A method and device for identifying elevator call intention and elevator call control system

Abstract

The present invention relates to a method for identifying the intention of an external call to take an elevator, comprising: backtracking the surveillance video of a passenger pressing a button and extracting key frames to obtain an elevator key frame; performing button detection and directional intention recognition on a number of elevator key frames in sequence to obtain a button action frame and directional intention information, and establishing associations between the button action frame and a number of elevator key frames according to the directional intention information to obtain a monitoring list of the same batch; according to the monitoring list of the same batch, performing target tracking on a number of elevator key frames, and judging whether the current passenger has a change in the intention to take the elevator based on the result of the target tracking: if not, continuing the target tracking; if so, assuming that the passenger intends to stop calling the elevator. The method for identifying the intention of an external call to take an elevator described in the present invention has the advantages of significantly reducing misjudgments in complex elevator waiting area scenarios; effectively reducing the computing resources required for target tracking; more efficient resource utilization; and improving the stability and real-time performance of the identification of the intention to take the elevator.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation and elevator dispatching, and in particular to a method and device for identifying an elevator call intention, an elevator call control system, an electronic device and a computer storage medium. Background Art

[0002] With the continuous development of social economy and urbanization, elevators have become an indispensable means of vertical transportation in modern buildings. The basic working principle of general elevators is: the elevator call control system receives the elevator passengers pressing the up or down elevator call button on the elevator call box, that is, the elevator call button, such as Figure 1 As shown in the figure, the elevator call control system will control the elevator to arrive at the floor according to the corresponding elevator call button, so that passengers can take the elevator. Figure 2 As shown; then, the passengers click the target floor on the internal floor panel, and the elevator moves to the target floor smoothly at a constant speed or with uniform acceleration. When the elevator is moving, if a passenger on a floor on its path clicks the elevator call button in the same direction, the elevator will slow down and arrive at that floor to pick up passengers in the same moving direction.

[0003] However, in actual applications, the intention of passengers often changes, resulting in the elevator running empty. Specific situations include: passengers press the wrong elevator call button for various reasons, passengers on the lower floors wait for the elevator for too long and take the stairs instead, passengers temporarily change their itinerary and stop riding the elevator, etc., resulting in the elevator receiving the call command and arriving at the corresponding floor, but no passengers get on the elevator. This situation not only seriously affects the riding experience of other passengers, but also prolongs the waiting time for the next round of elevators, thereby reducing the overall operating efficiency of the elevator.

[0004] Based on this, in order to avoid the phenomenon of empty elevator running, the existing technology usually uses methods such as passenger identity recognition and passenger intention recognition to optimize elevator scheduling. Among them, passenger identity recognition is suitable for residential buildings and office buildings with strong privacy, but it has limited effect in buildings with extensive public areas. The passenger intention recognition is mainly performed through the following steps: real-time acquisition of surveillance video outside each elevator car for preprocessing; posture estimation of the preprocessed surveillance video to obtain the human skeleton nodes and the number of human skeletons of each passenger; and the key point coordinates of the human skeleton node graph are analyzed through the behavior analysis model of the support vector machine to obtain whether the current passenger has the intention to take the elevator; the elevator external call control system calls the elevator according to the intention to take the elevator, such as: whether to move the elevator to the current passenger's waiting floor for the passenger to enter the elevator.

[0005] However, if there are buildings with multiple public areas, such as shopping malls, hospitals, etc., where there is a large flow of people and frequent movement, the existing technology that only uses support vector machines to judge the intention of taking the elevator is prone to misjudgment, and the implementation of surveillance video collection, preprocessing, posture estimation and behavior analysis requires high computing resources. When the computing load increases, it may cause the system to respond slowly, affecting the real-time and accuracy of elevator scheduling. Summary of the invention

[0006] Based on this, the object of the present invention is to provide a method for identifying the intention of an external call to take the elevator.

[0007] A method for identifying an elevator call intention comprises the following steps:

[0008] S1: Backtrack the surveillance video of the passenger pressing the button and extract the key frame to obtain the key frame of taking the elevator;

[0009] S2: Perform key detection and direction intention recognition on several elevator key frames in sequence to obtain key action frames and direction intention information, and establish associations between the key action frames and several elevator key frames according to the direction intention information to obtain a monitoring list of the same batch;

[0010] S3: According to the monitoring list of the same batch, target tracking is performed on several elevator key frames, and based on the target tracking results, it is determined whether the current passenger has a change in the intention to take the elevator: if not, target tracking continues; if so, it is considered that the passenger intends to stop calling the elevator.

[0011] Compared with the prior art, the method for identifying the intention of taking the elevator by calling the elevator described in the present invention adopts a two-way single-round detection strategy, i.e., key action detection, by combining the actual application scenario, and only focuses on identifying and tracking the intention of the first batch of passengers to take the elevator, thereby representing the waiting passengers in the same direction but in different batches, so that the system can significantly reduce the misjudgment in complex waiting area scenarios. In addition, by tracking the target based on several key frames of taking the elevator, the system effectively reduces the computing resources required for target tracking, achieves more efficient resource utilization, and improves the stability and real-time performance of the recognition of the intention to take the elevator, providing a more intelligent solution for elevator scheduling in complex scenarios.

[0012] Furthermore, the step S1 includes the following sub-steps:

[0013] S11: Tracing back the surveillance video of the passenger pressing buttons to obtain relevant video segments;

[0014] Among them, for the current time t, the relevant video segment is specifically expressed as

[0015] t related ∈[ts,e],s <t

[0016] In the formula, t related is the relevant video segment; [ts,e] represents the video segment within s seconds from the current time t, s is the preset interaction time, and e represents the time of the analysis result of the elevator intention, that is, the current time of the monitoring video;

[0017] Get relevant video segments relatrd Corresponding video frame sequence Frame n , and use the inter-frame difference method to extract key frames to obtain several ladder key frames. The specific calculation expression is as follows:

[0018] For continuous frames i and Frame i+1 The average difference intensity D i It is expressed as:

[0019]

[0020] In the formula, Frame i (x, y) represents the pixel value of the i-th frame at the pixel coordinate (x, y), W and H represent the width and height of each frame respectively; i belongs to the interval [1, n], and n is the total number of the video frame sequence, which is calculated according to the monitored refresh rate fps, specifically calculated as: fps×(e-t+s); then, a difference threshold is used to calculate all the average difference intensity sets D={D1,D2,…,D n-1} to filter and obtain several ladder key frames, whose difference threshold is:

[0021] D Th =max(D)-ΔD var

[0022] In the formula, max(·) represents the maximum average difference intensity in the current set D of all average difference intensities; ΔD var is a custom differential intensity variable; several ladder keyframes are represented as:

[0023] KeyFrame={(Frame i ,t j )|i∈[1,n],t j ∈[ts,e]}

[0024] In the formula, Frame i represents the i-th frame in the current video frame sequence; t j Represented as a video frame Frame i The timestamp of the belonging.

[0025] The present invention backtracks based on the instant time when the indicator light turns on after the passenger presses the button to ensure that the passenger's button action is successfully captured within the time period of the surveillance video, and highlights the key action frames through average differential intensity, providing representative data for subsequent target recognition and target tracking, ensuring that the system can still accurately identify the intention to take the elevator in complex scenarios and reduce the possibility of misjudgment. In addition, by only processing the key frames within the relevant time period, the system effectively reduces the amount of data processing and optimizes the use of computing resources, thereby significantly improving the real-time performance and response speed of the system, thereby enabling the system to accurately identify the intention to take the elevator in complex waiting scenarios and quickly respond to the elevator's scheduling situation to prevent the elevator from running empty, thereby improving the efficiency and reliability of the overall system.

[0026] Furthermore, the step S2 includes the following sub-steps:

[0027] S21: Use an image classification model to classify the image at timestamp t j ∈[ts,t] to perform key detection on several key frames of the elevator to obtain key action frames;

[0028] S22: Determine whether the current key action frame is a single frame or a double frame: if it is a single frame, obtain direction intention information according to the direction corresponding to the indicator light triggered when the passenger presses the key; if it is a double frame, determine the timestamp sequence corresponding to the double frames according to the sequence of the directions corresponding to the indicator lights triggered when the passenger presses the key, and obtain the direction intention information;

[0029] S23: According to the direction intention information, the unique feature information of the passenger who performs the key action in the key action frame is identified and used as the key passenger KeyID, which is represented as KeyID = {Feature KeyID}; Based on the unique characteristic information of key passengers, the human skeleton spatiotemporal relationship model M is used r Perform correlation detection on several elevator key frames to obtain the PID of the group of passengers traveling together, which is specifically expressed as:

[0030] PID={M r (Feature LeyID ,(Frame i ,t j ))|i∈[1,n],t j ∈[ts,t]}={PID1,PID2,…,PID i}

[0031] Where M r (Feature KeyID ,(Frame i ,t j )) is expressed as the unique characteristics of the key passengers according to the input; (Framei ,t j ) is the key frame of taking the elevator in the time interval [ts, t];

[0032] The key passengers and the set of traveling passengers are associated to obtain a monitoring list of the same batch, which is expressed as:

[0033] Group up ={KeyID,PID1,PID2,…,PID i}

[0034] and / or,

[0035] Gruop down ={KeyID,PID1,PID2,…,PID j}

[0036] In the formula, Group up Used to indicate the same batch monitoring list in the upward direction, Group down It is used to indicate the monitoring list of the same batch in the downward direction; i and j represent the total number of passengers in the upward and downward directions respectively; among them, PID i Includes the posture characteristics and action characteristics of the i-th passenger.

[0037] The present invention captures the passenger who presses the first button as the key passenger, and associates the key passenger with the fellow passengers, so that the system can focus on the key passenger and his fellow passengers, and effectively judge the intention of taking the elevator before the current elevator arrives. At the same time, it reduces the need for comprehensive monitoring of all passengers waiting for the elevator, and centrally processes the intention of the first batch of passengers in the same direction to take the elevator, further reducing unnecessary resource consumption and increasing the real-time and response speed of the system. In addition, the association strategy of the present invention improves the robustness and stability of the system in complex scenarios, ensuring that even when multiple batches of passengers are waiting for the elevator at the same time, the system can still accurately identify and respond to the actual needs of passengers, and accurately track and manage key passengers and fellow passengers.

[0038] Furthermore, the step S3 includes the following sub-steps:

[0039] S31: According to the same batch monitoring list Group up and / or Group down , using the target tracking model M Tracking Target tracking is performed on several passenger key frames to obtain several real-time action Act and track Track lists. For the key passenger KeyID in the same batch monitoring list, the specific expression of target tracking is as follows:

[0040] M Tracling (KeyID,(Framei ,t j )),t j ∈[et,e]

[0041] ={Act KeyID ={act1,…,act i},Track KeyID ={position1,…,position i}}

[0042] In the formula, act i Indicates the action type corresponding to the key passenger in the i-th frame, position i Indicates the position of the key passenger in the i-th frame; the e in the time interval [et,e] is updated in real time according to the current time of the surveillance video;

[0043] S32: judging whether the current passenger has a change in the intention to take the elevator according to the real-time action and trajectory list: if so, it is considered that the passenger intends to stop calling the elevator; if not, executing step S31;

[0044] Among them, for the same batch monitoring list Group up and / or Group down The judgment expression of the jth passenger's intention to take the elevator is as follows:

[0045]

[0046] In the formula, act i is the action type of the i-th frame, and act i The action list of the jth passenger in the same batch monitoring list; distance(·) is used to calculate the distance between two input data, and left is the distance threshold for judging the position of leaving the elevator; and It is used to determine whether, among all subsequent frames k, there is at least one subsequent frame k where the distance between the passenger and the elevator is less than or equal to the distance threshold; and It is used to determine whether the distance between the passengers and the elevator in all subsequent frames k is greater than the distance threshold; E is the coordinate position of the elevator; position k is the passenger trajectory position of the kth frame, and position k The trajectory list of the jth passenger belonging to the same batch monitoring list; where k∈[i,n], n is the total number of video frame sequences.

[0047] The present invention conducts a detailed analysis of the movements and trajectories of all passengers in the same batch monitoring list and sets clear conditional judgments, namely, when and only when all passengers in the same batch monitoring list are in a moving state and have reached a certain distance away from the elevator, it is determined that the passengers collectively intend to leave the elevator waiting area, and this is used as a condition for stopping the elevator call, thereby ensuring that the intention of a single passenger will not affect the elevator experience of other waiting passengers in the same batch. At the same time, it also significantly reduces the empty running phenomenon of the elevator due to misjudgment, thereby optimizing the overall scheduling efficiency of the elevator system.

[0048] A device for identifying the intention of an elevator call, comprising a monitoring key frame acquisition unit, a same batch target direction intention association unit and an elevator intention identification unit;

[0049] The monitoring key frame acquisition unit is used to trace back the monitoring video of the passenger pressing the key and extract the key frame to obtain the elevator key frame;

[0050] The same batch target direction intention association unit is used to perform key detection and direction intention recognition on a number of elevator key frames in sequence, obtain key action frames and direction intention information, and establish association between the key action frames and a number of elevator key frames according to the direction intention information to obtain a same batch monitoring list;

[0051] The elevator intention recognition unit is used to track the target of several elevator key frames according to the same batch monitoring list, and judge whether the current passenger has changed his intention to take the elevator based on the target tracking result: if not, continue to track the target; if so, it is considered that the passenger intends to stop calling the elevator.

[0052] Further, the monitoring key frame acquisition unit includes a monitoring backtracking module and a monitoring key frame extraction module;

[0053] The monitoring backtracking module is used to backtrack the monitoring video of the passenger's keystrokes to obtain relevant video segments;

[0054] Among them, for the current time t, the relevant video segment is specifically expressed as

[0055] t related ∈[ts,e],s <t

[0056] In the formula, t related is the relevant video segment; [ts,e] represents the video segment within s seconds from the current time t, s is the preset interaction time, and e represents the time of the analysis result of the elevator intention, that is, the current time of the monitoring video;

[0057] The monitoring key frame extraction module is used to obtain the relevant video segment t related Corresponding video frame sequence Frame n, and use the inter-frame difference method to extract key frames to obtain several ladder key frames. The specific calculation expression is as follows:

[0058] For continuous frames i and Frame i+1 The average difference intensity D i It is expressed as:

[0059]

[0060] In the formula, Frame i (x, y) represents the pixel value of the i-th frame at the pixel coordinate (x, y), W and H represent the width and height of each frame respectively; i belongs to the interval [1, n], and n is the total number of the video frame sequence, which is calculated according to the monitored refresh rate fps, specifically calculated as: fps×(e-y+s); then, a difference threshold is used to calculate all the average difference intensity sets D={D1,D2,…,D n-1} to filter and obtain several ladder key frames, whose difference threshold is:

[0061] D Th =max(D)-ΔD var

[0062] In the formula, max(·) represents the maximum average difference intensity in the current set D of all average difference intensities; ΔD var is a custom differential intensity variable; several ladder keyframes are represented as:

[0063] KeyFrame={(Frame i ,t j )|i∈[1,n],t j ∈[ts,e]}

[0064] In the formula, Frame i represents the i-th frame in the current video frame sequence; t j Represented as a video frame Frame i The timestamp of the belonging.

[0065] Further, the same batch target direction intention association unit includes a key detection module, a direction intention labeling module and a direction intention association module;

[0066] The key detection module is used to use an image classification model to detect the key at the time stamp t j ∈[ts,t] to perform key detection on several key frames of the elevator to obtain key action frames;

[0067] The direction intention labeling module is used to determine whether the current key action frame is a single frame or a double frame: if it is a single frame, the direction intention information is obtained according to the direction corresponding to the indicator light triggered when the passenger presses the key; if it is a double frame, the time stamp sequence corresponding to the double frame is determined according to the sequence of the directions corresponding to the indicator lights triggered when the passenger presses the key, and the direction intention information is obtained;

[0068] The direction intention association module is used to identify the unique feature information of the passenger who performs the key action in the key action frame according to the direction intention information, and use it as the key passenger KeyID, which is expressed as KeyID={Feature KeyID}; Based on the unique characteristic information of key passengers, the human skeleton spatiotemporal relationship model M is used r Perform correlation detection on several elevator key frames to obtain the PID of the group of passengers traveling together, which is specifically expressed as:

[0069] PID={M r (Feature LeyID ,(Frame i ,t j ))|i∈[1,n],t j ∈[ts,t]}={PID1,PID2,…,PID i}

[0070] Where M r (Feature KeyID ,(Frame i ,t j )) is expressed as the unique characteristics of the key passengers according to the input; (Frame i ,t j ) is the key frame of taking the elevator in the time interval [ts, t];

[0071] The key passengers and the set of traveling passengers are associated to obtain a monitoring list of the same batch, which is expressed as:

[0072] Group up ={KeyID,PID1,PID2,…,PID i}

[0073] and / or,

[0074] Gruop down ={KeyID,PID1,PID2,…,PID j}

[0075] In the formula, Group up Used to indicate the same batch monitoring list in the upward direction, Group downIt is used to indicate the monitoring list of the same batch in the downward direction; i and j represent the total number of passengers in the upward and downward directions respectively; among them, PID i Including the posture characteristics and action characteristics of the i-th passenger;

[0076] The elevator intention recognition unit includes a same batch target tracking module and a same batch target elevator intention judgment module;

[0077] The same batch target tracking module is used to monitor the same batch of targets according to the list Group up and / or Group down , using the target tracking model M Tracking Target tracking is performed on several passenger key frames to obtain several real-time action Act and track Track lists. For the key passenger KeyID in the same batch monitoring list, the specific expression of target tracking is as follows:

[0078] M Tracking (KeyID,(Frame i ,t j )),t j ∈[et,e]

[0079] ={Act KeyID ={act1,…,act i},Track KeyID ={position1,…,position i}}

[0080] In the formula, act i Indicates the action type corresponding to the key passenger in the i-th frame, position i Indicates the position of the key passenger in the i-th frame; the e in the time interval [et,e] is updated in real time according to the current time of the surveillance video;

[0081] The same batch target elevator intention judgment module is used to judge whether the current passenger has an elevator intention change according to the real-time action and trajectory list: if so, it is considered that the passenger intends to stop calling the elevator; if not, step S31 is executed;

[0082] Among them, for the same batch monitoring list Group up and / or Gruop down The judgment expression of the jth passenger's intention to take the elevator is as follows:

[0083]

[0084] In the formula, act i is the action type of the i-th frame, and act iThe action list of the jth passenger in the same batch monitoring list; distance(·) is used to calculate the distance between two input data, and left is the distance threshold for judging the position of leaving the elevator; and It is used to determine whether, among all subsequent frames k, there is at least one subsequent frame k where the distance between the passenger and the elevator is less than or equal to the distance threshold; and It is used to determine whether the distance between the passengers and the elevator in all subsequent frames k is greater than the distance threshold; E is the coordinate position of the elevator; position k is the passenger trajectory position of the kth frame, and position k The trajectory list of the jth passenger belonging to the same batch monitoring list; where m∈[i,n], n is the total number of video frame sequences.

[0085] An elevator call control system includes a plurality of camera devices, a plurality of elevator call buttons, and an elevator call intention recognition device electrically and / or communicatively connected to the plurality of camera devices and the plurality of elevator call buttons;

[0086] The plurality of camera devices are arranged in the elevator waiting area to monitor the elevator call button on the corresponding floor and send the monitoring video to the elevator call intention recognition device;

[0087] The plurality of elevator call buttons are arranged in the elevator waiting area, and are used for passengers to press buttons in corresponding directions to call an elevator to the corresponding elevator waiting floor;

[0088] The elevator call intention recognition device is used to identify whether the current passenger has changed his / her intention to take the elevator from the monitoring video: if not, continue to identify; if so, cancel the elevator call button corresponding to the elevator waiting area and cancel the elevator call;

[0089] Among them, the device for identifying the intention of an outside call to take the elevator is the device for identifying the intention of an outside call to take the elevator described above.

[0090] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 This is a schematic diagram of the appearance of the elevator call button;

[0092] Figure 2 A schematic diagram showing an example of the elevator riding process for elevator passengers;

[0093] Figure 3 It is a structural schematic diagram of the device for identifying the intention of calling an elevator to board the elevator according to the present invention;

[0094] Figure 4It is a schematic flow chart of the method for identifying the intention of calling an elevator to board the elevator according to the present invention;

[0095] Figure 5 Schematic diagram of the strategy for identifying and tracking objects for the purpose of the present invention. DETAILED DESCRIPTION

[0096] In order to solve the problem of insufficient real-time and accuracy in judging the intention of taking the elevator in the prior art, the present invention obtains the key frames of taking the elevator in the monitoring video, and identifies the key actions of the passengers in the key frames of taking the elevator, obtains the key action frames and the direction intention information of the corresponding passengers, establishes associations between the key action frames and the key frames of taking the elevator according to the direction intention information, obtains the same batch monitoring list, and tracks the corresponding passengers according to the list to obtain the real-time action and trajectory list. According to the real-time action and trajectory list, it is judged in real time whether the current passenger has changed his intention to take the elevator: if so, a cancel call control instruction is sent to the elevator call control system. Accordingly, the present invention enables the system to track and analyze the behavior and trajectory of the passenger group more accurately by constructing the same batch monitoring list, thereby reducing the misjudgment caused by individual behavior. At the same time, by performing association analysis based on the key frames of taking the elevator, not only the accuracy of the system in judging the intention of taking the elevator is improved, but also the unnecessary calculation burden of the system caused by misjudgment is reduced, and the real-time performance of the system is significantly improved.

[0097] Based on the above design, the present invention proposes a method for identifying the intention of an outside call to take the elevator, and based on this method, proposes a device for identifying the intention of an outside call to take the elevator.

[0098] See also Figure 3 and Figure 4 , Figure 3 Schematic diagram of the structure of the device for identifying the intention of calling the elevator according to the present invention, Figure 4 The present invention is a flow chart of the method for identifying the intention of an elevator call.

[0099] The device for identifying the intention of an external call to take the elevator comprises a monitoring key frame acquisition unit 1, a same-batch target direction intention association unit 2 and an elevator intention identification unit 3.

[0100] The monitoring key frame acquisition unit 1 is used to execute step S1: backtrack the monitoring video of the passenger pressing the button and extract the key frame to obtain the elevator key frame.

[0101] Specifically, the monitoring key frame acquisition unit 1 includes a monitoring backtracking module 11 and a monitoring key frame extraction module 12 .

[0102] The monitoring backtracking module 11 is used to execute step S11: backtrack the monitoring video of the passenger pressing the button to obtain the relevant video segment.

[0103] Specifically, for the current time t, the relevant video segment is specifically expressed as

[0104] t related ∈[ts,e],s <t

[0105] In the formula, t related is the relevant video segment; [ts,e] represents the video segment traced back from the current time t within s seconds, s is the preset interaction time, the default is 1-2 seconds, and e represents the result time of the elevator intention analysis, that is, the current time (real-time) of the monitoring video until the result of the elevator intention analysis is obtained.

[0106] Depending on different elevator systems or actual scenarios, such as shopping malls and residences, the preset interaction time s is selected differently, and the present invention is not specifically limited to this. It is only necessary to ensure that the relevant video segment includes the time period from the passenger pressing the button to the indicator light turning on.

[0107] The present invention captures the time segment when the passenger presses the elevator call button by backtracking, thereby ensuring that all behaviors and situations related to the passenger's key action are captured, providing key data for subsequent intent analysis.

[0108] The monitoring key frame extraction module 12 is used to execute step S12: extract key frames from relevant video segments to obtain a number of elevator riding key frames.

[0109] Specifically, obtain the relevant video segment t related Corresponding video frame sequence Frame n , and use the inter-frame difference method to extract key frames to obtain several ladder key frames. The specific calculation expression is as follows:

[0110] For continuous frames i and Frame i+1 The average difference intensity D i It is expressed as:

[0111]

[0112] In the formula, Frame i (x, y) represents the pixel value of the i-th frame at the pixel coordinate (x, y), W and H represent the width and height of each frame respectively; i∈[1, n] and n is the total number of the video frame sequence, which is calculated according to the monitored refresh rate fps×(e-t+s), and n increases in real time according to the time e of the result of the elevator intention analysis. Among them, the refresh rate fps has different values ​​according to the performance of the device, and the present invention is not specifically limited here.

[0113] Next, a difference threshold is used to evaluate all the average difference strength sets D = {D1, D2, ..., D n-1} to screen and obtain several ladder key frames, whose difference threshold D Th for:

[0114] D Th =max(D)-ΔD var

[0115] In the formula, max(·) represents the maximum average difference intensity in the current set D of all average difference intensities; ΔD var It is a user-defined differential strength variable. Different differential strength variables can be defined according to specific needs to adjust the acquisition interval of the elevator key frame. The present invention does not specifically limit the value of the differential strength variable.

[0116] Its several key frames can be expressed as:

[0117] KeyFrame={(Frame i ,t j )|i∈[1,n],t j ∈[ts,e]}

[0118] In the formula, Frame i represents the i-th frame in the current video frame sequence; t j Represented as a video frame Frame i The timestamp of the belonging.

[0119] The present invention extracts key frames that best reflect passenger behavior from a large number of surveillance video frames through key frame extraction, so as to highlight the effective information of the most passengers' intention to take the elevator and related contextual content, and effectively remove redundant information in the video frames, thereby improving the accuracy of the system's judgment of the intention to take the elevator. At the same time, the amount of image data that the system needs to process is reduced to reduce the demand for computing resources, thereby improving the overall processing efficiency of the system.

[0120] In addition, the present invention uses differential strength variables to adjust the screening range of elevator key frames, ensuring that the system can select the most representative key frames according to actual conditions, rather than relying solely on a fixed differential strength threshold, to adapt to a variety of complex situations and optimize the effect of key frame extraction, thereby improving the performance and reliability of the overall system.

[0121] The same batch target direction intention association unit 2 is used to execute step S2: perform key detection and direction intention recognition on several elevator key frames in turn, obtain key action frames and direction intention information, and establish association between the key action frames and several elevator key frames according to the direction intention information to obtain a same batch monitoring list.

[0122] Specifically, the same-batch target direction intention association unit 2 includes a key detection module 21 , a direction intention labeling module 22 and a direction intention association module 23 .

[0123] The key detection module 21 is used to execute step S21: perform key detection on a number of elevator riding key frames to obtain key action frames.

[0124] Specifically, an image classification model Model is used a , for the time stamp t j ∈[ts,t] to perform key detection on several ladder key frames and obtain key action frames.

[0125] The image classification model a The system comprises a feature extraction module, a feature fusion module and a key classification module. The feature extraction module is used to extract features from a number of elevator key frames to obtain feature maps of different scales. The feature fusion module is used to perform feature fusion on a number of fused feature maps to obtain fused feature maps of different scales. The key classification module is used to perform key detection on fused feature maps of different scales to obtain key action frames.

[0126] Among them, the image classification model Model a You can choose YOLOv10 (You Only Look Once Version 10), and manually label the buttons and human movements in the elevator car video to build a training data set. Then, use this training data set to classify the image model Model a Train and verify to make Model a If the accuracy of key recognition and elevator waiting meets the standard, the trained image classification model Model is obtained. a .

[0127] Since different image classification models can be selected according to different hardware conditions, computing resources or accuracy requirements, the present invention does not specifically limit the structure and selection of the model.

[0128] It should be emphasized that the key action frame obtained by the present invention is actually the video frame of the indicator light turning on after the first passenger presses the key, and if the indicator light is not off, the repeated key actions of other passengers in the same batch will not be captured. Therefore, the present invention uses the indicator light turning on after pressing the key as the starting point for detecting a batch of passengers waiting for the elevator.

[0129] The direction intention labeling module 22 is used to execute step S22: labeling the key action frame according to the time correspondence to obtain the direction intention information.

[0130] Specifically, it is determined whether the current key action frame is a single frame or a double frame: if it is a single frame, the direction intention information is obtained according to the direction corresponding to the indicator light triggered when the passenger presses the key;

[0131] If it is a double frame, the order of the timestamps corresponding to the double frames is determined according to the order of the directions of the indicator lights triggered when the passenger presses the button, and the direction intention information is obtained.

[0132] Since there is a time sequence between the indicator lights lighting up after the passenger presses the button, and the passenger's goal may only be to go up or down, and will not choose both directions at the same time, the key passenger's directional intention can be accurately judged based on the order in which the indicator lights light up corresponding to the button action frame, so that the directional intention labeling can be completed efficiently and accurately, ensuring the intelligent response of the elevator dispatching system.

[0133] The direction intention association module 23 is used to execute step S23: according to the direction intention information, the key action frame is associated with a plurality of elevator riding key frames to obtain a monitoring list of the same batch.

[0134] Specifically, according to the direction intention information, the unique feature information of the passenger who performs the key action in the key action frame is identified and used as the key passenger KeyID, which is expressed as KeyID = {Feature KeyID}.

[0135] According to the unique characteristic information of key passengers, the human skeleton spatiotemporal relationship model M is used. r Perform correlation detection on several elevator key frames to obtain the PID of the group of passengers traveling together, which is specifically expressed as:

[0136] PID={M r (Feature KeyID ,(Frame i ,t j ))|i∈[1,n],t j ∈[ts,t]}={PID1,PID2,…,PID i}

[0137] Where M r (Feature KeyID ,(Frame i ,t j )) is expressed as the unique characteristics of the key passengers according to the input; (Frame i ,t j ) is the key frame of taking the elevator in the time interval [ts,t].

[0138] The key passengers and the set of traveling passengers are associated to obtain a monitoring list of the same batch, which is expressed as:

[0139] Group up ={KeyID,PID1,PID2,…,PID i}

[0140] and / or,

[0141] Group down ={KeyID,PID1,PID2,…,PID j}

[0142] In the formula, Group up Used to indicate the same batch monitoring list in the upward direction, Group down It is used to indicate the monitoring list of the same batch in the downward direction. i and j represent the total number of passengers in the same batch in the upward and downward directions respectively.

[0143] The same batch monitoring list can determine the passenger group belonging to the upward, downward or both upward and downward directions according to the direction intention information, which is specifically determined according to the key action frame corresponding to the given direction intention information.

[0144] Wherein, the human skeleton spatiotemporal relationship model m r It includes a posture estimation module, a spatiotemporal feature dimension extraction module, a spatiotemporal feature fusion module and a PID generation module. The posture estimation module is used to extract the posture of the key frames of the elevator in the time interval [ts, t] to obtain the posture features of all passengers.

[0145] The spatiotemporal feature extraction module is used to extract spatial and temporal features from the posture features of all passengers and obtain a number of spatiotemporal feature information.

[0146] The spatiotemporal feature fusion module is used to fuse the spatiotemporal feature information of each passenger to generate the motion feature of each passenger.

[0147] The PID generation module is used to match and identify the unique characteristics of the key passenger with the action characteristics of all passengers, obtain the characteristics of the passengers traveling with the key passenger, assign the corresponding PID to the traveling passengers, and output the PID of the traveling passenger set. i}, and PID i Including the posture features and action features of the i-th fellow passenger, which are used to provide sufficient information for subsequent target tracking.

[0148] Furthermore, the elevator key frame in the time interval [ts, t] input by the human skeleton spatiotemporal relationship model can be dynamically expanded, specifically, according to the current elevator key frame (Frame i ,tj ), extended to a sequence of adjacent video frames as input, the expression of the sequence of adjacent video frames is as follows:

[0149] {Frame i-c ,…,Frame i ,…,Frame i+c |c≤fps}

[0150] By providing more temporal information to the human skeleton spatiotemporal model through adjacent video frame sequences, the model's recognition accuracy of fellow passengers is improved.

[0151] In addition, the human skeleton spatiotemporal relationship model M r Spatial Temporal Graph Convolutional Networks (ST-GCN) can be selected and trained based on a self-constructed training data set of fellow passengers to identify and assign PIDs to fellow passengers. However, users can select different spatiotemporal relationship models to identify and assign PIDs to fellow passengers according to different needs and different key frame extraction methods. Therefore, the present invention does not specifically limit the selection of the human skeleton spatiotemporal relationship model.

[0152] In the actual elevator scene, when the first batch of passengers press the up or down button, the indicator light will turn on, and the subsequent passengers will not press the button again, but only need to wait until the indicator light turns off. Therefore, for the target of this study, we only need to judge the intention of the first batch of passengers to characterize whether there are passengers waiting for the elevator in the current round. Figure 5 As shown, the computational burden of target recognition and target tracking is reduced, and the effectiveness of the system in identifying changes in elevator intentions is ensured.

[0153] The elevator intention recognition unit 3 is used to execute step S3: according to the same batch monitoring list, target tracking is performed on several elevator key frames, and based on the target tracking result, it is judged whether the current passenger has any change in the elevator intention: if not, the target tracking is continued; if so, it is considered that the passenger intends to stop calling the elevator.

[0154] Specifically, the elevator intention recognition unit 3 includes a same-batch target tracking module 31 and a same-batch target elevator intention judgment module 32.

[0155] The same batch target tracking module 31 is used to execute step S31: according to the same batch monitoring list, target tracking is performed on a number of elevator riding key frames to obtain a number of real-time action and trajectory lists.

[0156] Specifically, according to the same batch monitoring list Group up and / or Groupdown , using the target tracking model M Tracking Target tracking is performed on several passenger key frames to obtain several real-time action Act and track Track lists. For the key passenger KeyID in the same batch monitoring list, the specific expression of target tracking is as follows:

[0157] M Tracking (1KeyID,(Frame i ,t j )),t j ∈[et,e]

[0158] ={Act KeyID ={act1,…,act i},Track KeyID ={position1,…,position i}}

[0159] In the formula, act i Indicates the action type corresponding to the key passenger in the i-th frame, position i Indicates the position of the key passenger in the i-th frame. In the time interval [et,e], e is a dynamic value, that is, it is updated in real time according to the current time, and is used to represent the target tracking interval from the associated passenger to the current time. For example, the target tracking interval t of the first cycle j ∈[10s,20s], in the next cycle when e=e+10, the target tracking interval is updated to t j ∈[20s,30s].

[0160] Wherein, the target tracking model M Tracking It includes feature extraction module, feature fusion module, target association module, action recognition module and trajectory recognition module.

[0161] The feature extraction module is used to extract features from a number of passenger key frames to obtain a number of feature vectors.

[0162] The feature fusion module is used to perform feature fusion on a plurality of feature vectors to obtain a plurality of passenger fusion feature vectors.

[0163] The target association module is used to match the passenger fusion feature vector with the passenger information in the same batch monitoring list to obtain the matching fusion feature vector to ensure continuous tracking of the target passenger.

[0164] The action recognition module is used to perform action recognition on the matched fusion feature vectors to obtain an action list corresponding to each passenger, specifically:

[0165] Act j={act1,…,act i}

[0166] Where j is the jth passenger in the same batch monitoring list, act i is the action type corresponding to the i-th frame. The action type may be moving and waiting for an elevator. Different action types may be selected according to different scenes and requirements, and the present invention does not specifically limit this.

[0167] The trajectory recognition module is used to track the matching fusion feature vectors and obtain a trajectory list corresponding to each passenger, specifically:

[0168] Track j ={position1,…,position i}

[0169] In the formula, j is the jth passenger in the same batch monitoring list, position i is the trajectory coordinate corresponding to the i-th frame, and the trajectory coordinate may be a joint coordinate, a detection frame coordinate, etc., which is not specifically limited in the present invention.

[0170] In addition, the target tracking model M Tracking The deep multi-target tracking algorithm (DeepSORT) can be used in combination with the human skeleton spatiotemporal relationship model M r However, according to different requirements and different types of data input, users can select different target tracking models to achieve multi-target tracking. Therefore, the present invention does not specifically limit the selection of target tracking models.

[0171] The same batch target taking the elevator intention judgment module 32 is used to execute step S32: based on the real-time action and trajectory list, determine whether the current passenger has a change in the intention to take the elevator: if so, send a cancel call control instruction to the elevator call control system; if not, call the same batch target tracking module 31.

[0172] Specifically, for the same batch monitoring list Group up and / or Group down The judgment expression of the jth passenger's intention to take the elevator is as follows:

[0173]

[0174] In the formula, act i is the action type of the i-th frame, and act iThe action list of the jth passenger in the same batch monitoring list; distance(·) is used to calculate the distance between two input data, left is the distance threshold for judging the distance from the elevator position, and different distance thresholds can be defined in actual situations. The present invention does not specifically limit the selection of the distance threshold; and It is used to determine whether, among all subsequent frames k, there is at least one subsequent frame k where the distance between the passenger and the elevator is less than or equal to the distance threshold; and It is used to determine whether the distance between the passengers and the elevator in all subsequent frames k is greater than the distance threshold; e is the coordinate position of the elevator, which is known data and is defined according to the actual scenario; position k is the passenger trajectory position of the kth frame, and position k The trajectory list of the jth passenger belonging to the same batch monitoring list. Where k∈[i,n], n is the total number of video frame sequences.

[0175] Based on this, if the same batch monitoring list Group up and / or Group down If there is at least one passenger whose intention of taking the elevator is to wait for the elevator, it is considered that there is no change in the intention of the current passenger to take the elevator, and the same batch target tracking module 31 is called. up and / or Group down If all passengers intend to leave, it is considered that the current passenger's intention to take the elevator has changed, and a cancel hall call control instruction is sent to the elevator hall call control system, so that the elevator hall call control system cancels the elevator call.

[0176] The present invention analyzes real-time actions and trajectories to accurately distinguish whether passengers are waiting for the elevator or leaving the elevator. In addition, in order to ensure that passengers waiting for the elevator will not be misjudged, the present invention only determines that the intention to take the elevator has changed if all passengers in the same batch monitoring list show signs of leaving, thereby balancing the passenger experience of passengers waiting for the elevator and passengers in the elevator, thereby improving passenger satisfaction.

[0177] Compared with the existing technology, the present invention combines the actual elevator riding scene, focuses on the target identification and tracking of the first batch of elevator passengers, and represents the intention of other batches of passengers in the same direction to take the elevator, so that the system can effectively identify the true intention of passengers in complex scenarios and significantly reduce the misjudgment rate. At the same time, the present invention performs behavioral analysis based on the elevator key frame to ensure that the analysis results are representative, thereby reducing the system's computing resource consumption and improving the real-time and accuracy of the elevator dispatch response.

[0178] Based on the same inventive concept, the present application also provides an electronic device, which may be a terminal device such as a server, a desktop computing device or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet computer, a netbook, etc.). The device includes one or more processors and a memory, wherein the processor is used to execute a program to implement the method for recognizing the intention of calling an elevator in an embodiment of the present invention; and the memory is used to store a computer program that can be executed by the processor.

[0179] Based on the same inventive concept, the present application also provides a computer-readable storage medium, corresponding to an embodiment of the aforementioned method for identifying an intention of an outside call for the elevator, wherein the computer-readable storage medium stores a computer program thereon, and when the program is executed by a processor, the steps of the method for identifying the intention of an outside call for the elevator recorded in any of the above embodiments are implemented.

[0180] The present application may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-usable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be achieved by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0181] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, and the present invention is also intended to include these modifications and modifications.

Claims

1. A method for identifying an elevator call intention, characterized in that: The following steps are involved: S1: Backtrack the surveillance video of the passenger pressing the key and extract the key frame to obtain the elevator key frame; wherein, step S1 includes the following sub-steps: S11: Tracing back the surveillance video of the passenger pressing buttons to obtain relevant video segments; Among them, for the current time t, the relevant video segment is specifically expressed as t relatea ∈[ts,e],s <t Where, t related is the relevant video segment; [ts,e] represents the video segment within s seconds from the current time t, s is the preset interaction time, and e represents the time of the analysis result of the elevator intention, that is, the current time of the monitoring video; S12: Obtain relevant video segment t related Corresponding video frame sequence Frame n , and use the inter-frame difference method to extract key frames to obtain several ladder key frames. The specific calculation expression is as follows: For continuous frames i and Frame i+1 The average difference intensity D i It is expressed as: In the formula, Frame i (x, y) represents the pixel value of the i-th frame at the pixel coordinate (x, y), W and H represent the width and height of each frame respectively; i belongs to the interval [1, n], and n is the total number of the video frame sequence, which is calculated according to the monitored refresh rate fps, specifically calculated as: fps×(e-t+s); then, a difference threshold is used to calculate all the average difference intensity sets D={D1,D2,…,D n-1 } to screen and obtain several ladder key frames, whose difference threshold D Th for: D Th =max(D)-ΔD var In the formula, max(·) represents the maximum average difference intensity in the current set D of all average difference intensities; ΔD var is a custom differential intensity variable; several ladder keyframes are represented as: KeyFrame={(Frame i ,t j )|i∈[1,n],t j ∈[t-s,e]} In the formula, Frame i represents the i-th frame in the current video frame sequence; t j Represented as a video frame Frame i timestamp of belonging; S2: performing key detection and direction intention recognition on a plurality of elevator key frames in sequence, obtaining key action frames and direction intention information, and establishing associations between the key action frames and a plurality of elevator key frames according to the direction intention information, and obtaining a monitoring list of the same batch; wherein, step S2 includes the following sub-steps: S21: Use an image classification model to classify the image at timestamp t j ∈[ts,t] to perform key detection on several key frames of the elevator to obtain key action frames; S22: Determine whether the current key action frame is a single frame or a double frame: if it is a single frame, obtain direction intention information according to the direction corresponding to the indicator light triggered when the passenger presses the key; if it is a double frame, determine the timestamp sequence corresponding to the double frames according to the sequence of the directions corresponding to the indicator lights triggered when the passenger presses the key, and obtain the direction intention information; S23: According to the direction intention information, the unique feature information of the passenger who performs the key action in the key action frame is identified and used as the key passenger KeyID, which is represented as KeyID = {Feature KeyID }; Based on the unique characteristic information of key passengers, the human skeleton spatiotemporal relationship model M is used r Perform correlation detection on several elevator key frames to obtain the PID of the group of passengers traveling together, which is specifically expressed as: PID={M r (Feature KeyID ,(Frame i ,t j ))|i∈[1,n],t j ∈[t-s,t]}={PID1,PID2,…,PID i } Where M r (Feature KeyID ,(Frame i ,t j )) is expressed as the unique characteristics of the key passengers according to the input; (Frame i ,t j ) is the key frame of taking the elevator in the time interval [ts, t]; The key passengers and the set of traveling passengers are associated to obtain a monitoring list of the same batch, which is expressed as: Group up ={KeyID,PID1,PID2,…,PID i } and / or, Group down ={KeyID,PID1,PID2,…,PID j } In the formula, Group up Used to indicate the same batch monitoring list in the upward direction, Group down It is used to indicate the monitoring list of the same batch in the downward direction; i and j represent the total number of passengers in the upward and downward directions respectively; among them, PID i Including the posture characteristics and action characteristics of the i-th passenger; S3: According to the monitoring list of the same batch, target tracking is performed on several elevator key frames, and based on the target tracking results, it is determined whether the current passenger has a change in the intention to take the elevator: if not, target tracking continues; if so, it is considered that the passenger intends to stop calling the elevator.

2. The method for identifying the intention of calling an elevator according to claim 1, characterized in that: The step S3 comprises the following sub-steps: S31: According to the same batch monitoring list Group up and / or Group down , using the target tracking model M Tracking Target tracking is performed on several passenger key frames to obtain several real-time action Act and track Track lists. For the key passenger KeyID in the same batch monitoring list, the specific expression of target tracking is as follows: M Tracking (KeyID,(Frame i ,t j )),t j ∈[e-t,e] ={Act KeyID ={act1,…,act i },Track KeyID ={Position1,…,position i }} In the formula, act i Indicates the action type corresponding to the key passenger in the i-th frame, position i Indicates the position of the key passenger in the i-th frame; the e in the time interval [et,e] is updated in real time according to the current time of the surveillance video; S32: judging whether the current passenger has a change in the intention to take the elevator according to the real-time action and trajectory list: if so, it is considered that the passenger intends to stop calling the elevator; if not, executing step S31; Among them, for the same batch monitoring list Group up and / or Group down The judgment expression of the jth passenger's intention to take the elevator is as follows: In the formula, act i is the action type of the i-th frame, and act i The action list of the jth passenger in the same batch monitoring list; distance(·) is used to calculate the distance between two input data, and left is the distance threshold for judging the position of leaving the elevator; and It is used to determine whether, among all subsequent frames k, there is at least one subsequent frame k where the distance between the passenger and the elevator is less than or equal to the distance threshold; and It is used to determine whether the distance between the passengers and the elevator in all subsequent frames k is greater than the distance threshold; E is the coordinate position of the elevator; position k is the passenger trajectory position of the kth frame, and position k The trajectory list of the jth passenger belonging to the same batch monitoring list; where k∈[i,n], n is the total number of video frame sequences.

3. A device for identifying the intention of calling an elevator, characterized in that: It includes a monitoring key frame acquisition unit, a same batch target direction intention association unit and an elevator intention recognition unit; The monitoring key frame acquisition unit is used to trace back the monitoring video of the passenger pressing the key and extract the key frame to obtain the elevator key frame; wherein the monitoring key frame acquisition unit includes a monitoring backtracking module and a monitoring key frame extraction module; The monitoring backtracking module is used to backtrack the monitoring video of the passenger's keystrokes to obtain relevant video segments; Among them, for the current time t, the relevant video segment is specifically expressed as t related ∈[ts,e],s <t Where, t related is the relevant video segment; [ts,e] represents the video segment within s seconds from the current time t, s is the preset interaction time, and e represents the time of the analysis result of the elevator intention, that is, the current time of the monitoring video; The monitoring key frame extraction module is used to obtain the relevant video segment t related Corresponding video frame sequence Frame n , and use the inter-frame difference method to extract key frames to obtain several ladder key frames. The specific calculation expression is as follows: For continuous frames i and Frame i+1 The average difference intensity D i It is expressed as: In the formula, Frame i (x, y) represents the pixel value of the i-th frame at the pixel coordinate (x, y), W and H represent the width and height of each frame respectively; i belongs to the interval [1, n], and n is the total number of the video frame sequence, which is calculated according to the monitored refresh rate fps, specifically calculated as: fps×(e-t+s); then, a difference threshold is used to calculate all the average difference intensity sets D={D1,D2,…,D n-1 } to screen and obtain several ladder key frames, whose difference threshold D Th for: D Th =max(D)-ΔD var In the formula, max(·) represents the maximum average difference intensity in the current set D of all average difference intensities; ΔD var is a custom differential intensity variable; several ladder keyframes are represented as: KeyFrame={(Frame i ,t j )|i∈[1,n],t j ∈[t-s,e]} In the formula, Frame i represents the i-th frame in the current video frame sequence; t j Represented as a video frame Frame i timestamp of belonging; The same batch target direction intention association unit is used to perform key detection and direction intention recognition on a number of elevator key frames in sequence, obtain key action frames and direction intention information, and establish associations between the key action frames and a number of elevator key frames according to the direction intention information to obtain a same batch monitoring list; wherein, the same batch target direction intention association unit includes a key detection module, a direction intention labeling module and a direction intention association module; The key detection module is used to use an image classification model to detect the key at the time stamp t j ∈[ts,t] to perform key detection on several key frames of the elevator to obtain key action frames; The direction intention labeling module is used to determine whether the current key action frame is a single frame or a double frame: if it is a single frame, the direction intention information is obtained according to the direction corresponding to the indicator light triggered when the passenger presses the key; if it is a double frame, the time stamp sequence corresponding to the double frame is determined according to the sequence of the directions corresponding to the indicator lights triggered when the passenger presses the key, and the direction intention information is obtained; The direction intention association module is used to identify the unique feature information of the passenger who performs the key action in the key action frame according to the direction intention information, and use it as the key passenger KeyID, which is expressed as KeyID={Feature KeyID }; Based on the unique characteristic information of key passengers, the human skeleton spatiotemporal relationship model M is used r Perform correlation detection on several elevator key frames to obtain the PID of the group of passengers traveling together, which is specifically expressed as: PID={M r (Feature KeyID ,(Frame i ,t j ))|i∈[1,n],t j ∈[t-s,t]}={PID1,PID2,…,PID i } Where M r (Feature KeyID ,(Frame u ,t j )) is expressed as the unique characteristics of the key passengers according to the input; (Frame i ,t j ) is the key frame of taking the elevator in the time interval [ts, t]; The key passengers and the set of traveling passengers are associated to obtain a monitoring list of the same batch, which is expressed as: Group up ={KeyID,PID1,PID2,…,PID i } and / or, Group down ={KeyID,PID1,PID2,…,PID j } In the formula, Group up Used to indicate the same batch monitoring list in the upward direction, Group down It is used to indicate the monitoring list of the same batch in the downward direction; i and j represent the total number of passengers in the upward and downward directions respectively; among them, PID i Including the posture characteristics and action characteristics of the i-th passenger; The elevator intention recognition unit is used to track the target of several elevator key frames according to the same batch monitoring list, and judge whether the current passenger has changed his intention to take the elevator based on the target tracking result: if not, continue to track the target; if so, it is considered that the passenger intends to stop calling the elevator.

4. The device for recognizing the intention of calling an elevator according to claim 3, characterized in that: The elevator intention recognition unit includes a same batch target tracking module and a same batch target elevator intention judgment module; The same batch target tracking module is used to monitor the same batch of targets according to the list Group up and / or Group down , using the target tracking model M Tracking Target tracking is performed on several passenger key frames to obtain several real-time action Act and track Track lists. For the key passenger KeyID in the same batch monitoring list, the specific expression of target tracking is as follows: M Tracking (KeyID,(Frame i ,t j )),t j ∈[e-t,e] ={Act KeyID ={act1,…,act i },Track KeyID ={position1,…,position i }} In the formula, act i Indicates the action type corresponding to the key passenger in the i-th frame, position i Indicates the position of the key passenger in the i-th frame; the e in the time interval [et,e] is updated in real time according to the current time of the surveillance video; The same batch target elevator intention judgment module is used to judge whether the current passenger has an elevator intention change according to the real-time action and trajectory list: if so, it is considered that the passenger intends to stop calling the elevator; if not, step S31 is executed; Among them, for the same batch monitoring list Group up and / or Group down The judgment expression of the jth passenger's intention to take the elevator is as follows: In the formula, act i is the action type of the i-th frame, and act i The action list of the jth passenger in the same batch monitoring list; distance(·) is used to calculate the distance between two input data, and left is the distance threshold for judging the position of leaving the elevator; and It is used to determine whether, among all subsequent frames k, there is at least one subsequent frame k where the distance between the passenger and the elevator is less than or equal to the distance threshold; and It is used to determine whether the distance between the passengers and the elevator in all subsequent frames k is greater than the distance threshold; E is the coordinate position of the elevator; position k is the passenger trajectory position of the kth frame, and position k The trajectory list of the jth passenger belonging to the same batch monitoring list; where k∈[i,n], n is the total number of video frame sequences.

5. An elevator call control system, characterized in that: It includes a plurality of camera devices, a plurality of elevator call buttons, and an elevator call intention recognition device electrically and / or communicatively connected to the plurality of camera devices and the plurality of elevator call buttons; The plurality of camera devices are arranged in the elevator waiting area to monitor the elevator call button on the corresponding floor and send the monitoring video to the elevator call intention recognition device; The plurality of elevator call buttons are arranged in the elevator waiting area, and are used for passengers to press buttons in corresponding directions to call an elevator to the corresponding elevator waiting floor; The elevator call intention recognition device is used to identify whether the current passenger has changed his / her intention to take the elevator from the monitoring video: if not, continue to identify; if so, cancel the elevator call button corresponding to the elevator waiting area and cancel the elevator call; Wherein, the device for identifying the intention of an external call to take the elevator is the device for identifying the intention of an external call to take the elevator as described in any one of claims 3-4.

6. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a method for identifying an intention to call an elevator as described in any one of claims 1 to 2 when executing the computer program.

7. A computer-readable storage medium storing computer-executable instructions, characterized in that: The computer executable instructions are used for a method for identifying an elevator call intention as described in any one of claims 1 to 2.

Images