is used to generate event reports and upload them to the city operation management platform; The disposal result upload module is used to receive on-site images after disposal. The before-and-after comparison analysis module is used to analyze the changes in images before and after processing; The case closure module is used to determine whether to close a case or trigger secondary actions based on changes in circumstances.

6. The system as described in claim 5, characterized in that, The event recognition module includes a visual feature extraction unit and a semantic understanding unit.

7. The system as described in claim 5, characterized in that, The before-and-after comparison analysis module includes a target detection unit, a scene recognition unit, and a change determination unit.

8. The system as described in claim 5, characterized in that, The intelligent reporting module includes an event element generation unit and a geographic location matching unit.

9. A computer-readable storage medium having a program stored thereon, the program being executed by a processor to implement the method of any one of claims 1 to 4.

10. The storage medium as claimed in claim 9, characterized in that, The program includes an instruction module for performing event recognition, image comparison analysis, and automatic case closure.

Images