Target object detection method, device, electronic device and storage medium

Abstract

The present disclosure provides a method, device, electronic device and storage medium for detecting target objects, and relates to the field of data processing technology, in particular to the technical fields of computer vision, image processing, artificial intelligence and the like. The specific implementation scheme is as follows: based on the surveillance video of the target area, the number of existing target objects in the target area is detected; when it is determined that the number of target objects in the target area has decreased based on the number of existing target objects, the number of missing target objects is determined based on the number of existing target objects and the number of target objects that should exist in the target area that is pre-stored; based on the surveillance video of the target area, suspected missing objects are determined to obtain a set of suspected missing objects; based on the missing amount, the actual missing objects are determined from the set of suspected missing objects. In the embodiment of the present disclosure, when the number of target objects has decreased, the detection of the actual missing objects can be triggered, and the actual missing objects can be further determined based on the missing amount of the target objects.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to technical fields such as computer vision, image processing, and artificial intelligence. Background Art

[0002] Monitoring is a crucial aspect of daily production and operations. Currently, a large number of cameras are installed to monitor key targets for easier management. However, the challenge of automatically analyzing missing objects based on surveillance video remains unresolved. Summary of the Invention

[0003] The present disclosure provides a method, device, electronic device, and storage medium for detecting a target object.

[0004] According to one aspect of the present disclosure, a method for detecting a target object is provided, comprising:

[0005] Based on the surveillance video of the target area, detect the number of existing target objects in the target area;

[0006] In a case where it is determined that the number of target objects in the target area is reduced based on the number of existing target objects, determining a missing amount of target objects based on the number of existing target objects and a pre-stored number of target objects that should exist in the target area;

[0007] Based on the surveillance video of the target area, suspected missing objects are determined to obtain a set of suspected missing objects;

[0008] Based on the missing amount of the target object, the actual missing object is determined from the set of suspected missing objects.

[0009] According to another aspect of the present disclosure, there is provided a device for detecting a target object, comprising:

[0010] A detection module is used to detect the number of existing target objects in the target area based on the surveillance video of the target area;

[0011] a target object number determination module, configured to determine a missing amount of target objects based on the number of existing target objects and a pre-stored number of target objects that should exist in the target area, when it is determined that the number of target objects in the target area is reduced based on the number of existing target objects;

[0012] A first determination module is configured to determine suspected missing objects based on the surveillance video of the target area and obtain a set of suspected missing objects;

[0013] The second determining module is configured to determine the actual missing object from the set of suspected missing objects based on the missing amount of the target object.

[0014] According to another aspect of the present disclosure, there is provided an electronic device, comprising:

[0015] at least one processor; and

[0016] a memory communicatively connected to the at least one processor; wherein,

[0017] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method of any embodiment of the present disclosure.

[0018] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method according to any embodiment of the present disclosure.

[0019] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the method according to any embodiment of the present disclosure when executed by a processor.

[0020] In the embodiment of the present disclosure, it is achieved to focus on the overall situation of the target area first. Only a small number of surveillance cameras are required to be arranged in the same target area, and there is no need to configure a separate camera near each target object, thereby reducing the number of cameras. Compared with focusing on the specific information of each object (such as the movement trajectory), the solution of focusing on the number of target objects is relatively simple and easy to implement, and when the number of target objects decreases, the detection of the actual missing objects is triggered, and it can also be ensured that when the target object leaves the target area, the actual missing object can be further determined based on the missing amount of the target object. In addition, the solution proposed in the embodiment of the present disclosure can be applied to scenarios such as parks, factory areas, and corporate office buildings, and has universality.

[0021] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0023] Figure 1a is a schematic diagram of an application scenario of a target object detection method according to an embodiment of the present disclosure;

[0024] Figure 1b is a flowchart of a method for detecting a target object according to an embodiment of the present disclosure;

[0025] Figure 2is a schematic diagram of a scene of objects in other working areas according to another embodiment of the present disclosure;

[0026] Figure 3 is a schematic diagram of obtaining missing probability according to another embodiment of the present disclosure;

[0027] Figure 4 is a schematic structural diagram of a target object detection device according to another embodiment of the present disclosure;

[0028] Figure 5 is a schematic structural diagram of a target object detection device according to another embodiment of the present disclosure;

[0029] Figure 6 It is a block diagram of an electronic device used to implement the target object detection method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0031] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout the present disclosure, "plurality" means two or more, unless otherwise specifically defined.

[0032] In related technologies, a camera is installed near a target object, for example, in front of the target object. The camera then processes the key features of the target object to determine whether the target object is within the target area. This approach requires the purchase of a large amount of hardware and is not suitable for large-scale deployment.

[0033] In view of this, a method for detecting a target object is proposed in an embodiment of the present disclosure. Figure 1a The diagram shows an application scenario applicable to the embodiment of the present disclosure. The application scenario includes multiple areas, including area 1...area n, and each area includes at least one target object. Figure 1aEach five-pointed star in the figure represents a target object. The target object in the embodiment of the present disclosure may be a staff member in a key position. The key position is, for example, a position involving personal or property safety. Once the staff member of this position leaves, it may cause major safety hazards. Alternatively, the key position is a position with high precision requirements in the production process. Once the staff member of this position leaves the position, the precision of key core components may not meet the requirements. Of course, it should be noted that the embodiment of the present disclosure is not only applicable to the aforementioned key positions, but also to any position in need.

[0034] like Figure 1a As shown, each monitoring area includes at least one monitoring device 11 , and all monitoring devices 11 establish communication connections with electronic devices 12 through a network. Terminal devices 13 also establish communication connections with electronic devices 12 through a network.

[0035] The network used to establish the communication connection may be a wireless or wired network.

[0036] The monitoring device 11 is used to obtain a video stream of a corresponding area and upload it to the electronic device 12 so that the electronic device 12 can detect the target object.

[0037] The electronic device 12 may be a server, a server cluster consisting of several servers, or a cloud computing center. The electronic device may provide target object detection services for one unit or multiple units.

[0038] The terminal device 13 includes but is not limited to electronic devices such as desktop computers, mobile phones, mobile computers, tablet computers, media players, smart wearable devices, and smart TVs.

[0039] In the embodiment of the present disclosure, the processing method for each monitoring area is the same. Taking one monitoring area as the target area as an example, the detection method of the target object in the embodiment of the present disclosure is described. Figure 1b FIG. 5 is a flow chart of the method, comprising:

[0040] S101 , based on a surveillance video of a target area, detecting the number of existing target objects in the target area.

[0041] In some embodiments, the number of target objects in the target area may be determined based on a human body tracking technology of the target object, or may be determined based on a technology for detecting key features of key parts, such as a face, and key features such as facial features.

[0042] For some key positions, it is necessary to ensure that the staff in that position are on duty in real time. The target objects are the staff in that position. Because each staff member has different facial features, face detection can be performed on that staff member, and the number of target objects in the target area can be determined. Of course, the disclosed embodiments do not limit the method for detecting the number of target objects in the target area.

[0043] Taking the example of determining the number of target objects in a target area based on the target object's outline features, the outline features of the target objects in the target area are detected to obtain the UID (User Identification) of each object in the target area. By counting the number of UIDs in the target area, the number of target objects existing in the target area can be determined. For example, taking the example of different staff members having different body shapes, the outline features can be human body features determined based on human body tracking technology for the target objects.

[0044] S102 : When it is determined that the number of target objects in the target area is reduced based on the number of existing target objects, determine the missing amount of target objects based on the number of existing target objects and the pre-stored number of target objects that should exist in the target area.

[0045] The number of target objects that should be present can be determined based on the duty schedule and time period. For example, in the duty schedule, the target objects that should be present in area A from 8:00 AM to 12:00 PM are object A, object B, and object C. The target objects that should be present in area A from 12:00 PM to 4:00 PM are object D, object E, object F, and object G. At 9:00 AM, the number of target objects that should be present at that time is 3. If the number of detected existing target objects is 2, the difference between the detected number and the pre-stored number of target objects that should be present is calculated, and the number of missing target objects is determined to be 1.

[0046] S103: Determine suspected missing objects based on the surveillance video of the target area to obtain a set of suspected missing objects.

[0047] S104: Based on the missing amount of the target object, determine the actual missing object from the set of suspected missing objects.

[0048] The actual missing objects are determined from the set of suspected missing objects with the missing amount of the target object as the upper limit, that is, the total number of actual missing objects does not exceed the missing amount of the target object.

[0049] In the disclosed embodiments, the number of existing target objects can be analyzed based on the surveillance video of the target area. If the number of existing target objects decreases, it can be determined that a target object has left the target area, and suspected missing objects can be analyzed from the surveillance video. To accurately determine which target object has left the target area, the actual missing object can be determined from the set of suspected missing objects based on the number of missing target objects. The entire process achieves the goal of first focusing on the overall situation of the target area, that is, first focusing on the number of existing target objects in the target area, rather than focusing on the specific characteristics of each object in the target area. In practice, only a small number or even a single surveillance camera can be deployed in the same target area, eliminating the need to configure a separate camera on the target object, thereby reducing the number of cameras. Compared to focusing on the specific information of each target object (such as its movement trajectory), the solution of focusing on the number of target objects is relatively simple and easy to implement. When the number of target objects decreases, it triggers the detection of the actual missing object. It also ensures that if a target object leaves the target area, the actual missing object can be further determined based on the number of missing target objects. In addition, the solution proposed in the disclosed embodiments is applicable to scenarios such as parks, factories, and corporate office buildings, and has universality.

[0050] In the embodiment of the present disclosure, the detection process of the target object can be divided into a process of determining a suspected missing object in S103 and a process of determining an actual missing object in S104. The two processes are described in detail below:

[0051] 1) Identify suspected missing objects

[0052] In the embodiment of the present disclosure, the suspected missing object may be determined based on the outline features of the existing target object, or the existing target object may be determined first and then the suspected missing object may be determined.

[0053] Solution 1: Determine the suspected missing object based on the outline features of the existing target object

[0054] In some embodiments, determining a suspected missing object based on surveillance video of a target area may be implemented as follows:

[0055] Step A1: Taking a target time point as a reference, a target video segment of a target duration is captured from the surveillance video of a target area; the target time point is the time point when the number of target objects in the target area decreases.

[0056] The time range covered by the target duration includes the time before the target time point and the time after the target time point, so that the target video segment can include the scenes before and after the actual missing object leaves the target area.

[0057] Since the target area is monitored in real time, the time point at which the number of target objects in the target area decreases is determined as the target time point. Taking the target time point as 9:00:30 as an example, the duration 20 seconds before and 20 seconds after the target time point is taken as the target duration. Therefore, 9:00:10-9:00:50 can be used as the target duration. The video clip within this target duration is captured from the surveillance video of the target area as the target video clip.

[0058] It should be noted that the target duration can also be 30 seconds before and after the target time point, or 10 seconds before and 20 seconds after the target time point. The target duration needs to be determined based on actual conditions, and the embodiments of the present disclosure do not limit this.

[0059] Step A2: determining the missing object features based on the tracking result of the target object in the target video clip.

[0060] The tracking result of the target object may be an outline feature of the target object.

[0061] It should be noted that when the tracking method is not used to detect the number of existing target objects in the target area, the human body tracking method can be used to detect the existing target objects on the target video clip, determine the UID list of the disappeared objects, and then obtain the object features corresponding to the UID of the disappeared objects (such as the human body features of the staff).

[0062] When the existing target object tracking method is used to detect the number of existing target objects in the target area, the tracking result of the target object of the target video clip can be reused to determine the object feature corresponding to the UID of the disappeared object.

[0063] The object features corresponding to the UID of the disappeared object can be extracted from the video image containing the UID, or the shape contour features corresponding to the UID can be obtained from a pre-stored shape contour feature library. Specifically, since the target object needs to punch in before entering the target area, when the punch card machine detects that the target object is punching in, the key features of the target object can be identified from the surveillance video of the target area to determine the position of the target object in the video image, and then the shape contour features of the target object are obtained based on the position and stored in the shape contour feature library. Afterwards, the shape contour features corresponding to the UID are extracted from the surveillance video image, and the shape contour features are compared with the shape contour features in the shape contour feature library to obtain the shape contour features corresponding to the UID in the shape contour feature library.

[0064] In the embodiment of the present disclosure, a correspondence between outline features and objects can be pre-stored and stored in an outline feature library, thereby determining the object corresponding to the missing outline feature. Then, in step A3, the object corresponding to the missing outline feature is determined as a suspected missing object.

[0065] In the embodiment of the present disclosure, based on the time point when the number of target objects decreases, a target video segment of the target duration is captured from the surveillance video of the target area, and the disappearing shape contour features are sorted out based on the tracking results of the target objects in the target video segment, so that the suspected missing objects can be accurately obtained, thereby improving the accuracy of target object detection.

[0066] Solution 2: Identify suspected missing objects based on existing target objects

[0067] In some embodiments, it can be implemented as follows:

[0068] Step B1: Obtain a video stream captured after a target time point, which is the time point at which the number of target objects is determined to decrease.

[0069] Step B2: Identify the object identifiers of existing target objects in the video stream to obtain a set of existing target objects.

[0070] The object identifier may be the real name information of the existing target object, or the object number of the existing target object. Any information that can determine the identity of the existing target object can be used as the object identifier.

[0071] In some embodiments, identifying the object identification of an existing target object in a video stream can be implemented as follows: detecting key features of key parts of the existing target object, which key features can uniquely identify the existing target object; matching the key features of the existing target object with the key features of a known object to obtain the object identification of the existing target object.

[0072] In some embodiments, information about a target object (i.e., a known object) may be pre-stored. This information may include the real name, image, communication method, and other information of the target object. After obtaining key features of the existing target object, a match is performed between the key features and the key features of the known object. If a match is successful, the object identifier of the existing target object is obtained.

[0073] In the disclosed embodiment, based on matching the key features of the existing target object with the key features of the known object, the object identification of the existing target object can be accurately obtained, providing a basis for subsequent determination of the actual missing object.

[0074] In some embodiments, in addition to using the key features of an existing target object to determine the object identification, the object identification of the target object can also be obtained based on the shape contour features of the existing target object. This method can be implemented as follows: detecting the shape contour features of the existing target object; matching the shape contour features of the existing target object with the shape contour features of a known object to obtain the object identification of the existing target object.

[0075] In some embodiments, it may also be pre-stored information about a target object (i.e., a known object). This information may also include appearance and contour features. Since the target object needs to clock in before entering the target area, when the clock-in machine detects that the target object is clocking in, the appearance and contour features of the working object today can be obtained and stored. The appearance and contour features may include, for example, posture, gait, etc., and may also include the clothing features of the target object. Based on the appearance and contour features of the target object, the appearance and contour features are matched with the appearance and contour features of the known object. If the match is successful, the object identification of the target object is obtained.

[0076] In the disclosed embodiment, the object identification of the target object can be accurately obtained based on matching the outer contour features of the target object with the outer contour features of the known object, which provides a basis for subsequent determination of the actual missing object.

[0077] In short, in the disclosed embodiments, a key feature library and a contour feature library for known objects can be established. As shown in Table 1, the key feature library includes the object identifiers and corresponding key features of known objects, while the contour feature library includes the object identifiers and corresponding contour features of known objects. Therefore, by matching the key feature library with the contour feature library, the corresponding object identifier can be determined.

[0078] Table 1

[0079]

[0080]

[0081] Step B3: Determine the difference between the set of target objects that should exist and the set of existing target objects to obtain suspected missing objects.

[0082] The set of target objects that should exist includes the target objects that should currently exist within the target area. It should be noted that because the methods for obtaining the key features and outline features of the target objects have certain errors, if a suspected missing object determined by the tracking method is included in the set of existing target objects, the suspected missing object will be deleted from the set of existing target objects. In other words, in order to accurately determine the actual missing object, suspected missing objects determined by any method will be included in the set of suspected missing objects.

[0083] In the disclosed embodiment, based on the object identifiers of existing target objects, a set of existing target objects can be obtained. Based on the difference between the set of expected target objects and the set of existing target objects, suspected missing objects can be obtained. By combining the number of existing target objects with the number of expected target objects, the suspected missing objects determined in this way are accurate.

[0084] 2) Determine the actual missing objects

[0085] In the embodiment of the present disclosure, the actual missing object may be determined based on the object conditions in other working areas, or the actual missing object may be determined based on the missing probability of the suspected missing object.

[0086] Solution 1: Determine the actual missing objects based on the objects in other work areas

[0087] In some embodiments, based on the missing amount of the target object, determining the actual missing object from the set of suspected missing objects can be implemented as follows:

[0088] Step C1: Acquire a set of remote objects, including object identifiers in other work areas and / or object identifiers that have left the target area through an access control device.

[0089] The remote object set includes object identifiers that should exist in other work areas, and may also include object identifiers of non-working objects in other work areas.

[0090] For example, if Figure 2 As shown, the target area for key positions originally had four target objects. However, object D, who was working in the target area, left the target area and moved into another work area. Since D was not a work object in the other work area, object D became a non-work object in the other work area. At this point, the number of objects in the target area has dropped to three, and the number of objects in the other work areas has increased from three to four.

[0091] The object identification of non-working objects in other work areas can be obtained based on the key features of the objects. First, based on the monitoring of other work areas, a video stream of the other work areas is obtained. The key features of the key parts of the existing target objects are detected in the video stream. The detected key features are matched with the key features of the target objects in other work areas. If a match is not successful, the object identification of the non-working objects in the other work areas can be obtained.

[0092] Of course, the object identification of the non-working object can also be obtained based on the outline features of the object. The determination method is similar to the method of determining the object identification of the object based on the key features mentioned above, and will not be described in detail here.

[0093] When entering and exiting the target area requires passing through an access control device, the object information of the target object will basically be entered through the access control device. In this way, the access control device can be linked to obtain the target object that leaves the target area through the access control device, and then the object identifier corresponding to the target object is obtained and added to the remote object collection.

[0094] Step C2: if any suspected missing object is included in the remote object set, the suspected missing object is determined as an actual missing object.

[0095] Step C3: When the total number of actual missing objects is equal to the missing amount of the target objects, determine that all actual missing objects are identified.

[0096] In the embodiment of the present disclosure, considering that the target object may appear in a different place (i.e., non-target area), whether the suspected missing object is the actual missing object is determined based on the monitoring and access control equipment in the different place. Based on this method, the object identification of the actual missing object can be accurately determined.

[0097] In addition, based on the off-site object set, the location where the actual missing object appears can also be determined.

[0098] In some embodiments, there may be a situation where the total number of actual missing objects is less than the missing amount of target objects. In this case, an alarm message may be output to instruct to check the objects that have not been confirmed to have left the target area.

[0099] In some embodiments, for example, using the alarm information in the form of a list, for unconfirmed objects leaving the target area, all unconfirmed suspected missing objects can be consolidated into a list and sent to a management staff member to facilitate manual correction of the actual missing objects leaving the target area. Other forms are also possible, such as outputting images of the unconfirmed suspected missing objects, and the present disclosure is not limited thereto.

[0100] Since the detection of key features of key parts of existing target objects may contain errors during the actual detection process, it is possible that the actual missing object may appear in other working areas but not be detected. Therefore, in order to ensure the accuracy of the actual missing objects, when the total number of actual missing objects is less than the number of missing target objects, the difference between the set of suspected missing objects and the set of actual missing objects can be determined to obtain the objects to be verified; the shape contour features of the objects to be verified are matched with the set of foreign shape contour features. If the shape contour features of any object to be verified are included in the set of foreign shape contour features, any object to be verified is determined to be an actual missing object.

[0101] The off-site outline feature set includes outline features of candidate objects in other areas, but the object identifier of the candidate object cannot be identified. In other words, the candidate object also appears in the off-site surveillance video, but because its object identifier cannot be identified, the candidate object is not included in the off-site object set.

[0102] When the total number of actual missing objects is less than the number of missing target objects, the outline features of the suspected missing objects that have not been determined to be actual missing objects are calculated to be similar to the outline features in the foreign outline feature set. The cosine similarity can be used to calculate the similarity between the two. When the similarity between the two is greater than a preset threshold, the suspected missing object is determined to be the actual missing object. For example, the preset threshold can be 90%, which can be determined based on actual conditions and is not limited in the embodiments of the present disclosure.

[0103] The similarity between the two may also be calculated using Euclidean distance, Pearson correlation coefficient, Hamming distance, Manhattan distance, etc., which is not limited in the present embodiment.

[0104] In the embodiment of the present disclosure, taking into account the situation that the actual missing object may not be included in the off-site object set, the shape contour features of the suspected missing object and the off-site shape contour feature set are matched, and it is also possible to accurately identify whether the suspected missing object is the actual missing object based on the shape contour features. This method of collecting off-site object sets can comprehensively consider the situation of actual missing objects and can improve the accuracy of target object detection.

[0105] If the total number of missing objects actually obtained based on the set of foreign objects and the set of foreign outline features is still less than the number of missing target objects, an alarm message may be output to indicate that the suspected missing objects have not been verified. The format of the alarm message is the same as described above and will not be described in detail in this embodiment.

[0106] In the embodiment of the present disclosure, in order to facilitate management personnel to manage the target object, the target object can be notified in a timely manner using alarm information so that the management system can be improved.

[0107] Solution 2: Determine the actual missing objects based on the missing probability

[0108] Among them, data analysis and mining can be performed on multiple historical missing objects to build a missing object model. The model includes multiple reference features, such as Figure 3 As shown, the missing time, missing location, missing object, behavior of leaving the target area, and reasonable behavior within the target area in the historical information are modeled, so as to analyze and mine the behavioral feature set of leaving the target area, the whitelist behavioral feature set, the missing time period set of each object in the regional dimension, and the missing number set of each object in the regional dimension.

[0109] For each historically missing object, we can analyze the video footage for a target duration around a target time point to determine the missing object's behavior before leaving the target area. For example, if the target object disappears from the video after collecting their personal belongings, this behavior can be identified as leaving the target area, and the behavior feature extraction model can be used to extract the departure behavior.

[0110] The behavioral feature extraction model is used to extract behavioral features from videos of historically missing objects. These extracted behavioral features may not necessarily indicate leaving the target area, so whitelisted behaviors can be interpreted as legitimate behaviors within the target area. If the object does not disappear from the video after the behavior occurs, it can be confirmed as a whitelisted behavior, and whitelisted behavior features can be extracted.

[0111] Finally, these reference features obtained constitute the feature set of historically missing objects.

[0112] Accordingly, based on the missing amount of the target object, determining the actual missing object from the set of suspected missing objects can be implemented as follows:

[0113] Step D1: Obtain the feature set of the historically missing object.

[0114] Step D2 is performed for each suspected missing object in the suspected missing object set: determining feature information of the suspected missing object based on the feature set, matching the feature information of the suspected missing object with the feature set, and determining the missing probability of the suspected missing object based on the matching result.

[0115] In some embodiments, since the feature set includes multiple reference features, each with a corresponding initial probability, determining the missing probability of a suspected missing object based on the matching results can be implemented by: obtaining the initial probability corresponding to the matched reference feature; and determining the missing probability of the suspected missing object based on the influence rule of the initial probability corresponding to the reference feature on the missing probability. In this influence rule, if the reference feature is a feature that leaves the target area, the missing probability is positively correlated with the initial probability corresponding to the reference feature; if the reference feature is a feature of reasonable behavior within the target area, the missing probability is negatively correlated with the initial probability corresponding to the reference feature.

[0116] Continue with Figure 3 For example, the behavior to be matched of the suspected missing object is matched with the behavior feature set of leaving the target area to obtain a first probability; the behavior to be matched of the suspected missing object is matched with the whitelist behavior feature set to obtain a second probability; the missing time point of the suspected missing object is matched with the missing time period set of each object in the regional dimension. If the suspected missing object exists in the set, it is determined that the suspected missing object successfully matches the missing time point set; if the suspected missing object does not exist in the set, it is determined that the missing time point set is not successfully matched, thereby obtaining a third probability; the suspected missing object is matched with the missing number set of each object in the regional dimension. If the suspected missing object exists in the set, it is determined that the suspected missing object successfully matches the missing number set; if the suspected missing object does not exist in the set, it is determined that the missing number set is not successfully matched, thereby obtaining a fourth probability. Then, the missing probability of the suspected missing object is determined based on the first probability, the second probability, the third probability, and the fourth probability.

[0117] Taking the calculation of the first probability of a suspected missing object as an example, based on obtaining multiple behavioral features to be matched of the suspected missing object, the calculation method of the first probability is as shown in Expression (1):

[0118]

[0119] Among them, P(B) represents the first probability of the suspected missing object, P(B1) represents the initial probability of the first behavior of leaving the target area that the suspected missing object is successfully matched, f1 represents the weight of the first behavior of leaving the target area of the suspected missing object, P(B2) represents the initial probability of the second behavior of leaving the target area that the suspected missing object is successfully matched, f2 represents the weight of the second behavior of leaving the target area of the suspected missing object, and so on. P(B3),…,P(Bk), and the meanings of other expressions of f3,…,fk are similar to the above, and n represents that the behavior feature set of leaving the target area includes n behaviors of leaving the target area.

[0120] Taking the calculation of the second probability of the suspected missing object as an example, based on obtaining multiple behavioral features to be matched of the suspected missing object, the calculation method of the second probability is shown in Expression (2):

[0121]

[0122] Among them, P(WB) represents the second probability of the suspected missing object, P(WB1) represents the initial probability of the first whitelist behavior of the suspected missing object successfully matching, f1 represents the weight of the first whitelist behavior of the suspected missing object, P(WB2) represents the initial probability of the second whitelist behavior of the suspected missing object successfully matching, f2 represents the weight of the second whitelist behavior of the suspected missing object, and so on. P(WB3),…,P(WBq), and the meanings of other expressions of f3,…,fq are similar to the above, and w represents that the whitelist behavior feature set includes w whitelist behaviors.

[0123] The third probability can be obtained as follows: In the target area, object A was missing between 9:00 and 10:00, and object B was missing between 10:00 and 10:30. Object A is missing at 9:20, confirming that object A matches the missing time period set. Therefore, the initial probability of object A at the missing time period is the third probability.

[0124] Similarly, the fourth probability can be obtained as described in the following example: object A is missing 6 times, object B is missing 4 times, and object C is missing 8 times. If object C is detected to be missing, it can be determined that object C successfully matches the set of missing times, and the initial probability corresponding to its missing times is obtained as the fourth probability.

[0125] On the basis of obtaining the first probability, the second probability, the third probability and the fourth probability of the suspected missing object, these four probabilities may be added together to obtain the missing probability of the suspected missing object.

[0126] In order to further improve the accuracy of missing object detection, in the embodiment of the present disclosure, the missing probability of the historical missing object can also be obtained based on expression (3):

[0127] Pi=max(0,(P(B)+P(WB))+P(T)+P(N))(3)

[0128] Among them, Pi represents the i-th suspected missing object among the suspected missing objects, that is, any suspected missing object, P(B) represents the first probability of the i-th suspected missing object, P(WB) represents the second probability of the i-th suspected missing object, P(T) represents the third probability of the i-th suspected missing object, and P(N) represents the fourth probability of the i-th suspected missing object.

[0129] In the disclosed embodiments, by setting the corresponding initial probability for any reference feature, the missing probability of the suspected missing object can be obtained, thereby laying the data foundation for determining the actual missing object. The missing probability calculation method is simple and easy to implement, and can quickly and accurately determine the missing probability based on the features of each suspected missing object, thereby improving the accuracy and efficiency of target object detection.

[0130] In step D3, the suspected missing objects with missing amounts of the target object are screened out from the set of suspected missing objects in descending order of missing probabilities as actual missing objects.

[0131] For example, the missing probability of suspected missing object 1 is 50%, the missing probability of suspected missing object 2 is 30%, the missing probability of suspected missing object 3 is 65%, and the missing probability of suspected missing object 4 is 85%. The resulting sorting order of suspected missing objects is: suspected missing object 4 - suspected missing object 3 - suspected missing object 1 - suspected missing object 2. If the number of missing objects of the target object is 2, suspected missing objects 4 and 3 are determined to be the actual missing objects.

[0132] In the embodiment of the present disclosure, feature information of the suspected missing object is determined based on a feature set, and the feature information of the suspected missing object is matched with the feature set. The missing probability of the suspected missing object can be determined based on the matching result. The missing probability of the suspected missing object obtained based on this method is accurate, and the actual missing object obtained is relatively accurate.

[0133] In some embodiments, before screening out suspected missing objects with missing amounts of the target object from the set of suspected missing objects in descending order of missing probabilities, the method further includes:

[0134] Normalize the missing probability of each suspected missing object in the suspected missing object set so that the sum of the missing probabilities of all suspected missing objects is the target value.

[0135] Among them, taking the target value as 1 as an example, the normalization method can be shown as expression (4):

[0136]

[0137] Among them, the suspected missing object set includes m suspected missing objects, P1 represents the first suspected missing object, P2 represents the second suspected missing object, and so on. The meanings of P3, P4, ..., Pm are similar to the above and are not repeated here.

[0138] The missing probabilities of the m suspected missing objects are normalized by expression (4) so that the sum of the missing probabilities of the m suspected missing objects is a target value, which may be 1.

[0139] In the disclosed embodiment, the data is normalized, and subsequent processing can be performed based on the normalized data, thereby alleviating jumps between data and improving the accuracy of the missing probability of each suspected missing object.

[0140] In some embodiments, before the missing duration of the actual missing object reaches a preset duration, if there is a returned object, the key features of the returned object are compared with the key features of the actual missing object. If it is determined that they are not the same object, it is determined that a replacement behavior has occurred.

[0141] After comparing the key features of the returned object with the key features of the actually missing object and confirming that they are the same object, the actually missing object is returned to the job.

[0142] In the embodiment of the present disclosure, when a missing object returns to work, the replacement behavior can be automatically detected by comparing the key features of the returned object with the key features of the actual missing object.

[0143] In some embodiments, if the missing person has been missing for a preset period of time, an alarm can be issued. The alarm information can be sent to the management object so that the management personnel can record the behavior of the missing person. The alarm information can also be sent to the missing person to remind them to return as soon as possible to avoid causing a safety accident. The alarm method can use SMS, APP (Application) notification, speaker, sound and light alarm, etc.

[0144] In this disclosed embodiment, using personnel in key positions as target objects, cameras are installed in the work areas of these positions. Human tracking is performed based on the video footage captured by the cameras to determine the number of personnel in these positions. If the number of personnel in these positions decreases, detection of target objects is initiated to determine which target objects have actually left the target area. This allows for monitoring the presence of personnel in these key positions and effectively mitigates potential safety hazards.

[0145] Based on the same technical concept, the embodiment of the present disclosure also provides a target object detection device 400, such as Figure 4 As shown, the device includes:

[0146] A detection module 401 is configured to detect the number of target objects existing in a target area based on a surveillance video of the target area;

[0147] A target object number determining module 402 is configured to determine a missing amount of target objects based on the number of existing target objects and a pre-stored number of target objects that should exist in the target area, when it is determined that the number of target objects in the target area is reduced based on the number of existing target objects;

[0148] A first determining module 403 is configured to determine suspected missing objects based on the surveillance video of the target area and obtain a set of suspected missing objects;

[0149] The second determining module 404 is configured to determine an actual missing object from the set of suspected missing objects based on the missing amount of the target object.

[0150] In some embodiments, the first determining module is configured to:

[0151] Taking the target time point as a benchmark, a target video segment of a target duration is captured from the surveillance video of the target area; the target time point is the time point when the number of target objects in the target area decreases;

[0152] determining features of the disappeared object based on a tracking result of the target object in the target video clip;

[0153] The object corresponding to the disappeared object feature is determined as a suspected missing object.

[0154] In some embodiments, Figure 4 On the basis of Figure 5 As shown, it also includes an identification module 501, which is used to:

[0155] Get the video stream collected after the target time point;

[0156] Identify the object identifiers of existing target objects in the video stream and obtain a set of existing target objects;

[0157] Determine the difference between the set of target objects that should exist and the set of existing target objects to obtain suspected missing objects.

[0158] In some embodiments, the identification module is configured to:

[0159] Detect key features of key parts of existing target objects, which can uniquely identify existing target objects;

[0160] The key features of the existing target object are matched with the key features of the known object to obtain the object identification of the existing target object.

[0161] In some embodiments, the identification module is further configured to:

[0162] Detect the contour features of existing target objects;

[0163] The shape contour features of the existing target object are matched with the shape contour features of the known object to obtain the object identification of the existing target object.

[0164] In some embodiments, the second determining module is configured to:

[0165] Obtaining a collection of remote objects; the collection of remote objects includes object identifiers in other areas and / or object identifiers that have left the target area through the access control device;

[0166] In the case where any suspected missing object is included in the off-site object set, any suspected missing object is determined to be an actual missing object;

[0167] In a case where the total number of actual missing objects is equal to the missing amount of the target object, it is determined that all actual missing objects are identified.

[0168] In some embodiments, the second determining module is further configured to:

[0169] When the total number of actual missing objects is less than the missing amount of the target object, the difference between the set of suspected missing objects and the set of actual missing objects is determined to obtain the objects to be verified;

[0170] The shape contour features of the object to be checked are matched with the off-site shape contour feature set. If the shape contour features of any object to be checked are included in the off-site shape contour feature set, it is determined that any object to be checked is an actual missing object. The off-site shape contour feature set includes the shape contour features of candidate objects in other areas, and the candidate object fails to identify the object identity.

[0171] In some embodiments, Figure 4 On the basis of Figure 5 As shown, the system further includes an alarm module 502 for:

[0172] When the total number of actual missing objects is less than the missing amount of target objects, an alarm message is output, and the alarm message is used to instruct to check the unconfirmed missing objects.

[0173] In some embodiments, the second determining module is configured to:

[0174] Get the feature set of historically missing objects;

[0175] For each suspected missing object in the suspected missing object set, respectively performing: determining feature information of the suspected missing object based on the feature set, matching the feature information of the suspected missing object with the feature set, and determining a missing probability of the suspected missing object based on the matching result;

[0176] According to the order of missing probability from high to low, the suspected missing objects with the missing amount of the target object are screened out from the set of suspected missing objects as the actual missing objects.

[0177] In some embodiments, the feature set includes multiple reference features, each reference feature has a corresponding initial probability, and the missing probability of the suspected missing object is determined based on the matching results. The feature set also includes a probability acquisition module 503 for:

[0178] Get the initial probability corresponding to the matched reference feature;

[0179] Determine the missing probability of the suspected missing object based on the influence rule of the initial probability corresponding to the reference feature on the missing probability;

[0180] In the impact rule: when the reference feature is the behavior feature of leaving the target area, the missing probability is positively correlated with the initial probability corresponding to the reference feature; when the reference feature is the reasonable behavior feature within the target area, the missing probability is negatively correlated with the initial probability corresponding to the reference feature.

[0181] In some embodiments, before screening out suspected missing objects with missing amounts of the target object from the set of suspected missing objects in descending order of missing probabilities, the method further includes:

[0182] Normalize the missing probability of each suspected missing object in the suspected missing object set so that the sum of the missing probabilities of all suspected missing objects is the target value.

[0183] In some embodiments, the identification module is also used to: before the missing duration of the actual missing object reaches a preset duration, if there is a returned object, compare the key features of the returned object with the key features of the actual missing object, and if it is determined that they are not the same object, determine that a replacement behavior has occurred.

[0184] For the description of specific functions and examples of each module and submodule of the device in the embodiment of the present disclosure, please refer to the relevant description of the corresponding steps in the above method embodiment, which will not be repeated here.

[0185] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0186] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0187] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0188] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for the operation of the device 600 can also be stored in the RAM 603. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0189] Various components in device 600 are connected to I / O interface 605, including an input unit 606, such as a keyboard, mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, optical disk, etc.; and a communication unit 609, such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows device 600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0190] The computing unit 601 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the target object detection method. For example, in some embodiments, the target object detection method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the target object detection method described above can be performed. Alternatively, in other embodiments, the computing unit 601 can be configured to perform the target object detection method in any other appropriate manner (e.g., by means of firmware).

[0191] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0192] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0193] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0194] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0195] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0196] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0197] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0198] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for detecting a target object, comprising: Detecting the number of existing target objects in the target area based on the surveillance video of the target area; When it is determined based on the number of existing target objects that the number of target objects in the target area is reduced, determining a missing amount of target objects based on the number of existing target objects and a pre-stored number of target objects that should exist in the target area; Determining suspected missing objects based on the surveillance video of the target area to obtain a set of suspected missing objects; Determining an actual missing object from the set of suspected missing objects based on the missing amount of the target object includes: Get the feature set of historically missing objects; For each suspected missing object in the set of suspected missing objects, respectively, the following steps are performed: determining feature information of the suspected missing object based on the feature set, matching the feature information of the suspected missing object with the feature set, and determining a missing probability of the suspected missing object according to the matching result, including: obtaining an initial probability corresponding to a matched reference feature; determining the missing probability of the suspected missing object based on an influence rule of the initial probability corresponding to the reference feature on the missing probability; wherein, in the influence rule, when the reference feature is a behavior feature of leaving a target area, the missing probability is positively correlated with the initial probability corresponding to the reference feature; and when the reference feature is a reasonable behavior feature within the target area, the missing probability is negatively correlated with the initial probability corresponding to the reference feature; In descending order of missing probability, the suspected missing objects having the missing amount of the target object are screened out from the set of suspected missing objects as actual missing objects.

2. The method according to claim 1, wherein The determining of a suspected missing object based on the surveillance video of the target area includes: Taking a target time point as a benchmark, intercepting a target video segment of a target duration from the surveillance video of the target area; the target time point is the time point when the number of target objects in the target area decreases; determining features of the disappeared object based on a tracking result of the target object in the target video clip; The object corresponding to the disappeared object feature is determined as a suspected missing object.

3. The method according to claim 2, further comprising: Obtaining a video stream collected after the target time point; Identifying object identifiers of existing target objects in the video stream to obtain a set of existing target objects; A difference between the target object set that should exist and the target object set that already exists is determined to obtain a suspected missing object.

4. The method according to claim 3, wherein: The identifying the object identifier of the target object existing in the video stream includes: Detecting key features of key parts of an existing target object, where the key features can uniquely identify the existing target object; The key features of the existing target object are matched with the key features of the known object to obtain the object identifier of the existing target object.

5. The method according to claim 3 or 4, wherein: The identifying the object identifier of the target object existing in the video stream further includes: Detect the contour features of existing target objects; The shape contour features of the existing target object are matched with the shape contour features of a known object to obtain an object identifier of the existing target object.

6. The method according to claim 1, wherein The determining of the actual missing object from the set of suspected missing objects based on the missing amount of the target object further includes: Acquire a collection of remote objects; the collection of remote objects includes object identifiers in other areas and / or object identifiers that have left the target area through an access control device; If any suspected missing object is included in the remote object set, determining the suspected missing object as an actual missing object; In a case where the total number of actual missing objects is equal to the missing amount of the target object, it is determined that all actual missing objects are identified.

7. The method according to claim 6, further comprising: When the total number of actual missing objects is less than the missing amount of the target object, determining the difference between the set of suspected missing objects and the set of actual missing objects to obtain the objects to be verified; The shape contour features of the object to be checked are matched with the off-site shape contour feature set. If the shape contour features of any object to be checked are included in the off-site shape contour feature set, the object to be checked is determined to be an actual missing object. The off-site shape contour feature set includes the shape contour features of candidate objects in other areas, and the object identification of the candidate objects cannot be identified.

8. The method according to claim 6 or 7, further comprising: In the case that the total number of actual missing objects is less than the missing amount of the target objects, an alarm message is output, where the alarm message is used to instruct to check the unconfirmed missing objects.

9. The method according to claim 1, wherein Before screening out suspected missing objects having a missing amount of the target object from the set of suspected missing objects in descending order of missing probabilities, the method further includes: The missing probability of each suspected missing object in the suspected missing object set is normalized so that the sum of the missing probabilities of all suspected missing objects is a target value.

10. The method according to claim 1, further comprising: If there is a returned object before the missing duration of the actual missing object reaches the preset duration, the key features of the returned object are compared with the key features of the actual missing object. If it is determined that they are not the same object, it is determined that a replacement behavior has occurred.

11. A device for detecting a target object, comprising: A detection module, configured to detect the number of target objects existing in the target area based on a surveillance video of the target area; a target object quantity determination module, configured to determine a missing amount of target objects based on the number of existing target objects and a pre-stored number of target objects that should exist in the target area, when it is determined based on the number of existing target objects that the number of target objects in the target area is reduced; A first determining module is configured to determine suspected missing objects based on the surveillance video of the target area to obtain a set of suspected missing objects; The second determination module is used to obtain a feature set of historically missing objects; For each suspected missing object in the set of suspected missing objects, respectively, performing: determining feature information of the suspected missing object based on the feature set, matching the feature information of the suspected missing object with the feature set, and determining a missing probability of the suspected missing object based on the matching result; Screening out suspected missing objects with missing amounts of the target object from the set of suspected missing objects in descending order of missing probabilities as actual missing objects; It also includes a probability acquisition module for: Get the initial probability corresponding to the matched reference feature; Determining the missing probability of the suspected missing object based on an influence rule of the initial probability corresponding to the reference feature on the missing probability; In the impact rule: when the reference feature is a behavioral feature of leaving the target area, the missing probability is positively correlated with the initial probability corresponding to the reference feature; when the reference feature is a reasonable behavioral feature within the target area, the missing probability is negatively correlated with the initial probability corresponding to the reference feature.

12. The device according to claim 11, wherein The first determining module is configured to: Taking a target time point as a benchmark, intercepting a target video segment of a target duration from the surveillance video of the target area; the target time point is the time point when the number of target objects in the target area decreases; determining features of the disappeared object based on a tracking result of the target object in the target video clip; The object corresponding to the disappeared object feature is determined as a suspected missing object.

13. The apparatus according to claim 12, further comprising an identification module configured to: Obtaining a video stream collected after the target time point; Identifying object identifiers of existing target objects in the video stream to obtain a set of existing target objects; A difference between the target object set that should exist and the target object set that already exists is determined to obtain a suspected missing object.

14. The device according to claim 13, wherein The identification module is used to: Detecting key features of key parts of an existing target object, where the key features can uniquely identify the existing target object; The key features of the existing target object are matched with the key features of the known object to obtain the object identifier of the existing target object.

15. The device according to claim 13 or 14, wherein The identification module is further used to: Detect the contour features of existing target objects; The shape contour features of the existing target object are matched with the shape contour features of a known object to obtain an object identifier of the existing target object.

16. The device according to claim 11, wherein The second determining module is further configured to: Acquire a collection of remote objects; the collection of remote objects includes object identifiers in other areas and / or object identifiers that have left the target area through an access control device; If any suspected missing object is included in the remote object set, determining the suspected missing object as an actual missing object; In a case where the total number of actual missing objects is equal to the missing amount of the target object, it is determined that all actual missing objects are identified.

17. The apparatus according to claim 16, wherein the second determining module is further configured to: When the total number of actual missing objects is less than the missing amount of the target object, determining the difference between the set of suspected missing objects and the set of actual missing objects to obtain the objects to be verified; The shape contour features of the object to be checked are matched with the off-site shape contour feature set. If the shape contour features of any object to be checked are included in the off-site shape contour feature set, the object to be checked is determined to be an actual missing object. The off-site shape contour feature set includes the shape contour features of candidate objects in other areas, and the object identification of the candidate objects cannot be identified.

18. The apparatus according to claim 16 or 17, further comprising an alarm module, configured to: In the case that the total number of actual missing objects is less than the missing amount of the target objects, an alarm message is output, where the alarm message is used to instruct to check the unconfirmed missing objects.

19. The device according to claim 11, wherein Before screening out suspected missing objects having a missing amount of the target object from the set of suspected missing objects in descending order of missing probabilities, the method further includes: The missing probability of each suspected missing object in the suspected missing object set is normalized so that the sum of the missing probabilities of all suspected missing objects is a target value.

20. The apparatus according to claim 11, wherein the identification module is further configured to: If there is a returned object before the missing duration of the actual missing object reaches the preset duration, the key features of the returned object are compared with the key features of the actual missing object. If it is determined that they are not the same object, it is determined that a replacement behavior has occurred.

21. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 10.

22. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-10.

23. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 10.

Images