Elevator floor display abnormality identification method, device and computer-readable storage medium

Abstract

The present invention relates to a method, device and computer-readable storage medium for identifying abnormalities in elevator floor displays. When an abnormality in operating speed and / or acceleration is detected, an image detection sample is obtained, and a floor display detection model performs target detection on the image detection sample to determine whether the elevator has an operating abnormality. When the elevator is judged to be operating abnormally, the intelligent interactive device of the elevator is controlled to play a soothing video and / or display a circular information in the elevator, and continuously send the elevator abnormality information to maintenance personnel. The present invention adopts Internet of Things technology, artificial intelligence and big data technology, and realizes 24-hour online monitoring of the elevator through a camera installed in the car. When the elevator floor display is judged to be abnormal, the elevator abnormality information is continuously sent to maintenance personnel. When someone is detected in the car, a pre-set voice prompt is further used to comfort the passengers or play a soothing video to reduce the passengers' anxiety and thus reduce the occurrence of secondary accidents.

Description

Technical Field

[0001] The present invention relates to the field of elevator monitoring, and in particular to a method and device for identifying abnormalities in elevator floor displays, and a computer-readable storage medium. Background Art

[0002] With the gradual implementation of urbanization, high-rise buildings have sprung up in cities. Urban residents rely on elevators for daily commuting, bringing great convenience to people. During an elevator ride, the floor display normally displays numbers and an arrow indicating the current floor, indicating the current floor. However, when the elevator control board malfunctions, this information is lost and the floor display may display abnormalities such as "--" or "E." This can easily cause panic among passengers in the confined cabin, leading to emotional fluctuations and, in severe cases, aggressive behavior in the elevator, resulting in secondary accidents.

[0003] In this context, there is an urgent need for a system that proactively monitors elevator floor display anomalies in real time. When an abnormality occurs, a prompt alert can be issued, the elevator can be located, and corresponding video images can be used to support the problem, allowing for timely resolution of potential elevator failures. Furthermore, when passengers are present in the elevator, voice reassurance can be provided to prevent more serious accidents, ensuring a safe and comfortable ride for passengers and increasing the warmth of the city. Summary of the Invention

[0004] The purpose of the present invention is to provide an elevator floor display abnormality recognition method that can provide voice comfort to passengers when there are people in the car to avoid causing more serious accidents.

[0005] To achieve the above-mentioned object, the present invention provides a method for identifying abnormalities in elevator floor display, comprising:

[0006] S10. When the running speed and / or the acceleration are detected to be abnormal, obtaining an image detection sample;

[0007] S20. Perform target detection on the image detection sample through the layer display detection model to determine whether the elevator is operating abnormally;

[0008] S30. When it is determined that the elevator is operating abnormally, the intelligent interactive device of the elevator is controlled to play a soothing video and / or display a circular message in the elevator, and continuously send elevator abnormality information to maintenance personnel.

[0009] According to one aspect of the present invention, obtaining an image detection sample specifically includes:

[0010] S21. When the running speed and / or the acceleration is detected to be abnormal, the video information of the layer display screen corresponding to the time period is collected;

[0011] S22. Perform frame capture processing on the video information to obtain image detection samples.

[0012] According to one aspect of the present invention, in step S30, after performing frame capture processing on the video, the method further includes:

[0013] S23 performs illumination correction processing on each frame of the image to obtain a corrected output image;

[0014] The illumination correction process adopts an adaptive method.

[0015]

[0016] In the above formula, I(x,y) is the illumination component extracted from the input image F(x,y), O(x,y) is the brightness value of the corrected output image, m is the mean brightness of the illumination component, and k and r are custom coefficients for correcting γ. k and r are iteratively learned using machine learning methods based on the actual lighting conditions in the elevator car.

[0017] According to one aspect of the present invention, the layer display detection model is trained through an offline model, specifically including:

[0018] S41. Obtaining image training samples by performing data enhancement and feature transformation on the image detection samples;

[0019] S42. Create a target detection dataset and use labelImg to mark images containing layer abnormalities;

[0020] S43. Use target detection algorithm to train the model;

[0021] S44. By setting a stopping strategy, training stops when the specified number of iterations is reached or the accuracy of the model on the validation set reaches the set value.

[0022] According to one aspect of the present invention, in step S41, data enhancement is performed on the image detection sample, and the data enhancement method includes at least replication, mosaic enhancement, and inversion. The corrected output image is an 8-bit grayscale image, and the inversion transformation formula is:

[0023] s=255-1-R,R∈(0,1)

[0024] In the above formula, s represents the image after negative transformation, and R represents the original image.

[0025] According to one aspect of the present invention, in step S41, the feature transformation uses ZCA whitening to remove correlation and data redundancy from the image, including:

[0026] Step S411. Calculate the data set {x (1) ,x(2) ,…,x (m) The covariance matrix of

[0027] Step S412. Perform SVD decomposition on the covariance matrix to obtain the eigenvector U. T Get the result x after the data set is rotated rot =U T x;

[0028] Step S413. Perform PCA whitening so that the input features have unit variance, that is,

[0029] where λ i is x rot The value of the diagonal element of the covariance matrix; finally multiply it by the eigenvector on the left to obtain the ZCA whitened data set x ZcAWhite =Ux PCAWhite .

[0030] According to one aspect of the present invention, determining whether the elevator has an operational abnormality includes:

[0031] S71. Preset confidence threshold of confidence and non-maximum suppression threshold of nms;

[0032] S72. When the confidence level is greater than the confidence threshold and the non-maximum suppression is greater than the non-maximum suppression threshold, it is confirmed that there is an abnormality in the elevator corresponding to the frame image;

[0033] S73. When there are at least three corresponding frames of abnormal images in the video information, it is determined that the elevator operation is abnormal.

[0034] Another object of the present invention is to provide an elevator floor display abnormality identification device, comprising:

[0035] A speed monitoring module, used to detect the running speed of the elevator and / or the acceleration of the elevator;

[0036] Video acquisition module, used to capture images of elevator floor display devices;

[0037] Model training module, used to pre-train the layer-level detection model for loading and calling by the detection and recognition module;

[0038] The detection and identification module is used to load the corresponding floor display detection model to determine whether there is any abnormal operation of the elevator;

[0039] The control module and the communication module are used to control the intelligent interactive device of the elevator to play a soothing video and / or display the circular information in the elevator when it is determined that the elevator is operating abnormally, and continuously send the elevator abnormality information to the maintenance personnel.

[0040] According to one aspect of the present invention, the elevator floor display abnormality identification device further includes:

[0041] The image correction module is used to perform adaptive illumination correction on the image of the layer display device.

[0042] Another object of the present invention is a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned elevator floor display abnormality identification methods.

[0043] The present invention provides an elevator floor display abnormality recognition method, device and computer-readable storage medium. When an abnormality in running speed and / or acceleration is detected, an image detection sample is obtained, and a floor display detection model performs target detection on the image detection sample to determine whether the elevator has an operating abnormality. When the elevator is judged to be operating abnormally, the intelligent interactive device of the elevator is controlled to play a soothing video and / or display a circular information in the elevator, and the elevator abnormality information is continuously sent to maintenance personnel. The Internet of Things technology, artificial intelligence and big data technology are adopted, and 24-hour online monitoring of the elevator is achieved through a camera installed in the car. The floor display abnormality in the car is recognized through a computer vision intelligent target detection algorithm, and the corresponding video image information is intercepted. When the elevator floor display is determined to be abnormal, the elevator operation is confirmed to be abnormal, and the elevator abnormality information is continuously sent to maintenance personnel. When someone is detected in the car, a pre-set voice prompt is further used to comfort the passengers or play a soothing video to reduce the passengers' anxiety, thereby reducing the occurrence of secondary accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A block diagram schematically illustrates the steps of a method for identifying abnormalities in elevator floor display according to one embodiment of the present invention;

[0045] Figure 2 A flowchart schematically illustrates a method for identifying abnormalities in elevator floor display according to an embodiment of the present invention;

[0046] Figure 3 A flowchart schematically illustrates a method for identifying abnormalities in elevator floor display according to an embodiment of the present invention;

[0047] Figure 4 Schematically showing a flow chart of abnormality detection of elevator floor display according to one embodiment of the present invention;

[0048] Figure 5The figure schematically shows an elevator car diagram acquired by a video acquisition unit according to one embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The embodiments cannot be described one by one here, but the embodiments of the present invention are not limited to the following embodiments.

[0050] Combine Figure 1 and Figure 5 As shown, according to one embodiment of the present invention, a method for identifying abnormalities in elevator floor display of the present invention includes:

[0051] S10. When an abnormal running speed and / or acceleration is detected, an image detection sample is obtained;

[0052] S20. Perform target detection on the image detection sample using the layer display detection model to determine whether the elevator is operating abnormally;

[0053] S30. When it is determined that the elevator is operating abnormally, the intelligent interactive device that controls the elevator plays a soothing video and / or displays a circular message inside the elevator, and continuously sends elevator abnormality information to maintenance personnel.

[0054] In this embodiment, when an abnormality in running speed and / or acceleration is detected, an image detection sample is obtained, and a layer display detection model performs target detection on the image detection sample to determine whether the elevator has an abnormal operation. When the elevator is determined to be abnormal, the intelligent interactive device of the elevator is controlled to play a soothing video and / or display the circular information in the elevator, and the elevator abnormality information is continuously sent to the maintenance personnel. The Internet of Things technology, artificial intelligence and big data technology are adopted, and 24-hour online monitoring of the elevator is achieved through the camera installed in the car. The abnormality of the layer display in the car is identified through the computer vision intelligent target detection algorithm, and the corresponding video image information is intercepted. When the elevator layer display is determined to be abnormal, the elevator operation is confirmed to be abnormal, and the elevator abnormality information is continuously sent to the maintenance personnel. When someone is detected in the car, the pre-set voice prompts and soothes or plays a soothing video to reduce the anxiety of the passengers, thereby reducing the occurrence of secondary accidents.

[0055] Specifically, if Figure 1 and Figure 4As shown, when the elevator speed and / or acceleration is abnormal, it is inferred that the elevator operation status is abnormal. Therefore, by obtaining image detection samples, mainly shooting the elevator's floor display screen, and then calling the floor display detection model to perform target detection on the image detection samples, the elevator operation status is judged. When it is judged that the elevator operation is abnormal, first, the elevator abnormality information is continuously sent to the maintenance personnel, and secondly, the presence of passengers in the elevator is detected. When it is detected that there are passengers in the elevator car, a soothing video is played through the intelligent interactive device set in the car, and the soothing voice configured in advance is played at the same time.

[0056] It is worth noting that the above process can be implemented using Internet of Things technology. In the case of network anomalies, it can also detect whether there are any abnormalities in the elevator, avoiding the cloud platform's delayed processing of elevator operation abnormalities when the network is abnormal. It can truly realize real-time monitoring of the elevator's operating status, effectively improve the stability and safety of elevator operation, and enhance user experience.

[0057] For example, the analysis and judgment of acceleration and speed in this example is primarily based on signal statistical processing technology. This means that if the elevator is experiencing severe vibration or operating at an abnormally low speed, the suspected abnormal operation period is recalled. Severe vibration is determined by the speed suddenly dropping to zero within a short period of time, which in this case is 30 milliseconds. Abnormally low speed occurs when the speed is lower than the average speed of the previous 15 days.

[0058] In addition, after processing a frame by frame in an analysis video, if it is determined that there is a floor display abnormality in the elevator video, an alarm will be issued, and the corresponding video will be used as an alarm certificate and basis for subsequent handling.

[0059] like Figure 2 As shown, in one embodiment of the present invention, preferably, obtaining an image detection sample specifically includes:

[0060] S21. When an abnormal speed and / or acceleration is detected, the video information of the layer display screen corresponding to the time period is collected;

[0061] S22. Perform frame capture processing on the video information to obtain image detection samples.

[0062] In this embodiment, by monitoring the running speed and / or acceleration of the elevator in real time, the running status of the elevator can be timely understood, and the video information can be frame-cut to obtain image detection samples.

[0063] Specifically, IoT technologies are used to collect real-time sensor data for elevator monitoring. The sensor here primarily refers to a camera installed in the innermost, upper left corner of the elevator car, facing the elevator door. This camera provides 24-hour monitoring of the entire car, including the elevator keypad and floor display. The camera collects data at a frequency of 25 frames per second (fps).

[0064] In one embodiment of the present invention, preferably, in step S30, after performing frame capture processing on the video, the method further includes:

[0065] S23 performs illumination correction processing on each frame of the image to obtain a corrected output image;

[0066] The illumination correction process uses an adaptive method.

[0067]

[0068] In the above formula, I(x,y) is the illumination component extracted from the input image F(x,y), O(x,y) is the brightness value of the corrected output image, m is the mean brightness of the illumination component, and k and r are custom coefficients for correcting γ. k and r are iteratively learned using machine learning methods based on the actual lighting conditions in the elevator car.

[0069] In this embodiment, due to the influence of the elevator door opening and closing, the electronic advertising screen in the elevator, and the material of the floor display screen, low-illuminance images or high-illuminance images illuminated by strong lights are prone to appear at the floor display in the elevator. In particular, due to the influence of the light from the electronic advertising screen, adaptive illumination correction is performed on each frame of the image to improve the uniformity of the illumination and enhance the image quality, thereby ensuring recognition accuracy and improving stability and reliability.

[0070] like Figure 3 As shown, in one embodiment of the present invention, preferably, the layer display detection model is trained through an offline model, specifically including:

[0071] S41. Obtain image training samples by performing data enhancement and feature transformation on the image detection samples;

[0072] S42. Create a target detection dataset and use labelImg to mark images containing layer abnormalities;

[0073] S43. Use target detection algorithm to train the model;

[0074] S44. By setting a stopping strategy, training stops when the specified number of iterations is reached or the accuracy of the model on the validation set reaches the set value.

[0075] In this embodiment, first, data enhancement and feature transformation are performed on the above-mentioned image detection samples. Secondly, a target detection dataset is prepared, and labelImg is used to mark pictures containing layer-visible abnormalities, such as "-", "E", etc. that appear in the layer-visible area. Specifically, the abnormal layer-visible area is framed and the category is marked, and all incomplete contrast imaging is marked, and the model is trained using the target detection algorithm. Such as the YOLO series, Efficientdet, SSD algorithm, etc. A target detection model based on YOLOv4 can be built, and the transfer learning method is used to use the model trained on the coco dataset as the initial weight of the network backbone and randomize other parameters, and train on the above-mentioned labeled dataset. The loss function includes classification loss and regression function loss of the layer-visible abnormality boundary box position.

[0076] The backbone of the network is CSPDarknet53, the initial weights are the pre-trained weights of YOLOv4, the neck uses SPP, and the head uses the three output layers of YOLOv3.

[0077] In one embodiment of the present invention, preferably, in step S41, data enhancement is performed on the image detection sample, and the data enhancement method at least includes copying, mosaic enhancement, inversion, correcting the output image to an 8-bit grayscale image, and the inversion transformation formula,

[0078] s=255-1-R,R∈(0,1)

[0079] In the above formula, s represents the image after negative transformation, and R represents the original image.

[0080] For example, this example uses the YOLOv4 object detection algorithm. First, the image input size is set to 416*416 to reduce video memory usage. Secondly, during the transfer learning training phase, the first 50 layers of the backbone network are frozen, and only fine-tuning is performed to further reduce video memory usage. Other non-parameters include: batch_size is set to 8 and the learning rate is set to 1e-3 in the frozen layers, and to 4 and 1e-4 in the unfrozen layers.

[0081] In one embodiment of the present invention, preferably, in step S41, the feature transformation uses ZCA whitening to remove correlation and data redundancy from the image, including:

[0082] Step S411. Calculate the data set {x (1) ,x (2) ,…,x (m) The covariance matrix of

[0083] Step S412. Perform SVD decomposition on the covariance matrix to obtain the eigenvector U. T Get the result x after the data set is rotated rot =U T x;

[0084] Step S413. Perform PCA whitening so that the input features have unit variance, that is,

[0085] where λ i is x rot The value of the diagonal element of the covariance matrix; finally multiply it by the eigenvector on the left to obtain the ZCA whitened data set x ZcAWhite =Ux PCAWhite .

[0086] In one embodiment of the present invention, preferably, determining whether the elevator has an operational abnormality includes:

[0087] S71. Preset confidence threshold of confidence and non-maximum suppression threshold of nms;

[0088] S72. When the confidence level is greater than the confidence threshold and the non-maximum suppression is greater than the non-maximum suppression threshold, it is confirmed that there is an abnormality in the elevator corresponding to the frame image;

[0089] S73. When there are at least three corresponding frames of abnormal images in the video information, it is determined that the elevator operation is abnormal.

[0090] In this embodiment, by pre-setting the confidence threshold of confidence and the non-maximum suppression threshold of non-maximum suppression nms, when detecting each frame of the picture, the confidence and the non-maximum suppression nms are compared. When the confidence is greater than the confidence threshold and the non-maximum suppression is greater than the non-maximum suppression threshold, it is confirmed that there is an abnormality in the elevator in the corresponding frame picture. Then, it is confirmed that the same second of video information contains 3 or 3 corresponding frame pictures with abnormalities, and the thief determines that the elevator operation is abnormal.

[0091] For example, by setting the category confidence confidence = 0.5 and non-maximum suppression nms = 0.3 to judge the results and boxes, that is, when an image is inferred by YOLOv4, if its confidence exceeds 0.5 and its iou exceeds 0.3, it will be pushed out for layer-by-layer abnormality judgment.

[0092] An elevator floor display abnormality recognition device of the present invention comprises:

[0093] A speed monitoring module is used to detect the running speed of the elevator and / or the acceleration of the elevator;

[0094] Video acquisition module, used to capture images of elevator floor display devices;

[0095] Model training module, used to pre-train the layer-level detection model for loading and calling by the detection and recognition module;

[0096] The detection and identification module is used to load the corresponding floor display detection model to determine whether there is any abnormal operation of the elevator;

[0097] The control module and communication module are used to control the elevator's intelligent interactive device to play a soothing video and / or display the circular information inside the elevator when it is determined that the elevator is operating abnormally, and continuously send elevator abnormality information to maintenance personnel.

[0098] In this embodiment, when the speed monitoring module detects abnormal operating speed and / or acceleration, the video acquisition module obtains image detection samples, and the detection and recognition module loads the corresponding floor display detection model to determine whether the elevator has abnormal operation. When it is determined that the elevator is operating abnormally, the control module and the communication module control the elevator's intelligent interactive device to play a soothing video and / or display the circular information in the elevator, and continuously send the elevator abnormality information to the maintenance personnel. The Internet of Things technology, artificial intelligence and big data technology are used to realize 24-hour online monitoring of the elevator through the camera installed in the car. The computer vision intelligent target detection algorithm is used to realize the identification of the abnormal floor display in the car, and the corresponding video image information is intercepted. When it is determined that the elevator floor display is abnormal, the elevator operation is confirmed to be abnormal, and the elevator abnormality information is continuously sent to the maintenance personnel. When someone is detected in the car, the pre-set voice is further used to prompt and comfort or play a soothing video to reduce the passenger's anxiety, thereby reducing the occurrence of secondary accidents.

[0099] In one embodiment of the present invention, preferably, the elevator floor display abnormality identification device further includes:

[0100] The image correction module is used to perform adaptive illumination correction on the image of the layer display device.

[0101] In this embodiment, due to the influence of the elevator door opening and closing, the electronic advertising screen in the elevator, and the material of the floor display screen, low-illuminance images or high-illuminance images illuminated by strong lights are prone to appear at the floor display in the elevator. In particular, due to the influence of the light from the electronic advertising screen, the image correction module performs adaptive illumination correction on each frame of the image to improve the uniformity of the illumination and enhance the image quality, thereby ensuring recognition accuracy and improving stability and reliability.

[0102] A computer-readable storage medium of the present invention stores a computer program, which, when executed by a processor, implements any of the above-mentioned elevator floor display abnormality identification methods.

[0103] The present invention provides an elevator floor display abnormality recognition method, device and computer-readable storage medium. When an abnormality in running speed and / or acceleration is detected, an image detection sample is obtained, and a floor display detection model performs target detection on the image detection sample to determine whether the elevator has an operating abnormality. When the elevator is judged to be operating abnormally, the intelligent interactive device of the elevator is controlled to play a soothing video and / or display a circular information in the elevator, and the elevator abnormality information is continuously sent to maintenance personnel. The Internet of Things technology, artificial intelligence and big data technology are adopted, and 24-hour online monitoring of the elevator is achieved through a camera installed in the car. The floor display abnormality in the car is recognized through a computer vision intelligent target detection algorithm, and the corresponding video image information is intercepted. When the elevator floor display is determined to be abnormal, the elevator operation is confirmed to be abnormal, and the elevator abnormality information is continuously sent to maintenance personnel. When someone is detected in the car, a pre-set voice prompt is further used to comfort the passengers or play a soothing video to reduce the passengers' anxiety, thereby reducing the occurrence of secondary accidents.

[0104] The above contents are merely examples of specific solutions of the present invention. For devices and structures not described in detail, it should be understood that they can be implemented by adopting general devices and methods available in the art.

[0105] The above description is merely one embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for identifying abnormalities in elevator floor displays, characterized in that: include: S10. When an abnormal running speed and / or acceleration is detected, an image detection sample is obtained; S20. Perform target detection on the image detection sample through the layer display detection model to determine whether the elevator is operating abnormally; S30. When the elevator is judged to be operating abnormally, the intelligent interactive device of the elevator is controlled to play a soothing video and / or display a circular message in the elevator, and continuously send an abnormal elevator information to the maintenance personnel; The acquiring of image detection samples specifically includes: S21. When the running speed and / or the acceleration is detected to be abnormal, the video information of the layer display screen corresponding to the time period is collected; S22. The video information is frame-cut and processed to obtain an image detection sample; The step of determining whether the elevator has an operational abnormality includes: S71. Preset confidence threshold of confidence and non-maximum suppression threshold of nms; S72. When the confidence level is greater than the confidence threshold and the non-maximum suppression is greater than the non-maximum suppression threshold, it is confirmed that there is an abnormality in the elevator corresponding to the frame image; S73. When there are at least three corresponding frames of abnormal images in the video information, it is determined that the elevator operation is abnormal.

2. The elevator floor display abnormality recognition method according to claim 1, characterized in that: In step S30, after the video is subjected to frame capture processing, the following steps are further included: S23 performs illumination correction processing on each frame of the image to obtain a corrected output image; The illumination correction process adopts an adaptive method. In the above formula, I(x,y) is the illumination component extracted from the input image F(x,y), O(x,y) is the brightness value of the corrected output image, m is the mean brightness of the illumination component, and k and r are custom coefficients for correcting γ. k and r are iteratively learned using machine learning methods based on the actual lighting conditions in the elevator car.

3. The elevator floor display abnormality identification method according to claim 1, characterized in that: The layer-display detection model is trained through an offline model, specifically including: S41. Obtaining image training samples by performing data enhancement and feature transformation on the image detection samples; S42. Create a target detection dataset and use labelImg to mark images containing layer abnormalities; S43. Use target detection algorithm to train the model; S44. By setting a stopping strategy, training stops when the specified number of iterations is reached or the accuracy of the model on the validation set reaches the set value.

4. The elevator floor display abnormality recognition method according to claim 3 is characterized in that: In step S41, data enhancement is performed on the image detection sample, and the data enhancement method includes at least replication, mosaic enhancement, and inversion. The corrected output image is an 8-bit grayscale image, and the inversion transformation formula is: s=255-1-R,R∈(0,1) In the above formula, s represents the image after negative transformation, and R represents the original image.

5. The elevator floor display abnormality recognition method according to claim 3, characterized in that: In step S41, the feature transformation uses ZCA whitening to remove correlation and data redundancy from the image, including: Step S411. Calculate the data set {x (1) ,x (2) ,…,x (m) The covariance matrix of Step S412. Perform SVD decomposition on the covariance matrix to obtain the eigenvector U. T Get the result x after the data set is rotated rot =U T x; Step S413. Perform PCA whitening so that the input features have unit variance, that is, where λ i is x rot The value of the diagonal element of the covariance matrix; finally multiply it by the eigenvector on the left to obtain the ZCA whitened data set x ZCAWhite =Ux PCAWhite .

6. An elevator floor display abnormality recognition device for implementing the method described in claim 1, characterized in that: include: A speed monitoring module, used to detect the running speed of the elevator and / or the acceleration of the elevator; Video acquisition module, used to capture images of elevator floor display devices; Model training module, used to pre-train the layer-level detection model for loading and calling by the detection and recognition module; The detection and identification module is used to load the corresponding floor display detection model to determine whether there is any abnormal operation of the elevator; The control module and the communication module are used to control the intelligent interactive device of the elevator to play a soothing video and / or display the circular information in the elevator when it is determined that the elevator is operating abnormally, and continuously send the elevator abnormality information to the maintenance personnel.

7. The elevator floor display abnormality recognition device according to claim 6, characterized in that: Also includes: The image correction module is used to perform adaptive illumination correction on the image of the layer display device.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the elevator floor display abnormality identification method according to any one of claims 1 to 5 is implemented.

Images