Elevator car monitoring method, system, and storage medium

Abstract

The application discloses an elevator car monitoring method, system and storage medium, and ingeniously acquires image data in the elevator car, then positions and identifies feature objects in the image data, judges whether to have forbidden entry, personnel or animals other than people into the elevator car according to the identification result, selects to immediately output early warning information or judges whether to output early warning information in combination with further judgment according to different detection conditions, is flexible in implementation, reliable in application, further selects reference points in the car internal environment by feature points of the feature objects, judges whether there is an abnormal condition such as personnel falling, excessive crowding or hostage in the car according to the distance change between the feature points or between the feature points and the reference points, so that the background security personnel can improve the monitoring attention of the elevator car according to the early warning information, and the elevator monitoring is improved in humanization and pertinence.

Description

Technical Field

[0001] This invention relates to elevator car monitoring technology and video surveillance technology, and particularly to elevator car monitoring methods, systems and storage media. Background Technology

[0002] Elevators are common in shopping malls, office buildings, and residential buildings, providing a safe passage between floors. Because the elevator car is relatively enclosed, people cannot leave during the ascent or descent. In the event of an accident, the safety of those inside the elevator car is compromised. Therefore, safety monitoring inside the elevator car is crucial. Some researchers have proposed using video data from inside the elevator car to obtain the posture of people inside and determine if there are falls or unusual close proximity incidents. However, because the video monitoring equipment inside the elevator car captures images at an angle from top to bottom, there are blind spots. For example, in some areas, only certain areas can be captured. When only the upper body or head of a person is visible, it is difficult to infer their posture. In addition, when shooting at an angle, the posture of the person is obscured, especially when there are many people, and the lower bodies of different people may overlap. Therefore, obtaining the posture of people through image data often requires a large elevator car to achieve good results. If the imaging equipment is placed in the lower part of the car, the probability of collision increases and the sensory rejection of users increases. Therefore, how to improve the versatility of elevator car anomaly monitoring and reduce the basic requirements for judging abnormal conditions is a topic of positive practical significance. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide an elevator car monitoring method, system, and storage medium that is fast-responding, reliable in implementation, flexible in use, user-friendly, and provides good results.

[0004] To achieve the above-mentioned technical objectives, the technical solution adopted by this invention is as follows:

[0005] An elevator car monitoring method, comprising:

[0006] S01. Acquire image data inside the elevator car according to preset conditions. The captured image data shall at least cover the elevator car door.

[0007] S02. Acquire image data according to preset conditions, locate and identify feature objects in the image data, and generate recognition results;

[0008] S03. Based on the recognition results, feature points and reference points are selected for the feature objects in the image data, and feature points and reference points corresponding to the feature objects are generated.

[0009] S04. Output early warning information based on the identification results and / or the positional change data between the feature points and the reference points.

[0010] As one possible implementation, further, solution S01 includes:

[0011] S011. Real-time liveness detection inside the elevator car to generate liveness detection data;

[0012] S012. Acquire liveness detection data.

[0013] When the liveness detection data indicates that no living object is detected inside the elevator car, image data of the elevator car is acquired at a first preset acquisition frequency.

[0014] When the liveness sensor data indicates that there is a living object inside the elevator car, image data inside the elevator car is acquired at a second preset acquisition frequency, and elevator operating parameters are also acquired in real time.

[0015] As a preferred implementation option, this solution S02 preferably includes:

[0016] S021. Acquire liveness detection data. When the liveness detection data indicates that there is a living object inside the elevator car, acquire image data at a first preset acquisition frequency; otherwise, acquire image data at a second preset acquisition frequency.

[0017] S022. Extract image frames from the image data, then locate and exclude preset elements in the image frames, then perform overall recognition on the key frames and separate recognition on the remaining elements, generating one or more recognition results pointing to the whole and one or more pointing to the part corresponding to the key frames.

[0018] As a preferred implementation option, this solution S022 preferably includes:

[0019] S0221. Extract image frames from the image data using a keyframe extraction algorithm to obtain several keyframes;

[0020] S0222. Locate the elements in the keyframe, and then exclude the preset elements;

[0021] S0223. Perform overall recognition on the keyframes from which preset elements have been removed to obtain the overall recognition result;

[0022] S0224. Extract the remaining elements from the keyframes after removing the preset elements, and perform recognition processing on them respectively to obtain local recognition results.

[0023] As a preferred implementation option, solution S03 preferably includes:

[0024] S031. Obtain the local recognition results and judge them according to preset conditions.

[0025] When it meets the first preset condition, it is marked with first information and a warning message is generated and output.

[0026] When it meets the second preset condition, mark it with the second information and proceed to S032.

[0027] When it meets the third preset condition, mark it with the third information and proceed to S033;

[0028] S032. Locate the elements marked with second information and third information in multiple keyframes, set them as feature objects, and then judge the constraint relationship between the feature objects corresponding to the second information and the feature objects corresponding to the third information according to preset conditions. When the constraint relationship does not meet the preset requirements, generate warning information and use it for output.

[0029] S033. Select at least one feature point on the feature object in the keyframe. At the same time, select at least one reference point in the upper part of the image acquisition area inside the elevator car, and make each feature point correspond to at least one reference point, wherein the location of the reference point is higher than the top of the feature object.

[0030] As a preferred implementation option, preferably, the second preset condition of this solution is that the local identification result identifies the animal as a living animal that does not include humans; S032 includes:

[0031] S0321. Locate the elements marked with second and third information in multiple keyframes and set them as feature objects;

[0032] S0322. Remove the background from the keyframes, and then perform edge contour recognition on the feature objects marked with the second and third information respectively to generate edge contour data.

[0033] S0323. Obtain edge contour data.

[0034] When the distance between the edge contour data corresponding to the feature object marked with the second information and the edge contour data corresponding to the feature object marked with the third information is less than a preset value, the two are marked as a constraint relationship.

[0035] When the distance between the edge contour data corresponding to the feature object marked with the second information and the edge contour data corresponding to the feature object marked with the third information is greater than a preset value, the region between the two is extracted and the shape is recognized. When the recognition result is that there are stripes, the two are marked as a constraint relationship; otherwise, the relationship between the two is marked as an unconstrained relationship.

[0036] S0324. Statistical analysis of constraint relationships of feature objects marked with second information within a preset time period. When the number of statistical counts of unconstrained relationships exceeds a preset value, generate warning information and output it.

[0037] As a preferred implementation option, preferably, the first information in this solution is an object prohibited from entering the elevator; the third preset condition is that the local identification result identifies a person; S04 includes:

[0038] S041. Obtain the identification result, and when it matches the preset abnormal information and / or is marked as the first information, output the warning information.

[0039] As a preferred implementation option, this solution S04 preferably also includes:

[0040] A041. Real-time recording of the distance between the feature points and reference points on the feature objects identified by the local recognition results as human features, generating the first reference distance;

[0041] A042. Obtain the first reference spacing. When it is greater than the preset first threshold distance, time it to obtain the duration of the feature object being greater than the first preset threshold in the first reference spacing, and obtain the duration of the first abnormal spacing.

[0042] A043. Obtain the duration of the first abnormal interval. When it is greater than the preset first time threshold, generate personnel fall information as early warning information output.

[0043] As a preferred implementation option, this solution S04 preferably also includes:

[0044] B041. The local recognition results are identified as the feature points corresponding to human feature objects, and the spacing between them is recorded and monitored in real time to generate a second reference spacing.

[0045] B042. Obtain the second reference spacing. When it is greater than the preset second threshold distance, time it to obtain the duration of the feature object being greater than the second preset threshold at the second reference spacing, and obtain the duration of the second abnormal spacing.

[0046] B043. Obtain the duration of the second abnormal interval. When it is greater than the preset second time threshold, generate the abnormal interval of personnel as an early warning information output.

[0047] Based on the above, the present invention also provides an elevator car monitoring system, which includes:

[0048] The image acquisition unit is one or more, and is used to acquire image data inside the elevator car according to preset conditions. The captured image data covers at least the elevator car door.

[0049] The server is used to store the image data acquired by the image acquisition unit;

[0050] The data processing unit is used to acquire image data according to preset conditions, locate and identify feature objects in the image data, and generate recognition results.

[0051] The data labeling unit is used to select feature points and reference points for feature objects in image data based on the recognition results, and generate feature points and reference points corresponding to the feature objects;

[0052] An object tracking unit is used to track objects based on recognition results and / or positional change data between feature points and reference points;

[0053] The instruction generation unit is used to output early warning information based on location change data and / or recognition results according to preset conditions.

[0054] Based on the above, the present invention also provides a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the elevator car monitoring method described above.

[0055] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art: The ingenuity of this solution lies in acquiring image data inside the elevator car, then locating and identifying feature objects within the image data, and judging whether there are prohibited areas, personnel, or animals other than humans entering the elevator car based on the recognition results. Thus, depending on different detection situations, it can choose to immediately output warning information or combine further judgment to determine whether to output warning information. This solution is not only flexible in implementation and reliable in application, but also further judges whether there are abnormal situations such as people falling, overcrowding, or being held hostage by identifying feature points of feature objects and selecting reference points for the internal environment of the car, and combining the changes in the distance between feature points or between feature points and reference points. This allows back-end security personnel to increase their monitoring attention to the elevator car based on the warning information, thereby improving the humanization and targeting of elevator monitoring. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is one of the simplified implementation flowcharts of the monitoring method of the present invention;

[0058] Figure 2 This is a schematic diagram illustrating the principle of the monitoring method of the present invention, which uses grayscale data to determine the presence of objects inside the elevator car.

[0059] Figure 3 This is the second simplified implementation flowchart of the monitoring method of the present invention;

[0060] Figure 4 This is a schematic diagram of a person bringing prohibited objects into the elevator car in the monitoring method of the present invention;

[0061] Figure 5 This is a schematic diagram illustrating a person bringing a pet dog into an elevator car in the monitoring method of this invention;

[0062] Figure 6 This is a simplified schematic diagram of the monitoring method of the present invention when there are multiple people in the elevator car;

[0063] Figure 7 In the monitoring method of this invention, for Figure 6 The diagram shows the selection of feature points on the head of a person after removing the background.

[0064] Figure 8 This is a schematic diagram of selecting feature points on the head of a person and selecting reference points on the upper part of the car in the monitoring method of the present invention;

[0065] Figure 9 This is a simplified flowchart of one of the sub-steps included in step S04 of the monitoring method of the present invention.

[0066] Figure 10 This is a simplified flowchart of another sub-step included in step S04 of the monitoring method of the present invention.

[0067] Figure 11 This is a schematic diagram of the unit module connections of the monitoring system of the present invention. Detailed Implementation

[0068] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] like Figure 1 As shown in the figure, this embodiment provides a method for monitoring an elevator car, which includes:

[0070] S01. Acquire image data inside the elevator car according to preset conditions. The captured image data shall at least cover the elevator car door.

[0071] S02. Acquire image data according to preset conditions, locate and identify feature objects in the image data, and generate recognition results;

[0072] S03. Based on the recognition results, feature points and reference points are selected for the feature objects in the image data, and feature points and reference points corresponding to the feature objects are generated.

[0073] S04. Output early warning information based on the identification results and / or the positional change data between the feature points and the reference points.

[0074] In this solution, to better integrate image data and provide more timely feedback on the situation inside the elevator car when there are people inside, one possible implementation method is to combine... Figure 2 As shown, this solution can process the image data into grayscale and then compare the grayscale value of the elevator car when it is empty with the grayscale image frame after grayscale processing to determine whether an object has entered the elevator car based on the difference. However, this method requires image frame extraction and grayscale processing, which consumes significant hardware resources, making it costly and uneconomical for local processing. Alternatively, this solution can use a liveness sensor (commercially available hardware) to directly determine whether a living object has entered the elevator car. In this case, further... Figure 3 As shown, this solution S01 includes:

[0075] S011. Real-time liveness detection inside the elevator car to generate liveness detection data;

[0076] S012. Acquire liveness detection data.

[0077] When the liveness detection data indicates that no living object is detected inside the elevator car, image data of the elevator car is acquired at a first preset acquisition frequency.

[0078] When the liveness sensor data indicates that there is a living object inside the elevator car, image data inside the elevator car is acquired at a second preset acquisition frequency (greater than the first preset acquisition frequency), and elevator operating parameters are also acquired in real time.

[0079] This method allows for sufficient video feedback from inside the elevator during nighttime or other off-peak hours to meet basic backend monitoring requirements. When a living person enters, the acquisition frequency is increased to capture details of fast-moving actions (low-frequency acquisition may result in motion blur or interruptions, and some details may be lost). Acquiring elevator operating parameters allows backend personnel to promptly understand the elevator's operating status and facilitates retrieval of data in case of abnormalities.

[0080] Accordingly, in image processing, as a preferred implementation option, this solution S02 preferably includes:

[0081] S021. Acquire liveness detection data. When the liveness detection data indicates that there is a live object in the elevator car, acquire image data at a first preset acquisition frequency; otherwise, acquire image data at a second preset acquisition frequency. The first preset acquisition frequency and the second preset acquisition frequency are adapted to the first preset acquisition frequency and the second preset acquisition frequency.

[0082] S022. Extract image frames from the image data, then locate and exclude preset elements in the image frames, then perform overall recognition on the key frames and separate recognition on the remaining elements, generating one or more recognition results pointing to the whole and one or more pointing to the part corresponding to the key frames.

[0083] Because a large number of image frames are generated per second during image monitoring, processing each frame would require significant computing power. To simplify the processing and ensure that important and critical parts are not overlooked, this preferred implementation scheme S022 includes:

[0084] S0221. Extract image frames from the image data using a keyframe extraction algorithm to obtain several keyframes;

[0085] S0222. Locate the elements in the keyframe, and then exclude the preset elements;

[0086] S0223. Perform overall recognition on the keyframes from which preset elements have been removed to obtain the overall recognition result;

[0087] S0224. Extract the remaining elements from the keyframes after removing the preset elements, and perform recognition processing on them respectively to obtain local recognition results.

[0088] The keyframe extraction algorithm can be based on extraction at preset time intervals, or it can be assisted by existing keyframe neural networks.

[0089] In the recognition of image frame content, as a preferred implementation option, this solution S03 preferably includes:

[0090] S031. Obtain the local recognition results and judge them according to preset conditions.

[0091] When it meets the first preset condition, it is marked with first information and a warning message is generated and output.

[0092] When it meets the second preset condition, mark it with the second information and proceed to S032.

[0093] When it meets the third preset condition, mark it with the third information and proceed to S033;

[0094] S032. Locate the elements marked with second information and third information in multiple keyframes, set them as feature objects, and then judge the constraint relationship between the feature objects corresponding to the second information and the feature objects corresponding to the third information according to preset conditions. When the constraint relationship does not meet the preset requirements, generate warning information and use it for output.

[0095] S033. Select at least one feature point on the feature object in the keyframe. At the same time, select at least one reference point in the upper part of the image acquisition area inside the elevator car, and make each feature point correspond to at least one reference point, wherein the location of the reference point is higher than the top of the feature object.

[0096] As an example, when the first information in this solution is an object prohibited from entering the elevator; S04 includes:

[0097] S041. Obtain the identification result, and when it matches the preset abnormal information and / or is marked as the first information, output the warning information.

[0098] As an example, combined with Figure 4 As shown, when the local identification results indicate that an electric bicycle has entered the elevator car in the key frame, the first information is marked, a warning message is generated, and it is used for output.

[0099] Combination Figure 5 As shown, in the local identification results, when a pet enters the elevator car in the keyframe, it is marked with the second information, while when a person enters, it is marked with the third information.

[0100] To facilitate the marking of characteristic objects entering the elevator car and to facilitate the assessment of potential dangers posed by unleashed pets, as a preferred implementation option, the second preset condition of this solution is that the local identification result identifies live animals (such as dogs, cats, etc.) that do not include humans; S032 includes:

[0101] S0321. Locate the elements marked with second and third information in multiple keyframes and set them as feature objects;

[0102] S0322. Remove the background from the keyframes, and then perform edge contour recognition on the feature objects marked with the second and third information respectively to generate edge contour data.

[0103] S0323. Obtain edge contour data.

[0104] When the distance between the edge contour data corresponding to the feature object marked with the second information and the edge contour data corresponding to the feature object marked with the third information is less than a preset value, the two are marked as a constraint relationship.

[0105] When the distance between the edge contour data corresponding to the feature object marked with the second information and the edge contour data corresponding to the feature object marked with the third information is greater than a preset value, the region between the two is extracted and shape is recognized. When the recognition result is that there is a strip-shaped object (such as a rope), the two are marked as a constraint relationship; otherwise, the relationship between the two is marked as an unconstrained relationship.

[0106] S0324. Statistical analysis of constraint relationships of feature objects marked with second information within a preset time period. When the number of statistical counts of unconstrained relationships exceeds a preset value, generate warning information and output it.

[0107] Combination Figures 6 to 8 As one example of implementation options, the third preset condition in this solution is that when the local recognition result identifies a person, abnormal situations such as a person falling, fighting, or being held hostage may occur. This solution obtains an image containing only the person by acquiring keyframes and then removing the background and preset elements. This image can then be combined with an existing trained localization neural network to locate the person's head and generate feature points (such as...). Figure 7 The feature points can be the forehead, ears, nose, etc. (the hands are not convenient as feature points because they are difficult to capture images and people move frequently). Inside the elevator car, a reference point located at the top can be selected (such as the top of the elevator control panel). Then, the distance between the two points can be used to determine whether there is abnormal movement of people. Then, the cumulative time can be used to determine whether there is a fall or other situation.

[0108] Combination Figure 9 As shown, in determining the situation of a fall, this solution S04 includes:

[0109] A041. Real-time recording of the distance between the feature points and reference points on the feature objects identified by the local recognition results as human features, generating the first reference distance;

[0110] A042. Obtain the first reference spacing. When it is greater than the preset first threshold distance, time it to obtain the duration of the feature object being greater than the first preset threshold in the first reference spacing, and obtain the duration of the first abnormal spacing.

[0111] A043. Obtain the duration of the first abnormal interval. When it is greater than the preset first time threshold, generate personnel fall information as early warning information output.

[0112] For example, when a person is standing inside an elevator car, the distance between the feature point and the reference point should be relatively fixed. If a fall occurs, the distance between the two will change abruptly. By timing this, it can be determined whether the person has fallen and is unable to get up. If a fall occurs, the situation can be reported to the monitoring backend so that monitoring personnel can focus on and investigate it.

[0113] Combination Figure 10 As shown, for abnormal situations such as fighting or hostage-taking, as an example, this solution S04 also includes:

[0114] B041. The local recognition results are identified as the feature points corresponding to human feature objects, and the spacing between them is recorded and monitored in real time to generate a second reference spacing.

[0115] B042. Obtain the second reference spacing. When it is greater than the preset second threshold distance, time it to obtain the duration of the feature object being greater than the second preset threshold at the second reference spacing, and obtain the duration of the second abnormal spacing.

[0116] B043. Obtain the duration of the second abnormal interval. When it is greater than the preset second time threshold, generate the abnormal interval of personnel as an early warning information output.

[0117] Combination Figure 11 As shown, based on the above, this embodiment also provides an elevator car monitoring system, which includes:

[0118] The image acquisition unit is one or more, and is used to acquire image data inside the elevator car according to preset conditions. The captured image data covers at least the elevator car door.

[0119] The server is used to store the image data acquired by the image acquisition unit;

[0120] The data processing unit is used to acquire image data according to preset conditions, locate and identify feature objects in the image data, and generate recognition results.

[0121] The data labeling unit is used to select feature points and reference points for feature objects in image data based on the recognition results, and generate feature points and reference points corresponding to the feature objects;

[0122] An object tracking unit is used to track objects based on recognition results and / or positional change data between feature points and reference points;

[0123] The instruction generation unit is used to output early warning information based on location change data and / or recognition results according to preset conditions.

[0124] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0125] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] The above description is only a part of the embodiments of the present invention and does not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for monitoring an elevator car, characterized in that, It includes: S01. Acquire image data inside the elevator car according to preset conditions. The captured image data shall at least cover the elevator car door. S02. Acquire image data according to preset conditions, locate and identify feature objects in the image data, and generate recognition results; S03. Based on the recognition results, feature points and reference points are selected for the feature objects in the image data, and feature points and reference points corresponding to the feature objects are generated. S04. Output early warning information based on the identification results and / or the positional change data between feature points and reference points; S01 includes: S011. Real-time liveness detection inside the elevator car to generate liveness detection data; S012. Acquire liveness detection data. When the liveness detection data indicates that no living object is detected inside the elevator car, image data of the elevator car is acquired at a first preset acquisition frequency. When the liveness sensor data indicates that there is a living object in the elevator car, the image data inside the elevator car is acquired at the second preset acquisition frequency, and the elevator operating parameters are also acquired in real time. S02 includes: S021. Acquire liveness detection data. When the liveness detection data indicates that there is a living object inside the elevator car, acquire image data at a first preset acquisition frequency; otherwise, acquire image data at a second preset acquisition frequency. S022. Extract image frames from the image data, then locate and exclude preset elements in the image frames, then perform overall recognition on the key frames and separate recognition on the remaining elements, generating one or more recognition results pointing to the whole and one or more pointing to the part corresponding to the key frames. S022 includes: S0221. Extract image frames from the image data using a keyframe extraction algorithm to obtain several keyframes; S0222. Locate the elements in the keyframe, and then exclude the preset elements; S0223. Perform overall recognition on the keyframes from which preset elements have been removed to obtain the overall recognition result; S0224. Extract the remaining elements from the keyframes after removing the preset elements, and perform recognition processing on them respectively to obtain local recognition results; S03 includes: S031. Obtain the local recognition results and judge them according to preset conditions. When it meets the first preset condition, it is marked with first information and a warning message is generated and output. When it meets the second preset condition, mark it with the second information and proceed to S032. When it meets the third preset condition, mark it with the third information and proceed to S033; S032. Locate the elements marked with second information and third information in multiple keyframes, set them as feature objects, and then determine the constraint relationship between the feature objects corresponding to the second information and the feature objects corresponding to the third information according to preset conditions. When the constraint relationship does not meet the preset requirements, generate warning information and use it for output. S033. Select at least one feature point on the feature object in the keyframe. At the same time, select at least one reference point in the upper part of the image acquisition area inside the elevator car, and make each feature point correspond to at least one reference point, wherein the location of the reference point is higher than the top of the feature object.

2. The elevator car monitoring method as described in claim 1, characterized in that, The second preset condition is that the local recognition result identifies the animal as a living animal that does not include humans; S032 includes: S0321. Locate the elements marked with second and third information in multiple keyframes and set them as feature objects; S0322. Remove the background from the keyframes, and then perform edge contour recognition on the feature objects marked with the second and third information respectively to generate edge contour data. S0323. Obtain edge contour data. When the distance between the edge contour data corresponding to the feature object marked with the second information and the edge contour data corresponding to the feature object marked with the third information is less than a preset value, the two are marked as a constraint relationship. When the distance between the edge contour data corresponding to the feature object marked with the second information and the edge contour data corresponding to the feature object marked with the third information is greater than a preset value, the region between the two is extracted and the shape is recognized. When the recognition result is that there are stripes, the two are marked as a constraint relationship; otherwise, the relationship between the two is marked as an unconstrained relationship. S0324. Statistical analysis of constraint relationships of feature objects marked with second information within a preset time period. When the number of statistical counts of unconstrained relationships exceeds a preset value, generate warning information and output it.

3. The elevator car monitoring method as described in claim 1, characterized in that, The first piece of information indicates that the object is prohibited from entering the elevator. The third preset condition is that the local recognition result identifies the person as a human. S04 includes: S041. Obtain the identification result, and when it matches the preset abnormal information and / or is marked as the first information, output the warning information; S04 also includes: A041. Real-time recording of the distance between the feature points and reference points on the feature objects identified by the local recognition results as human features, generating the first reference distance; A042. Obtain the first reference spacing. When it is greater than the preset first threshold distance, time it to obtain the duration of the feature object being greater than the first preset threshold in the first reference spacing, and obtain the duration of the first abnormal spacing. A043. Obtain the duration of the first abnormal interval. When it is greater than the preset first time threshold, generate personnel fall information as early warning information output.

4. The elevator car monitoring method as described in claim 3, characterized in that, S04 also includes: B041. The local recognition results are identified as the feature points corresponding to human feature objects, and the spacing between them is recorded and monitored in real time to generate a second reference spacing. B042. Obtain the second reference spacing. When it is greater than the preset second threshold distance, time it to obtain the duration of the feature object being greater than the second preset threshold at the second reference spacing, and obtain the duration of the second abnormal spacing. B043. Obtain the duration of the second abnormal interval. When it is greater than the preset second time threshold, generate the abnormal interval of personnel as an early warning information output.

5. An elevator car monitoring system for implementing the elevator car monitoring method according to any one of claims 1 to 4, characterized in that, It includes: The image acquisition unit is one or more, and is used to acquire image data inside the elevator car according to preset conditions. The captured image data covers at least the elevator car door. The server is used to store the image data acquired by the image acquisition unit; The data processing unit is used to acquire image data according to preset conditions, locate and identify feature objects in the image data, and generate recognition results. The data labeling unit is used to select feature points and reference points for feature objects in image data based on the recognition results, and generate feature points and reference points corresponding to the feature objects; An object tracking unit is used to track objects based on recognition results and / or positional change data between feature points and reference points; The instruction generation unit is used to output early warning information based on location change data and / or recognition results according to preset conditions.

6. A computer-readable storage medium, characterized in that: The storage medium stores at least one instruction, at least one program, code set, or instruction set, and the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the elevator car monitoring method as described in any one of claims 1 to 4.

Images