A method and device for checking

Abstract

The present invention provides a verification method and device for solving the problem of automated verification. The method comprises: if the location corresponding to a first ID is within a first detection area, and a first face and a first biometric feature are detected in a first video stream; then it is determined that the target corresponding to the first ID is within the first detection area. The method improves the efficiency and accuracy of automated verification.

Description

Technical Field

[0001] The present invention relates to the field of data, and in particular to a verification method and device. Background Art

[0002] In certain special places and management environments, it is often necessary to frequently conduct regular or irregular inspections of the management objects, such as the daily management of criminals in prisons, the management of workers in factories, or the management of animals on farms.

[0003] In the prior art, a list of managed objects is usually used, and the management personnel conduct on-site inspections one by one, and count the number of managed objects at the same time. However, this traditional inspection method is not only inefficient, but also labor-intensive. Summary of the invention

[0004] In view of this, the embodiments of the present invention provide an inspection method and device, which realize an automated inspection method and can determine the location of an inspector who is not on site through a short-range wireless positioning method.

[0005] One aspect of the present invention provides a verification method, the method comprising:

[0006] Receiving a first instruction, wherein the first instruction indicates a first set of IDs that need to be named;

[0007] If the target corresponding to the first ID is within the first detection area, the first ID in the first ID set is deleted, and the next step is continued to determine whether the targets corresponding to the remaining first IDs in the first ID set appear in the first detection area, and the cycle is performed until a preset time;

[0008] Determine a list of absent persons based on the list of persons not detected in the first detection area;

[0009] Wherein, if the location corresponding to the first ID is within the first detection area, and the first biometric feature is detected in the first video stream, it is determined that the target corresponding to the first ID is within the first detection area;

[0010] in,

[0011] The first ID has a corresponding relationship with the first biometric feature;

[0012] The first video stream is a video stream generated by shooting the first detection area;

[0013] The target carries a positioning device corresponding to the first ID;

[0014] Among them, the positioning method is short-range wireless positioning, including:

[0015] Determine at least one region corresponding to each target, and determine the confidence between any two regions;

[0016] For each of the targets, multiple positioning detections are performed on each of the targets, and the confidence of each of the at least one preset area is determined by the positioning results of the multiple positioning detections and the transfer coefficients of any two preset areas;

[0017] Selecting the location area with the highest confidence as the positioning result corresponding to the target;

[0018] Among them, for each of the targets, multiple positioning detections are performed on each of the targets, and the confidence of each area in at least one preset area is determined by the positioning results of the multiple positioning detections and the transfer coefficients of any two preset areas, including:

[0019] For each positioning detection, based on the transfer coefficients of the positioning area corresponding to the current positioning detection and the positioning area corresponding to the previous positioning detection, the confidence of the positioning area corresponding to the current positioning detection and the confidence of other areas are adjusted respectively.

[0020] Optionally, the determining whether the target corresponding to the first ID is located within the first detection area includes:

[0021] If the location corresponding to the first ID is within the first detection area,

[0022] It is then detected whether the first biometric feature appears in the first video stream.

[0023] Optionally, the determining whether the target corresponding to the first ID is located within the first detection area includes:

[0024] Obtaining a first ID set located within the first detection area,

[0025] Obtaining a corresponding face set according to the first ID set;

[0026] Determining whether the face identified in the first video stream matches any face in the face set;

[0027] If there is a match, the first biometric feature is detected in the first video stream.

[0028] Optionally, the determining that the target corresponding to the first ID is within the first detection area includes:

[0029] Determining whether the first biometric feature is detected in a second video stream, the area captured by the second video stream at least including the first detection area;

[0030] If the location corresponding to the first ID is within the first detection area,

[0031] If the first biometric feature is detected in the second video stream, it is determined that the target corresponding to the first ID is located within the first detection area.

[0032] Optionally, a shooting angle between the first video stream and the second video stream is greater than 5 degrees.

[0033] Optionally, the first detection area is a one-way entrance and exit area.

[0034] Optionally, the determining whether the target corresponding to the first ID is located within the first detection area includes:

[0035] Determine whether the location corresponding to the first ID is within the first detection area through one-dimensional UWB positioning coordinates;

[0036] Optionally, the determining whether the target corresponding to the first ID is located within the first detection area includes:

[0037] obtaining the status of the positioning device corresponding to the first ID,

[0038] If the status of the positioning device corresponding to the first ID is wearing status,

[0039] If the location corresponding to the first ID is within the first detection area and the first biological feature is detected in the first video stream, it is determined that the target corresponding to the first ID is within the first detection area.

[0040] Optionally, obtaining the state of the positioning device corresponding to the first ID includes:

[0041] Detect the heart rate of the target by using the positioning device,

[0042] and / or detecting the state of the anti-disassembly circuit of the positioning device;

[0043] If the heart rate of the target is detected and / or the anti-disassembly circuit indicates that it is not disassembled, the state of the positioning device corresponding to the first ID is the wearing state.

[0044] Optionally, the positioning device is a UWB positioning device;

[0045] The method includes: the first biometric feature is a human face;

[0046] And / or the method includes: if it is determined that the target corresponding to the first ID does not appear in the first detection area, sending an alarm message;

[0047] And / or the method includes: if it is determined that the target corresponding to the first ID does not appear in the first detection area, obtaining a video corresponding to the positioning according to the positioning corresponding to the first ID;

[0048] And / or the method includes: if it is determined that the target corresponding to the first ID appears in the first detection area, starting to collect the first video stream.

[0049] In a second aspect, the present invention further provides a checking device, comprising:

[0050] Positioning node, positioning base station, first image acquisition device and server;

[0051] The first image acquisition device is used to provide a first video stream to the server;

[0052] The server includes one or more memories, one or more processors, and one or more modules, wherein the one or more modules are stored in the memories and configured to be executed by the one or more processors, and the one or more modules include instructions for or respectively for executing the following steps:

[0053] Solve the positioning signal sent by the positioning node to the positioning base station;

[0054] And, the instructions corresponding to the above-mentioned inspection method.

[0055] Optional,

[0056] Also included is a second image acquisition device, the second image acquisition device is used to provide a second video stream to the server; the one or more modules include instructions for or respectively for performing the following steps: before determining that the target corresponding to the first ID is within the first detection area, including:

[0057] Determining whether the first biometric feature is detected in a second video stream, the area captured by the second video stream at least including the first detection area;

[0058] If the location corresponding to the first ID is within the first detection area,

[0059] If the first biometric feature is detected in the second video stream, it is determined that the target corresponding to the first ID is located within the first detection area.

[0060] Optionally, an angle between the first image acquisition device and the second image acquisition device is greater than 5 degrees.

[0061] Optionally, the one or more modules include instructions for or respectively for executing the following steps: determining whether the location corresponding to the first ID is within the first detection area based on the one-dimensional UWB positioning coordinates.

[0062] Optionally, the positioning node includes a heart rate detection circuit and an anti-disassembly circuit, and the server receives heart rate detection data and an anti-disassembly circuit status signal;

[0063] The one or more modules include instructions for performing the following steps:

[0064] obtaining the status of the positioning device corresponding to the first ID,

[0065] If the status of the positioning device corresponding to the first ID is wearing status,

[0066] If the location corresponding to the first ID is within the first detection area and the first biological feature is detected in the first video stream, it is determined that the target corresponding to the first ID is within the first detection area.

[0067] Optionally, the positioning device is a UWB positioning device;

[0068] The server is further configured to execute, wherein the first biometric feature is a human face;

[0069] and / or: if it is determined that the target corresponding to the first ID does not appear in the first detection area, sending an alarm message;

[0070] and / or: if it is determined that the target corresponding to the first ID does not appear in the first detection area, obtaining a video corresponding to the positioning according to the positioning corresponding to the first ID;

[0071] And / or: If it is determined that the target corresponding to the first ID appears in the first detection area, start collecting the first video stream.

[0072] Optionally, the server includes at least two servers, and the two servers respectively include partial modules of the multiple modules.

[0073] The present invention enables the target to carry a positioning device, and determines that the location corresponding to the first ID is within the first detection area through the positioning signal sent by the positioning device, and at the same time obtains the face recognition result of the M:N control scene. The target is judged to be within the first detection area through the positioning device and at the same time, it is judged through image recognition that the target appears in the first detection area, thereby ultimately improving the accuracy of judging that the target is located in the first detection area. The above scheme overcomes the problem that the target may wear other people's positioning equipment and cause false detection, and at the same time reduces the number of cameras deployed, reduces the time required for target verification, and overcomes the problem of inaccurate face recognition in M:N dynamic control scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0075] Figure 1 A schematic diagram of a flow chart of a checking method provided by the present invention;

[0076] Figure 2 A schematic diagram of a flow chart of another inspection method provided by the present invention;

[0077] Figure 3 A schematic diagram of a scene for checking the activity area of ​​an object provided by an embodiment of the present invention is shown;

[0078] Figure 4 A schematic diagram of a scenario of checking by short-range wireless positioning provided by an embodiment of the present invention is shown;

[0079] Figure 5 A schematic diagram of a consistency detection scenario provided by an embodiment of the present invention is shown;

[0080] Figure 6 The figure is a schematic diagram of a scene corresponding to an embodiment of the present invention. DETAILED DESCRIPTION

[0081] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It can be understood that the embodiments in this case can be used in any combination unless they are obviously not suitable for combination, and the specific embodiments do not limit the combination.

[0082] In one embodiment of the present invention, Figure 1 As shown, the method includes:

[0083] If the location corresponding to the first ID is within the first detection area,

[0084] and a first biometric feature is detected in the first video stream;

[0085] It is determined that the target corresponding to the first ID is located within the first detection area;

[0086] in,

[0087] The first ID has a corresponding relationship with the first biometric feature;

[0088] The first video stream is a video stream generated by shooting the first detection area;

[0089] The target carries a positioning device corresponding to the first ID.

[0090] like Figure 6 As shown, in an indoor scene in an embodiment of the present invention, the hardware required to implement the above method includes positioning nodes 151-153 that send positioning signals; positioning base station 131 that sends reference signals; multiple base stations that receive the positioning signals and the reference signals; server 121 that calculates the node position; first image acquisition device 141 and a network connecting the base station and the server. In some embodiments, a management terminal 161 may also be included, and the management terminal is used to configure the system, issue instructions, receive alarms, etc. In some embodiments, other mobile devices that can be used to issue instructions, receive alarms or calculation results may also be configured through the management terminal.

[0091] In the present invention, the node that sends the positioning signal can be a positioning device in the form of a watch, a work badge, etc. The positioning signal sent is a UWB signal. Each node has an ID that can uniquely identify the node, also called the first ID; in some embodiments, the target is required to wear the node that sends the positioning signal, that is, the target's location can be known from the position of the node that sends the positioning signal.

[0092] In one embodiment of the present invention, the target is a person, and the first biometric feature is a face. The target is required to wear a positioning watch as a positioning device, and the server records the face image or feature value of the face image of the person wearing the positioning watch, that is, the face image or feature value of the face image of the person wearing the watch can be determined by the ID of the watch. However, in actual situations, it is common for the target to wear someone else's positioning device. For example, in a roll call scenario, in order to avoid roll call, colleagues are often asked to wear their own work badges or positioning watches, etc., which leads to errors in roll call through the positioning device.

[0093] The first biometric feature is a biometric feature used to uniquely identify a target, such as tiger patterns for tigers, human faces for humans, etc. The targets 101-103 can be humans or animals. It is understandable that if the ID of the positioning device uniquely corresponds to the target, it can be known that the positioning device ID and the first biometric feature corresponding to the target have a unique corresponding relationship without a doubt. In this article, the target is a person and the first biometric feature is a human face as an example to describe the present invention, but it is not limited to this.

[0094] The first detection area is a fixed area defined, or is related to the area captured by the camera. For example, the area captured by the first image acquisition device 141 and the area of ​​the positioning base station 131 both include a common area. If the positioning of the first ID is located in the first detection area, the target corresponding to the first ID is located in the first area. Multiple people may appear in the area at the same time, or even multiple people appear densely. This scene is an M:N dynamic control scene. M:N dynamic control is to perform facial recognition on all people in the scene by a computer, that is, M people in the picture and N faces in the database. It can be understood that compared with 1:1 and 1:N scenes, M:N dynamic control can significantly reduce user time and reduce the number of deployed cameras. However, due to the large amount of calculation of this method, and the limitation of the pixels captured by the camera, and the fact that the camera usually captures a small face at a long distance, the face is not located in the center of the image, and the distortion caused by the camera is serious, the accuracy of face recognition in the M:N dynamic control scene is not high, the delay is long, etc., which ultimately leads to less application of face recognition in the M:N dynamic control scene.

[0095] For the first detection area, an image acquisition device, such as a fixed camera or a patrol camera, is used to capture the first detection area, thereby generating a first video stream. It is understandable that the area captured by the first video stream includes the first detection area, and may also capture areas outside the first detection area. For example, the camera is set in a corner of the room, and the first detection area is the half of the room close to the camera, but the camera can capture the other half of the area.

[0096] The faces in the first video stream are detected by image recognition technology. If the faces appearing in the first video stream include the first face corresponding to the first ID, it is determined that the first face appears in the first video stream, that is, the target is determined to appear in the first detection area through the video. It can be understood that in some embodiments of the present invention, all the faces appearing in the first video stream are first obtained, and it is determined whether the positioning devices corresponding to these faces appear in the first detection area. If so, the targets corresponding to these positioning devices appear in the first detection area. In another embodiment of the present invention, the IDs corresponding to all the positioning devices appearing in the first detection area are first detected, and the database is queried according to the ID to obtain the corresponding face set. If the face in the face set appears in the detection result of the first video stream, the target corresponding to the face is located in the first detection area.

[0097] It can be understood that the time corresponding to the first video stream in this case corresponds to the time when the first ID is located in the first detection area. That is, the time correspondence may not need to be indistinguishable at the millisecond level, but needs to be limited to the maximum movement speed of a person to ensure the accuracy of the detection.

[0098] The present invention uses a positioning device 1 carried by the target, and determines that the location corresponding to the first ID is within the first detection area through the positioning signal sent by the positioning device, and at the same time obtains the face recognition result of the M:N control scene. The target is judged to be within the first detection area through the positioning device and at the same time, it is judged through image recognition that the target appears in the first detection area, thereby ultimately improving the accuracy of judging that the target is located in the first detection area. The above scheme overcomes the problem that the target may wear other people's positioning equipment and cause false detection, while reducing the number of cameras deployed, reducing the time required for target verification, and overcoming the problem of inaccurate face recognition in M:N dynamic control scenarios.

[0099] In one embodiment of the present invention, the method performed includes:

[0100] If the target corresponding to the first ID is within the first detection area,

[0101] It is then identified whether the first biometric feature appears in the first video stream.

[0102] That is, in this embodiment, the execution method is:

[0103] If the target corresponding to the first ID is within the first detection area,

[0104] identifying whether a first face appears in the first video stream;

[0105] If the location corresponding to the first ID is within the first detection area,

[0106] and a first face is detected in the first video stream;

[0107] It is determined that the target corresponding to the first ID is located within the first detection area;

[0108] That is, in this embodiment, the IDs corresponding to all positioning devices appearing in the first detection area are first detected, and the corresponding face set is obtained by querying the database based on the ID. If a face in the face set appears in the detection result of the first video stream, the target corresponding to the face is located in the first detection area.

[0109] When the target corresponding to the first ID is not located in the first detection area, the above method does not identify the first video stream. In some embodiments, the first ID can be the ID of any positioning device, that is, the presence of any positioning device in the first detection area triggers the detection of whether a face appears in the first video stream. Since identifying the first video stream requires more computing resources, the above method reduces the length of time that the computer needs to process the video stream, and reduces the system operation concurrency and load. In some embodiments, the first video stream identifies the face in the first video stream by purchasing a third-party service. The purchase of the service is usually charged in units of data volume or duration. Therefore, the above scheme also reduces the cost of running the above scheme on one hand.

[0110] In some embodiments, the ID may be an ID in the target set that needs to be clocked in the first detection area, rather than an arbitrary ID, that is, only when a specific ID appears in the first detection area, the first video stream is detected, thereby reducing the length of the video stream that needs to be identified, and reducing the number of faces that need to be identified to improve the accuracy of face recognition. In the embodiment of this case, the above scheme includes the steps of: receiving the first instruction, wherein,

[0111] The first instruction indicates the location of the first detection area;

[0112] The first instruction indicates a set of the second IDs.

[0113] The first ID is an ID in the second ID set;

[0114] If the target corresponding to the first ID is within the first detection area,

[0115] identifying whether a first face appears in the first video stream;

[0116] If the location corresponding to the first ID is within the first detection area,

[0117] and a first face is detected in the first video stream;

[0118] It is determined that the target corresponding to the first ID is located within the first detection area;

[0119] The above method avoids people who do not need to clock in the first detection area from entering the first detection area and falsely triggering the recognition of the first video stream by preselecting and setting the ID set, thereby further reducing the length of the video stream that needs to be recognized. M:N dynamic control is to perform facial recognition of all people in the scene by a computer, that is, M people in the picture and N faces in the database, that is, this solution also receives a first instruction, and the first instruction indicates the first ID set that needs to be named, which is equivalent to further limiting the number of N, thereby further reducing the number of faces that need to be recognized and improving the accuracy of face recognition.

[0120] In some embodiments, the method of execution includes: obtaining a first ID set located within the first detection area,

[0121] Obtaining a corresponding face set according to the first ID set;

[0122] Determining whether the face identified in the first video stream matches any face in the face set;

[0123] M:N dynamic control is to use a computer to perform facial recognition on all people in the scene, that is, there are M people in the picture and N faces in the database. In the above scheme, the corresponding face set is obtained according to the first ID set; judging whether the face identified in the first video stream matches any face in the face set is equivalent to reducing the number of M, that is, reducing the number of faces that need to be identified. Since the number of faces that need to be identified is limited, the accuracy of face recognition is improved. In some scenarios, it is also possible to accept a first instruction, which indicates the first ID set that needs to be named, which is equivalent to further limiting the number of N, thereby further reducing the number of faces that need to be identified and improving the accuracy of face recognition.

[0124] In some embodiments, before determining that the target corresponding to the first ID is within the first detection area, the method includes:

[0125] If the target corresponding to the first ID is within the first detection area, deleting the first ID in the first ID set;

[0126] Continue to the next step of determining whether the targets corresponding to the remaining first IDs in the first ID set appear in the first detection area;.

[0127] M:N dynamic control is to perform facial recognition on all people in the scene by computer, that is, M people in the picture and N faces in the database, that is, in the scene corresponding to M:N dynamic control, the above scheme deletes the first ID in the first set, which is equivalent to reducing the number of N, that is, reducing the number of faces that need to be recognized. Since the number of faces that need to be recognized is limited, the accuracy of face recognition is improved.

[0128] It is understandable that the faces of people in the first detection area in this case are different. For example, in a factory environment or in a prison walk-in scene, workers need to complete their work according to the requirements of their positions and have no time to pay attention to the camera, or it is impossible for everyone to face the camera at a certain moment. In the prison walk-in scene, the prisoners' wandering is random. Therefore, in the above scenarios, by extending the time of the loop execution of the method of this case until the preset time is reached, the list of absent persons is determined based on the list of people who are not detected in the first detection area. After the list of absent persons is determined, it can be notified by an alarm or recorded in a log.

[0129] In one embodiment of the present case, the first detection area is a one-way entrance and exit area. By setting a camera in the one-way entrance area, more faces of people can be captured faster, thereby shortening the time required for detection and reducing the length of the video to be identified.

[0130] In one embodiment of the present case, a plurality of cameras are arranged in the first detection area.

[0131] In some scenarios, multiple cameras are set up in a first detection area of ​​a larger area. For example, one camera is used to provide a first video stream, and another camera is used to provide a second video stream. It is determined whether the first face is detected in the second video stream or the first video stream. The areas captured by the first video stream and the second video stream respectively capture different first detection areas. If the location corresponding to the first ID is located in the first detection area, and the first face is detected in the second video stream and / or the first video stream, it is determined that the target corresponding to the first ID is located in the first detection area. The above scheme is possible through the cooperation of multiple cameras in a larger scene.

[0132] In another scenario, a first instruction is received, a first ID set is determined, the first ID is an ID in the first ID set, a certain camera is used to provide a first video stream, and another camera is used to provide a second video stream, the area captured by the first video stream and the second video stream includes at least the same first detection area, if the location corresponding to the first ID is within the first detection area, and the first face is detected in the second video stream and / or the first video stream, then it is determined that the target corresponding to the first ID is within the first detection area, then the first ID in the first ID set is deleted, and the next step is continued to determine whether the targets corresponding to the remaining first IDs in the first ID set appear in the first detection area. In the above scheme, the detection of the first video stream and the second video stream is performed concurrently, and the information of the first ID set that needs to be identified is shared, that is, if one party detects that a certain ID has been identified and is located in the first detection area, then it is not necessary to analyze whether the ID appears in the video stream when analyzing the other video stream. M:N dynamic control is to use a computer to perform facial recognition on all people in the scene, that is, M people in the picture and N faces in the database, that is, when reducing the analysis of the video stream in the scene, the number of faces that need to be recognized in the video stream is deleted through the recognition of other video streams, which is equivalent to reducing the number of N through the recognition of other video streams, thereby improving the recognition speed and recognition accuracy. As a result, the above scheme can be applied to scenes with a large number of people and requiring rapid roll call. For example, in the prior art, a single person and a single ticket are always passed through the gate machine, while the above method allows multiple people to pass through the gate at the same time, and the passing speed can be very fast, that is, by setting multiple cameras at the gate, respectively collecting multiple video streams, and through the above positioning and video combined verification method, multiple people can pass quickly while having a high accuracy rate.

[0133] Through the above method, the shooting angle between the first video stream and the second video stream is greater than 5 degrees. The area shot by the first video stream and the second video stream is a one-way entrance and exit area.

[0134] In some embodiments of the present case, the hardware required to implement the above method includes a node that sends a positioning signal; a plurality of base stations that receive the positioning signal; a server that calculates the position of the node; and a network connecting the base station and the server. The multiple base stations are one-dimensional positioning base stations. For example, one base station is installed in a room. The base station can only detect the distance between the node that sends the test signal and the base station. It is easy to understand that if the accurate spatial position between the node and the base station is needed, at least two-dimensional positioning is required, but the number of base stations required for two-dimensional positioning is greater than that of one-dimensional positioning, so the cost is higher than that of one-dimensional positioning. In the above embodiment, the first detection area is an area within a certain radius of the corresponding base station. If it is detected that the distance between the positioning device and the base station is less than a preset threshold, it is determined that the positioning device is located in the first detection area. Through the above method, the number of base stations deployed is reduced, but the accuracy of the detection is also guaranteed at the same time.

[0135] In one embodiment of the present case, the determining whether the target corresponding to the first ID is located within the first detection area includes:

[0136] detecting the state of the positioning device corresponding to the first ID,

[0137] If the status of the positioning device corresponding to the first ID is wearing status,

[0138] If the location corresponding to the first ID is within the first detection area, and a first face is detected in the first video stream, then it is determined that the target corresponding to the first ID is within the first detection area.

[0139] If the state of the positioning device corresponding to the first ID is not in the wearing state, it is determined that the positioning device corresponding to the ID is abnormal, and the method of determining whether the target corresponding to the first ID appears in the first detection area is not performed, or it is directly determined that the target corresponding to the first ID is not located in the first detection area. The state of the positioning device corresponding to the first ID is detected, including:

[0140] Detect the heart rate of the target by using the positioning device,

[0141] and / or detecting the state of the anti-disassembly circuit of the positioning device;

[0142] If the heart rate of the target is detected and / or the anti-disassembly circuit indicates that it is not disassembled, the state of the positioning device corresponding to the first ID is the wearing state.

[0143] It can be understood that the anti-dismantling circuit is relative to the wearer, that is, in the case of illegal disassembly, the wearer needs to destroy the anti-dismantling circuit to dismantle the positioning device. Illegal disassembly of the positioning device may detect the disconnection of the circuit, for example, detecting a low level through the GPIO port of the circuit and judging that the circuit is disconnected.

[0144] In one embodiment of the present case, the method performed includes:

[0145] If the target corresponding to the first ID is within the first detection area, the first video stream is started to be collected. That is, in this embodiment, the execution method is:

[0146] If the target corresponding to the first ID is located in the first detection area, the first video stream is collected. The first video stream is detected; whether the first face appears in the first video stream is identified; if the location corresponding to the first ID is located in the first detection area, and the first face is detected in the first video stream; it is determined that the target corresponding to the first ID appears in the first detection area.

[0147] The above method does not capture the first video stream when the target corresponding to the first ID is not located in the first detection area. In some embodiments, the first ID can be the ID of any positioning device, that is, if any positioning device appears in the first detection area, it triggers the capture of the first video stream and detects whether a face appears in the first video stream. In other embodiments, it can also be an ID in the ID set of the target set that needs to punch in the first detection area, rather than an arbitrary ID, that is, only when a specific ID appears in the first detection area, the first video stream is captured. Since the operation of the camera requires a certain amount of electricity, the above scheme reduces the cost of running the above scheme. It can be understood that the start of the capture can also be to start the camera from the standby state or the sleep state to start capturing the first video stream.

[0148] In one embodiment of the present case, the first instruction includes: inspection task information. The inspection request is used to trigger the execution of an inspection of whether the target corresponding to the first ID set or the first ID is located in the first detection area, wherein the inspection execution end can be a server or a device terminal. Among them, the inspection task information indicates the inspection area and the inspection object; the inspection area represents the first detection area, and the inspection object can be understood as a person or object that needs to be inspected, that is, the target, and the first ID set or the first ID can be determined according to the inspection object. In addition, the inspection task information may also include: inspection method, inspection type or inspection time, etc., wherein the inspection method includes: short-distance wireless positioning method, image recognition method or a combination of short-distance wireless positioning and image recognition, etc. Inspection types may include: automatic inspection and manual inspection. The inspection time may include the following two situations:

[0149] 1) The current time period of receiving the verification request, for example, the current time period may be within 10 seconds after receiving the verification request;

[0150] 2) A designated inspection time period, such as inspection from 12:00 noon to 12:01 p.m.

[0151] Among them, with regard to the inspection time in the inspection task, when the inspection is performed simultaneously through the positioning method and the image recognition method, the two can be triggered at the same time, or the inspection operation can be performed in parallel within the same time period, that is, the positioning inspection result and the image inspection result are obtained through the above two inspection methods within the same time period.

[0152] In one embodiment of the present invention, the positioning method adopted is a short-range wireless positioning method. Each time the positioning device is positioned, multiple positioning detections are performed within a certain period of time or based on a preset frequency, and the multiple positioning results are compared to determine the final positioning result of this positioning detection.

[0153] For example: when the inspection object is positioned, the number of positioning times is set based on the refresh rate of the positioning device. For example, if there are N positioning times, and n of them are positioned in area A, and m are positioned in other areas, then if n and m meet certain conditions, the positioning result of the inspection object is considered to be area A. For example, when n=N and m=0, the positioning result of the inspection object is considered to be area A.

[0154] In one embodiment of the present invention, for an inspection object that has not been successfully inspected, the location of the inspection object that has not been successfully inspected can be determined by image recognition and short-range wireless positioning.

[0155] The above short-range wireless positioning method can be UWB (English full name: Ultra Wide Band, Chinese full name: Ultra Wideband Technology). Among them, UWB includes two functions of data transmission and positioning, and has accurate positioning and fast transmission speed. In addition, short-range wireless positioning methods can also use Bluetooth or Wi-Fi.

[0156] In one embodiment of the present case, when using short-range wireless positioning, if the object to be inspected is in a certain position for a long time, or if the target is located at the boundary of two areas, the target positioning often jumps at the boundary of the two areas. In order to solve the above problem, the following positioning method is disclosed in this embodiment.

[0157] refer to Figure 2 , shows a schematic diagram of a process of checking by a short-distance wireless positioning method provided by an embodiment of the present invention. In this embodiment, the method includes:

[0158] S201: Determine at least one region corresponding to each target, and determine the confidence between any two regions;

[0159] In this embodiment, the at least one preset area can be understood as an area where the target may be active, such as Figure 3As shown, in the farm, the areas where cows move include the milking area, the activity area, and the feeding area, wherein the feeding area includes: feeding area 1, feeding area 2, feeding area 3, and feeding area 4.

[0160] In this embodiment, the transfer coefficient is preset, the transfer coefficient between two connected regions is x, the transfer coefficient between two unconnected regions is y, and x>y. For example, x=1, y=0.2. For example: Figure 3 As shown, the areas with connectivity include: the activity area and the milking area, the activity area and the feeding area 4, the feeding area 4 and the feeding area 3, the feeding area 3 and the feeding area 2, and the feeding area 2 and the feeding area 1.

[0161] S202: for each of the targets, perform multiple positioning detections on each of the targets, and determine the confidence of each of the at least one preset area through the positioning results of the multiple positioning detections and the transfer coefficients of any two preset areas;

[0162] When checking the target, multiple positioning detections are performed on each target, and each positioning detection determines an area. In each positioning detection, the confidence of the area positioned this time can be calculated based on the positioning result and the transfer coefficients of any two areas. Specifically, S201 includes: for each positioning detection, based on the transfer coefficients of the positioning area corresponding to the current positioning detection and the positioning area corresponding to the previous positioning detection, the confidence of the positioning area corresponding to the current positioning detection and the confidence of other areas are adjusted respectively.

[0163] The confidence level of the current positioning detected in the location area a can be calculated by the following formula 1):

[0164] 1) V a (k+1)=V a (k)+P*T b-a ;

[0165] Among them, V a (k+1) represents the confidence of the location area a in the k+1th positioning detection, V a (k) represents the confidence of location area a during the kth positioning detection, p represents the area weight, T b-a Represents the transfer coefficient between location area b and location area a.

[0166] The methods for determining regional weights include:

[0167] Method 1

[0168] In each positioning detection, multiple positioning communications are performed on the area based on the set frequency, and the regional weight is determined based on the positioning result values ​​of multiple positioning communications. For example, in a positioning detection, it is set that each positioning detection needs to perform 10 positioning communications (that is, 10 positioning result values ​​can be obtained for each positioning detection). For example: the number of positioning communications performed for each positioning detection is N, and the N positioning result values ​​obtained are all located in the location area a, then p = p max， When only part of the positioning result values ​​are located in location area a, then p < p max， For example, p = 1 / 2*p max The positioning result value of the positioning communication may specifically be a coordinate value.

[0169] Method 2

[0170] In each positioning detection, it is set that each positioning detection needs to perform one positioning communication (that is, each positioning detection can obtain one positioning result value), and the regional weight is determined according to the M consecutive positioning result values. For example: M = 3, the positioning result value of the Kth positioning communication is located in area a, and the positioning result values ​​of the K-2 and K-1 positioning communications are also located in location area a, then p = p max， When some of the results are in location area a; and when only some of the K-2, K-1, and K-th positioning results are in location area a, then p < p max，

[0171] For example, p = 1 / 2*p max .

[0172] Method 3

[0173] In each positioning detection, it is set that each positioning detection needs to perform one positioning communication (that is, each positioning detection can obtain one positioning result value), and the regional weight is determined according to the positioning result value of the positioning communication and the error value corresponding to the result value.

[0174] Taking the weighted least squares positioning algorithm as an example, the observation equation is z=Hx+v, where z is the measurement vector, H is the observation matrix, v is the measurement error, and the measurement variance E(vv T ) = R. The standard weighted least squares estimate is The error variance is The positioning error vector err can be set to 3 times the variance Then there is a 99.7% probability that the positioning result falls within the positioning error. Generally, the radius r of the error circle can be set according to the maximum value of err in the x and y dimensions, and the positioning result value is within the estimated value. If the circle is located entirely within the position area a, then the weight of the area p = p max，When some results are in location area a, then p<p max， For example, p = 1 / 2*p max .

[0175] In this embodiment, in some special cases, for example, the location area of ​​the last positioning is a and the location area of ​​the current positioning is also a, then the transfer coefficient is T a-a , the transfer coefficient V a-a The maximum value can be taken. In this case, it can be calculated by the following formula 2):

[0176] 2) V a (k+1)=V a (k)+P*T a-a ;

[0177] The confidence of other areas can be calculated by the following formula 3), that is, when the location area of ​​the last positioning is a and the location area of ​​the current positioning is also a, the confidence of other area b is:

[0178] 3) V b (k+1)=V b (k)-P*T b-a ;

[0179] Among them, V b (k+1) represents the confidence level of the location in area b during the k+1th location detection, V b (k) represents the confidence of region b during the kth detection, P represents the region weight, T b-a Represents the transfer coefficient between region b and region a.

[0180] Furthermore, in order to avoid the confidence of a certain area increasing infinitely due to the positioning frequency being too high, thereby reducing the calculation accuracy and speed, and also to avoid the confidence of a certain area decreasing infinitely, confidence thresholds are set, including a maximum confidence threshold and a minimum confidence threshold. For example, the maximum threshold can be set to 50, and the minimum threshold can be set to 0.

[0181] S203: Selecting a location area with the highest confidence as a first target location area corresponding to the target.

[0182] In this embodiment, the confidences of all activity areas are obtained through the calculations of S201 - S202 , and the area with the highest confidence is selected as the positioning result, that is, as the first target location area.

[0183] Further, such as Figure 4 As shown, the calculation of the above regional confidence is further explained based on the actual scenario:

[0184] 1) Assume that the target activity areas include: outdoor, factory and corridor, where the corridor and factory are connected through door 1, and the corridor and outdoor can be connected through door 2. The movement trajectory of the inspector is shown in the figure as trajectory 1 and trajectory 2. Trajectory 1 shows that the inspector walks from the corridor to the factory, and trajectory 2 shows that the inspector has been active in the factory, but is located at the boundary between the factory and the outdoors. Positions 4, 5, 6, and 7 in trajectory 2 are located floating points, and positions 8 and 9 in trajectory 1 are located floating points.

[0185] 2) Based on whether the regions are connected, define the transfer coefficient between any two regions, such as T 室外-走廊 =1,T 走廊-工厂 =1,T 室外-工厂 =0.2.

[0186] 3) For trajectory 1, position point 1 and its previous series of position points are all in the corridor, so a larger corridor confidence V is obtained 走廊 , V 室外 =0 and V 工厂 = 0, when location point 2 enters the factory, assuming that through at least one positioning communication, all possible areas of location point 2 are determined to be within the factory, then V 走廊 =V 走廊- T 走廊-工厂 *P, at this time if V 走廊 Still greater than V 工厂 , then the positioning result will still be constrained to the corridor. With the generation of new position points 3 and 4, V 工厂 Will increase rapidly, V 走廊 Rapidly decreases, and the trajectory is not located outdoors, then V 室外 =0, so that V 工厂 >V 走廊 The positioning result will be constrained to the factory, and the same is true for trajectory 2.

[0187] In the above embodiment, the target is positioned multiple times, and based on the results of the multiple positioning and the transfer coefficients of any two areas, the confidence of each area is calculated, and the one with the largest confidence is selected as the final positioning result. This greatly reduces the probability of obtaining an erroneous positioning result due to positioning drift.

[0188] In one embodiment of the present case, before executing the present case to determine whether the target corresponding to the first ID appears in the first detection area, a method for automatically associating the first ID and the first face is also included. In some scenarios, the objects entering the field of view of the image device may include one or more. In the case where there is only one object within the field of view of the image device, it is considered that the positioning device closest to the image device is the positioning device carried by the object. When there are multiple target positioning devices entering the image device, the correspondence between the positioning device and the carried object can be determined based on the running trajectory of the target positioning device entering the field of view of the image device. Furthermore, when the running trajectory of the field of view of the image device is consistent, the correspondence between the target positioning device and the carried object can be determined by the distance between the target positioning device and the image device, or the correspondence between the target positioning device and the carrying person (face) can be determined by the time when the target positioning device enters the image device.

[0189] Based on the method of the above embodiment, the first inspection area is an arbitrary inspection area, that is, the inspection area is not specified, and the target is considered to be present at the roll call only if it appears in front of any camera linked to the positioning system. In one embodiment, for the target that has not been successfully inspected in the target inspection result, that is, the target marked as "not arrived", the position of the target that has not been inspected can be determined by video linkage. Specifically, it also includes:

[0190] Acquire a third target location area of ​​the positioning device corresponding to the target that was not successfully checked in the target checking result;

[0191] Determine a third target image device by using the position of the positioning device of the target that has not been successfully checked;

[0192] Detecting whether the target that has not been successfully checked is within the detection range of the third imaging device;

[0193] If the target that has not been successfully inspected is within the detection range of the third target image device, the third target location area of ​​the positioning device is used as the location area of ​​the target that has not been successfully inspected. In some embodiments, the video corresponding to the location area can also be provided to the server.

[0194] The present invention includes a testing device, such as Figure 6 As shown, for implementing the method provided by the present invention, the apparatus includes positioning nodes 151-153, a positioning base station 131, a first image acquisition device 141, and a server 121;

[0195] The first image acquisition device is used to provide a first video stream to the server;

[0196] The server includes one or more memories, one or more processors, and one or more modules. The one or more modules are stored in the memory and configured to be executed by the one or more processors. The server should be understood as a general term for computers with computing capabilities, and should not be understood only as a physically independent server hardware, that is, the server can be an independent server hardware, or it can include at least two physically independent server hardware, and different server hardwares have different program modules for respectively executing some steps of the present invention. The server includes one or more modules, which include or include respectively executing the following steps or methods:

[0197] Solve the positioning signal sent by the positioning node to the positioning base station;

[0198] Instructions for the aforementioned verification methods.

[0199] In some embodiments, the apparatus further comprises a second image acquisition device, the second image acquisition device being configured to provide a second video stream to the server;

[0200] The one or more modules of the server include or respectively include for executing a verification method, the method comprising:

[0201] Determining whether the first human face and the first biometric feature are detected in a second video stream, wherein the area captured by the second video stream at least includes the first detection area;

[0202] If the location corresponding to the first ID is within the first detection area,

[0203] If the first biometric feature of the first face is detected in the second video stream, it is determined that the target corresponding to the first ID is located within the first detection area.

[0204] In some embodiments, the angle between the first image acquisition device and the second image acquisition device is greater than 5 degrees.

[0205] In some embodiments, the positioning node, positioning base station, and server constitute a positioning system, and the one or more modules include instructions for executing the following steps: determining whether the positioning corresponding to the first ID is within the first detection area based on the one-dimensional UWB positioning coordinates.

[0206] In some embodiments, the positioning node includes a heart rate detection circuit and an anti-tampering circuit, and the server receives heart rate detection data and an anti-tampering circuit status signal;

[0207] The one or more modules of the server include or respectively include for executing the following method, the verification method includes:

[0208] obtaining the status of the positioning device corresponding to the first ID,

[0209] If the status of the positioning device corresponding to the first ID is wearing status,

[0210] If the location corresponding to the first ID is within the first detection area, and the first face and the first biometric feature are detected in the first video stream, it is determined that the target corresponding to the first ID is within the first detection area.

[0211] In some embodiments, the positioning device is a UWB positioning device;

[0212] The one or more modules of the server include or respectively include for executing a verification method, the method comprising:

[0213] The first biometric feature is a human face;

[0214] If it is determined that the target corresponding to the first ID does not appear in the first detection area, sending an alarm message;

[0215] If it is determined that the target corresponding to the first ID does not appear in the first detection area, obtaining a video corresponding to the positioning according to the positioning corresponding to the first ID;

[0216] If it is determined that the target corresponding to the first ID appears in the first detection area, the first video stream is started to be collected.

[0217] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0218] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A verification method, characterized in that: include: Receiving a first instruction, wherein the first instruction indicates a first set of IDs that need to be named; If the target corresponding to the first ID is within the first detection area, the first ID in the first ID set is deleted, and the next step is performed to determine the remaining first IDs in the first ID set. Whether the corresponding target appears in the first detection area, and the loop is executed until a preset time; Determine a list of absent persons based on the list of persons not detected in the first detection area; Wherein, if the location corresponding to the first ID is within the first detection area, and the first biometric feature is detected in the first video stream, it is determined that the target corresponding to the first ID is within the first detection area; in, The first ID has a corresponding relationship with the first biometric feature; The first video stream is a video stream generated by shooting the first detection area; The target carries a positioning device corresponding to the first ID; Among them, the positioning method is short-range wireless positioning, including: Determine at least one area corresponding to each target, and determine the confidence between any two areas; for each target, perform multiple positioning detections on each target, and determine the confidence of each area in at least one preset area through the positioning results of the multiple positioning detections and the transfer coefficients of any two preset areas; Selecting the location area with the highest confidence as the positioning result corresponding to the target; Among them, for each of the targets, multiple positioning detections are performed on each of the targets, and the confidence of each area in at least one preset area is determined by the positioning results of the multiple positioning detections and the transfer coefficients of any two preset areas, including: For each positioning detection, based on the transfer coefficients of the positioning area corresponding to the current positioning detection and the positioning area corresponding to the previous positioning detection, the confidence of the positioning area corresponding to the current positioning detection and the confidence of other areas are adjusted respectively.

2. The method according to claim 1, characterized in that The determining that the target corresponding to the first ID is within the first detection area includes: If the location corresponding to the first ID is within the first detection area, It is then detected whether the first biometric feature appears in the first video stream.

3. The method according to claim 1, characterized in that The determining that the target corresponding to the first ID is within the first detection area includes: Obtaining a first ID set located within the first detection area, Obtaining a corresponding face set according to the first ID set; Determining whether the face identified in the first video stream matches any face in the face set; If there is a match, the first biometric feature is detected in the first video stream.

4. The method according to claim 1, characterized in that The determining that the target corresponding to the first ID is within the first detection area includes: Determining whether the first biometric feature is detected in a second video stream, the area captured by the second video stream at least including the first detection area; If the location corresponding to the first ID is within the first detection area, If the first biometric feature is detected in the second video stream, it is determined that the target corresponding to the first ID is located within the first detection area.

5. The method according to claim 4, characterized in that The shooting angle between the first video stream and the second video stream is greater than 5 degrees.

6. The method according to claim 1, characterized in that The first detection area is a one-way entrance and exit area.

7. The method according to claim 1, characterized in that The determining that the target corresponding to the first ID is within the first detection area includes: The one-dimensional UWB positioning coordinates are used to determine whether the positioning corresponding to the first ID is within the first detection area.

8. The method according to claim 1, characterized in that: The determining that the target corresponding to the first ID is within the first detection area includes: obtaining the status of the positioning device corresponding to the first ID, If the status of the positioning device corresponding to the first ID is wearing status, If the location corresponding to the first ID is within the first detection area and the first biological feature is detected in the first video stream, it is determined that the target corresponding to the first ID is within the first detection area.

9. The method according to claim 8, characterized in that The obtaining the state of the positioning device corresponding to the first ID includes: Detect the heart rate of the target by using the positioning device, and / or detecting the state of the anti-disassembly circuit of the positioning device; If the heart rate of the target is detected and / or the anti-disassembly circuit indicates that it is not disassembled, the state of the positioning device corresponding to the first ID is the wearing state.

10. The method according to claim 1, characterized in that The positioning device is a UWB positioning device; The method includes: the first biometric feature is a human face; And / or the method includes: if it is determined that the target corresponding to the first ID does not appear in the first detection area, sending an alarm message; And / or the method includes: if it is determined that the target corresponding to the first ID does not appear in the first detection area, obtaining a video corresponding to the positioning according to the positioning corresponding to the first ID; And / or the method includes: if it is determined that the target corresponding to the first ID appears in the first detection area, starting to collect the first video stream.

11. A testing device, characterized in that: include: Positioning node, positioning base station, first image acquisition device and server; The first image acquisition device is used to provide a first video stream to the server; The server includes one or more memories, one or more processors, and one or more modules, wherein the one or more modules are stored in the memories and configured to be executed by the one or more processors, and the one or more modules include instructions for or respectively for executing the following steps: Solve the positioning signal sent by the positioning node to the positioning base station; And, instructions corresponding to any one of the methods described in claims 1 to 3.

12. The device according to claim 11, characterized in that Also included is a second image acquisition device, the second image acquisition device is used to provide a second video stream to the server; the one or more modules include instructions for or respectively for performing the following steps: The instructions corresponding to the method of claim 4.

13. The device according to claim 12, characterized in that The included angle between the first image acquisition device and the second image acquisition device is greater than 5 degrees.

14. The device according to claim 11, characterized in that The one or more modules include instructions for or respectively for executing the following steps: determining whether the location corresponding to the first ID is within the first detection area according to the one-dimensional UWB positioning coordinates.

15. The device according to claim 11, characterized in that The positioning node includes a heart rate detection circuit and an anti-disassembly circuit, and the server receives heart rate detection data and an anti-disassembly circuit status signal; The one or more modules include instructions for executing the following steps: instructions corresponding to the method described in claim 8.

16. The device according to claim 11, characterized in that The positioning device is a UWB positioning device; The server is also used to execute the method of claim 10.

17. The device according to claim 11, characterized in that The server includes at least two servers, and the two servers respectively include partial modules of the multiple modules.

Images