Intelligent elevator management system based on video recognition and control method thereof

Abstract

The invention relates to the technical field of intelligent elevator management systems, in particular to an intelligent elevator management system based on video recognition and a control method of the intelligent elevator management system based on video recognition. A transport layer; the data analysis layer comprises an image preprocessing module, a feature extraction module, a classification algorithm module and a multi-target tracking module, and the image preprocessing module is used for denoising and enhancing the acquired image; the feature extraction module is used for extracting target features from the preprocessed image; the classification algorithm module is used for classifying and distinguishing the extracted features; the multi-target tracking module is used for tracking a plurality of target objects at the same time; and the control module is used for sending a control instruction to the elevator control system according to the judgment result of the data analysis layer, the cost and difficulty of intelligent transformation are greatly reduced by fully utilizing existing video monitoring equipment, and large-scale popularization becomes possible.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent elevator management systems, particularly an intelligent elevator management system based on video recognition and a control method thereof. BACKGROUND

[0002] With the acceleration of urbanization and the popularity of high-rise buildings, elevators have become an indispensable important facility in modern life. However, the traditional elevator management system has many shortcomings in the face of increasingly complex use scenarios and safety needs. In particular, in places such as densely populated commercial complexes and large public buildings, the safety management and operational efficiency of elevators have become increasingly prominent.

[0003] Currently, the mainstream elevator management solutions on the market mainly fall into two categories. The first category is a monitoring system based on simple sensors, such as weight sensing, access card swiping, etc. Although this type of system is simple to operate and has a relatively low cost, it has a single function and cannot cope with complex use scenarios. For example, it is difficult for them to effectively distinguish between people and electric bicycles, leading to a common problem in implementing the policy of separating people and bicycles. The second category is an artificial management system based on video monitoring. This approach, although it can monitor the inside of the elevator in real time, relies entirely on manual judgment, which not only requires a lot of labor but also has a slow response speed and is prone to oversight. In addition, the cost of manual management is high, making it difficult to apply in large-scale scenarios.

[0004] The common problem with these existing technical solutions is that they cannot effectively integrate intelligent recognition technology with elevator control systems. This results in a lack of ability to cope with complex demands such as people and bicycle separation, safety event early warning, peak period scheduling optimization, etc. In particular, when dealing with potential safety hazards such as electric bicycles entering elevators, existing systems either fail to identify them in a timely manner or react too slowly to effectively prevent accidents.

[0005] In addition, existing intelligent retrofitting solutions often require significant changes to elevator hardware, which not only increases the cost of the retrofit but also prolongs the construction period, causing many inconveniences to daily use. For a large number of old elevators already in use, this type of retrofitting is even more difficult to implement.

[0006] In the face of these problems, there is an urgent need for an elevator management system that can be both intelligent and easy to deploy. This system should make full use of existing hardware resources and use advanced video recognition technology and intelligent algorithms to achieve real-time and accurate monitoring and analysis of the inside of the elevator. At the same time, it should also have the ability to respond quickly and take appropriate measures immediately when an abnormal situation is detected, effectively preventing the occurrence of safety accidents. SUMMARY

[0007] The present application is proposed to solve the above technical problems. It ingeniously combines advanced computer vision technology, artificial intelligence algorithms with traditional elevator control systems, providing a comprehensive, intelligent and scalable elevator management solution. The system not only accurately identifies various situations inside the elevator, but also responds quickly according to preset scene patterns, effectively improving the safety and efficiency of the elevator.

[0008] The present application proposes an intelligent elevator management system based on video recognition, comprising:

[0009] A perception layer for acquiring real-time images inside the elevator through a camera installed in the elevator car;

[0010] A transmission layer for transmitting the images to the data analysis layer;

[0011] A data analysis layer including an image preprocessing module, a feature extraction module, a classification algorithm module and a multi-target tracking module, wherein:

[0012] The image preprocessing module is used for denoising and enhancing the acquired images;

[0013] The feature extraction module is used to extract target features from the preprocessed images;

[0014] The classification algorithm module is used to classify and identify the extracted features;

[0015] The multi-target tracking module is used to track multiple target objects simultaneously;

[0016] A control module for sending control instructions to the elevator control system according to the identification results of the data analysis layer.

[0017] Preferably, it further comprises:

[0018] A scene database for storing preset scene patterns;

[0019] The data analysis layer further comprises a scene comparison module for comparing the classification and identification results with the preset scene patterns in the scene database.

[0020] Preferably, the preset scene patterns in the scene database include:

[0021] People and vehicles on the same elevator, people and vehicles separated, fire mode, special event mode.

[0022] Preferably, the feature extraction module uses the YOLO model to extract target features;

[0023] The classification algorithm module uses the ResNet convolutional neural network for classification and identification.

[0024] Preferably, it further comprises:

[0025] A face recognition module for identifying the identity of the person entering the elevator;

[0026] An access control module for controlling the opening or closing of the elevator door according to the face recognition result.

[0027] Preferably, it further comprises:

[0028] A counting module for counting the number of people and vehicles entering the elevator;

[0029] A data analysis module for analyzing the elevator usage efficiency and passenger flow based on the counting result.

[0030] Preferably, it further comprises:

[0031] A fire detection module including a smoke sensing unit, a temperature measuring unit, and a flame recognition unit;

[0032] A fire alarm module for triggering an alarm and sending a fire report to the fire department when a fire is detected.

[0033] Preferably, it further comprises:

[0034] A special event recognition module for identifying abnormal behavior in the elevator;

[0035] An event alarm module for triggering an alarm and notifying the manager when a special event is identified.

[0036] Preferably, it further comprises:

[0037] A human-computer interaction module including a voice broadcast unit and a visual management interface;

[0038] The voice broadcast unit is used to issue a voice reminder when a violation is detected;

[0039] The visual management interface is used to display the elevator operating status and alarm information.

[0040] A control method based on the intelligent elevator management system, comprising the following steps:

[0041] Setting up a detection area and initializing detection parameters;

[0042] Obtaining real-time images inside the elevator car through a camera;

[0043] Preprocessing, feature extraction, and target recognition are performed on the obtained images;

[0044] Comparing the recognition result with a preset scene mode;

[0045] According to the comparison result, it is judged whether there is an abnormal situation;

[0046] If there is an abnormal situation, corresponding control instructions and alarm measures are triggered;

[0047] If there is no abnormal situation, the elevator is allowed to operate normally;

[0048] The above steps are repeatedly executed to continuously monitor the elevator operating state.

[0049] The beneficial effects of the present application mainly manifest in the following aspects:

[0050] The core advantage of the present application lies in its modular system architecture and intelligent algorithm design. The seamless cooperation between the perception layer, transmission layer, data analysis layer and control module enables the system to realize intelligent management of the whole process from image acquisition, data processing to decision execution. In particular, in the data analysis layer, by integrating multiple functional modules such as image preprocessing, feature extraction, classification algorithm and multi-target tracking, the system can efficiently and accurately identify various situations in complex scenarios.

[0051] This design not only solves the problems of slow reaction and limited recognition ability of traditional systems, but also ingeniously balances the contradiction between system performance and deployment difficulty. By making full use of existing video monitoring equipment, the present system greatly reduces the cost and difficulty of intelligent transformation, making large-scale promotion possible.

[0052] From a macro perspective, the present application provides a feasible technical path for the intelligent upgrading of urban public facilities. It not only improves the safety and operating efficiency of elevators, but also provides valuable data analysis tools for managers, which helps to optimize resource allocation and improve service quality. At the micro level, the various modules of the system form an organic synergistic effect. For example, the cooperation of the multi-target tracking module and the classification algorithm module enables the system to continuously and accurately identify and track multiple targets in complex environments, providing reliable technical support for functions such as separating people and vehicles.

[0053] In addition, the scalability of the present application is also worth noting. Through the pre-set scene database and flexible algorithm framework, the system can easily adapt to new management needs and identification tasks. This design not only prolongs the life cycle of the system, but also provides convenience for future functional upgrade and optimization.

[0054] In summary, the present application successfully solves many technical problems faced by traditional elevator management systems by integrating advanced video recognition technology and intelligent algorithms. It not only provides an efficient and reliable elevator intelligent management scheme, but also provides new ideas and possibilities for the intelligent upgrading of urban public facilities. This innovation is expected to significantly improve the safety level and service quality of public places, and will contribute important strength to the construction of smart cities. Attached Figure Description

[0055] Figure 1 This is the overall system logic block diagram of the present invention.

[0056] Figure 2 This is a logical block diagram of the data analysis layer of the present invention.

[0057] Figure 3 This is a logic block diagram of the face recognition and access control system of the present invention.

[0058] Figure 4 This is a logic block diagram of the counting statistics and data analysis of the present invention.

[0059] Figure 5 This is a logic block diagram of the fire detection function of the present invention.

[0060] Figure 6 This is a logic block diagram for the special event recognition of the present invention.

[0061] Figure 7 This is a logic block diagram of the human-computer interaction logic of the present invention.

[0062] Figure 8 This is a flowchart of the control method of the present invention.

[0063] Figure 9 This is the overall architecture of the intelligent system for non-pedestrian elevators of the present invention.

[0064] Figure 10 The present invention provides a flow chart for the human and vehicle recognition algorithm. Detailed Implementation

[0065] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0067] Example 1

[0068] See Figures 1-10 This invention discloses an intelligent elevator management system based on video recognition and its control method. First, we will describe the structure and function of the system in detail, and then explain its control method.

[0069] The intelligent elevator management system includes a perception layer 1, a transmission layer 2, a data analysis layer 3, and a control module 4. The perception layer 1 is mainly composed of high-definition cameras installed in the elevator car, which are used to capture image information inside the elevator in real time. Preferably, the resolution of these cameras is not less than 2 million pixels to ensure that the image quality meets the subsequent analysis requirements. The transmission layer 2 is responsible for safely and efficiently transmitting the collected image data to the data analysis layer 3. In one embodiment of the present application, the transmission layer 2 can use the existing elevator monitoring network architecture without additional construction, thereby greatly reducing the system deployment cost.

[0070] The data analysis layer 3 is the core of the system, which includes multiple functional modules, namely image preprocessing module 31, feature extraction module 32, classification algorithm module 33, and multi-target tracking module 34. First, the image preprocessing module 31 performs denoising and enhancement processing on the received raw images. This step is crucial because it can significantly improve the accuracy of subsequent analysis. For example, image enhancement can improve the visibility of the target in low light conditions, while denoising can eliminate image noise caused by camera jitter or electromagnetic interference.

[0071] Secondly, the feature extraction module 32 extracts key features from the preprocessed images. In the preferred embodiment of the present application, we use the YOLO (You Only Look Once) model for feature extraction. The advantage of the YOLO model is that it can quickly and accurately identify multiple targets in an image, which is very suitable for elevator scenarios that require real-time response. For example, it can simultaneously identify personnel and electric bicycles in the elevator, providing a basis for subsequent classification judgment.

[0072] Thirdly, the classification algorithm module 33 uses the extracted features for classification and discrimination. The present application uses a Res Net (Residual Network) convolutional neural network for classification. The deep structure of ResNet enables it to learn more complex features, thereby improving classification accuracy. In the context of elevator management, it can accurately distinguish between normal passengers, passengers carrying large items, electric bicycles, and other different categories.

[0073] Finally, the multi-target tracking module 34 is responsible for simultaneously tracking multiple target objects. The role of this module is to ensure that the system can continuously monitor all objects in the elevator, even if they are moving or partially obscured. For example, when multiple passengers and an electric bicycle enter the elevator at the same time, the system can continuously track the position and state of each object.

[0074] The control module 4 sends corresponding control instructions to the elevator control system according to the discrimination results of the data analysis layer 3. For example, if a person and an electric bicycle are detected entering the elevator at the same time, the control module 4 may send an instruction to prevent the elevator door from closing and trigger a voice warning. The specific steps of person-bicycle recognition are as follows: set the detection area, initialize the detection parameters, define the elevator area and initial parameters that the system needs to focus on; capture a single image frame from the video stream. Identify people and vehicles in the image using AI algorithms; eliminate results with low recognition accuracy to improve system reliability; filter out targets outside the detection area and ignore recognition results outside the set area to focus on the inside of the elevator; determine whether the person / vehicle target exists at the same time to determine whether there is a person-bicycle situation; if so, send an elevator door opening instruction: keep the elevator door open to prevent operation; if not, send an elevator closing instruction: allow the elevator to close normally and operate; the system continuously captures single image frames from the video stream, constantly obtaining new images for analysis.

[0075] In one specific embodiment of the present application, the present application further includes a scenario database 5 for storing various preset scenario modes. The scenario comparison module 35 in the data analysis layer 3 compares the classification discrimination results with these preset scenarios. The preset scenarios include but are not limited to the person-bicycle same elevator mode, the person-bicycle separate flow mode, the fire mode, and the special event mode. This design enables the system to flexibly cope with various complex situations. For example, in the person-bicycle separate flow mode, the system can automatically designate some elevators to allow only personnel to ride, while other elevators allow the transportation of electric bicycles.

[0076] In one specific embodiment of the present application, the present application further includes a face recognition module 6 that uses deep learning algorithms such as FaceNet to extract and compare facial features. The access control module 7 controls the access rights of the elevator according to the face recognition results. This function is particularly suitable for places that require strict access control, such as office buildings or high-end residences. For example, the system can be set to allow only registered residents to ride the elevator to a specific floor.

[0077] In one specific embodiment of the present application, the present application further includes a counting and statistical module 8 and a data analysis module 9. These two modules work together not only to statistically analyze elevator usage in real time, but also to conduct in-depth data analysis. For example, the system can identify the peak period of passenger flow every day and optimize the elevator dispatching strategy accordingly to improve operational efficiency. In large shopping malls or office buildings, this function can significantly improve user experience and reduce waiting time.

[0078] In one specific embodiment of the present application, the present application further comprises a fire detection module 10, which includes a smoke sensing unit 101, a temperature measuring unit 102, and a flame recognition unit 103. These three units work together to greatly improve the accuracy of fire detection and reduce the likelihood of false alarms. For example, the system can set the temperature threshold to 60°C, and when it detects that the temperature exceeds this threshold and at the same time detects smoke or flame, it will trigger a fire alarm. The special event recognition module 11 can identify abnormal behaviors such as fighting, damaging equipment, etc. Once these events are detected, the event alarm module 12 will immediately notify the management personnel so that the problem can be handled in a timely manner.

[0079] In one specific embodiment of the present application, the present application further comprises a human-computer interaction module 13, which includes a voice broadcast unit 131 and a visual management interface 132. The voice broadcast unit 131 issues clear voice reminders when it detects violations, such as "Please do not bring electric vehicles into the elevator." The visual management interface 132 provides a visual system monitoring and operation platform for management personnel. This design not only effectively regulates passenger behavior, but also improves management efficiency.

[0080] Finally, the present application also discloses a control method based on the system. The method includes the following steps: setting the detection area and initializing the detection parameters; obtaining real-time images inside the elevator car through the camera; pre-processing, feature extraction, and target recognition on the obtained images; comparing the recognition results with the preset scene patterns; determining whether there is an abnormal situation according to the comparison result; if there is an abnormal situation, triggering the corresponding control instructions and alarm measures; if there is no abnormal situation, allowing the elevator to operate normally; and repeatedly executing the above steps to continuously monitor the elevator operating state. First, set the detection area and initialize the parameters, then enter a continuous monitoring loop. In each loop, the system obtains images, analyzes them, compares scenes, judges abnormalities, and takes appropriate measures. This method ensures that the system can respond to various situations inside the elevator in real time, providing comprehensive safety protection and management support.

[0081] One key innovation of the present application is that it can significantly improve the management level of the elevator through intelligent algorithms without significantly modifying the existing elevator hardware. For example, when renovating old elevators, only high-definition cameras and the present system need to be installed to achieve advanced functions such as separating people and vehicles, safety monitoring, etc., greatly saving the renovation cost.

[0082] In addition, the modular design of the present system makes it highly expandable. As new AI algorithms continue to develop, the system can be easily upgraded to meet new demands. For example, in the future, if a luggage recognition function needs to be added, only the corresponding training model needs to be added in the classification algorithm module.

[0083] To verify the superiority of the invention, we conducted a three-month field test on the elevator group of a large commercial complex. The commercial complex has a total of 20 elevators, with an average daily passenger flow of about 50,000 people, which is a typical high-density, multi-scene test environment. We deployed the system on 10 of the elevators, and the other 10 as a control group, maintaining the original traditional management method.

[0084] The test focuses on the following core indicators: recognition accuracy, response time, safety incident rate, elevator operation efficiency, and user satisfaction. These indicators directly reflect the core innovation points and practical application value of the invention.

[0085] First, we tested the recognition accuracy of the system. Through manual annotation, we recorded the system's recognition of people, electric bicycles, and other large items. The results show that the system can still maintain a high recognition accuracy in complex real-world scenarios. In particular, in terms of separating people and bicycles, the system's accuracy reached 98.7%, much higher than the traditional manual management of about 85%. This result fully proves the reliability of the invention in practical application.

[0086] Second, we focused on the response time of the system. When detecting abnormal situations, the system can respond within an average of 0.5 seconds, including issuing warnings and controlling elevators. In contrast, the average response time of manual management is about 5 seconds. This fast response capability plays a key role in preventing safety accidents.

[0087] Third, we compared the safety incident rate. During the test period, the elevator group with the system did not have any electric bicycle-related safety accidents, while the control group had 3 minor accidents. This data powerfully proves the significant effect of the system in improving elevator safety.

[0088] In addition, we also tested the operation efficiency of the elevators. Through intelligent scheduling and flow analysis, the system successfully shortened the average waiting time from the original 45 seconds to 30 seconds, an increase of 33.3%. This not only improves user experience, but also increases the carrying efficiency of the elevators.

[0089] Finally, we conducted a user satisfaction survey. Among the randomly selected 1000 users, 89% of the users expressed "very satisfied" or "satisfied" with the new system, much higher than the 65% satisfaction rate of the traditional management method.

[0090] The following is a detailed data table of the test results:

[0091] Test index System of the present application Traditional management mode Promotion range Identification accuracy 98.7％ 85％ 16.1％ Average response time 0.5 seconds 5 seconds 90％ Safety event occurrence rate 0 3 per 3 months 100％ Average waiting time 30 seconds 45 seconds 33.3％ User satisfaction 89％ 65％ 36.9％

[0092] These test results fully demonstrate the superiority of the present application. First, the recognition accuracy of up to 98.7% indicates that the system can maintain stable and reliable performance in complex real-world environments. This not only ensures the effective implementation of people-car separation, but also provides a solid foundation for other intelligent management functions.

[0093] Second, the average response time of 0.5 seconds demonstrates the efficiency of the system. In the context of elevators, which require fast responses, this near-real-time response capability is crucial for preventing accidents and improving safety. For example, in the case of detecting a person-car together, the system can issue a warning and prevent operation before the elevator door is fully closed, effectively avoiding potential dangers.

[0094] Third, the significant reduction in the rate of safety incidents directly proves the outstanding performance of the system in improving elevator safety. This not only reduces the losses caused by accidents, but also greatly reduces the legal risks of the management side.

[0095] In addition, the reduction in average waiting time reflects the system's ability to optimize elevator scheduling. By intelligently analyzing passenger flow patterns and adjusting operation strategies in real time, the system successfully improves the carrying efficiency of elevators. This is particularly evident during peak hours, greatly alleviating congestion and improving user experience.

[0096] Finally, the higher user satisfaction not only verifies the actual effect of the system, but also indicates that users have a positive attitude towards this intelligent management method. This lays a good foundation for the widespread promotion of the system.

[0097] In summary, these test results comprehensively and specifically demonstrate the superiority of the present application in practical application. It not only exhibits outstanding performance in the technical aspect, but more importantly, it successfully solves key problems in actual operation, such as safety management, operation efficiency, and user experience. These achievements fully prove that the intelligent elevator management system provided by the present application not only has advanced technology, but also has high practical value and promotion potential.

[0098] In summary, the present application provides a comprehensive, intelligent, and scalable elevator management solution. It not only improves the safety and efficiency of elevator use, but also provides valuable data analysis for managers, which is of great significance for improving the intelligent level of urban public facilities.

[0099] It should be noted that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A video recognition-based intelligent elevator management system, characterized in that, include: The perception layer is used to acquire real-time images of the elevator interior via cameras installed inside the elevator car. The transport layer is used to transmit the image to the data analysis layer; The data analysis layer includes an image preprocessing module, a feature extraction module, a classification algorithm module, and a multi-object tracking module, among which: The image preprocessing module is used to perform noise reduction and enhancement processing on the acquired image; The feature extraction module is used to extract target features from the preprocessed image; The classification algorithm module is used to classify and distinguish the extracted features; The multi-target tracking module is used to track multiple target objects simultaneously; The control module is used to send control commands to the elevator control system based on the judgment results of the data analysis layer.

2. The intelligent elevator management system according to claim 1, characterized in that, Also includes: A scene database is used to store preset scene patterns; The data analysis layer also includes a scene comparison module, which is used to compare the classification results with preset scene patterns in the scene database.

3. The intelligent elevator management system according to claim 2, characterized in that, The preset scene modes in the scene database include: Pedestrian and vehicle co-elevator mode, pedestrian and vehicle separation mode, fire mode, and special event mode.

4. The intelligent elevator management system according to claim 1, characterized in that, The feature extraction module uses the YOLO model to extract target features; The classification algorithm module uses a ResNet convolutional neural network for classification and discrimination.

5. The intelligent elevator management system according to claim 1, characterized in that, Also includes: The facial recognition module is used to identify the people entering the elevator; The access control module is used to control the opening or closing of elevator doors based on facial recognition results.

6. The intelligent elevator management system according to claim 1, characterized in that, Also includes: The counting and statistics module is used to count the number of people and vehicles entering the elevator; The data analysis module is used to analyze elevator usage efficiency and passenger flow based on the counting results.

7. The intelligent elevator management system according to claim 1, characterized in that, Also includes: The fire detection module includes a smoke sensing unit, a temperature measuring unit, and a flame recognition unit; The fire alarm module is used to trigger an alarm and send a fire briefing to the fire department when a fire is detected.

8. The intelligent elevator management system according to claim 1, characterized in that, Also includes: The special event recognition module is used to identify abnormal behavior inside the elevator; The event alarm module is used to trigger alarms and notify administrators when special events are detected.

9. The intelligent elevator management system according to claim 1, characterized in that, Also includes: The human-computer interaction module includes a voice broadcast unit and a visual management interface; The voice broadcasting unit is used to issue a voice reminder when a violation is detected; The visual management interface is used to display the elevator's operating status and alarm information.

10. A control method based on the intelligent elevator management system according to any one of claims 1-9, characterized in that, Includes the following steps: Set the detection area and initialize the detection parameters; Real-time images of the elevator car are captured using a camera. The acquired images undergo preprocessing, feature extraction, and target recognition. The recognition results are compared with preset scene patterns; Determine whether there are any anomalies based on the comparison results; If an abnormal situation occurs, the corresponding control commands and alarm measures will be triggered; If there are no abnormalities, the elevator is allowed to operate normally; Repeat the above steps to continuously monitor the elevator's operating status.

Images