Lost Item Detection Method, Device, Equipment and Storage Medium

Abstract

The present disclosure provides a lost item detection method, which relates to the field of artificial intelligence. The method includes: obtaining N images at at least two moments within a predetermined time period, where at least one image is obtained at each of the at least two moments; determining the motion state of the item to be detected based on the trajectory data of the item to be detected in the N images; if the motion state is a specific state, obtaining the matching degree between the item to be detected and each of the M users; if the matching degree between the first user and the item to be detected is greater than or equal to a first threshold, determining the first user as a candidate user for the lost item. Embodiments of the present disclosure can achieve the effect of automatically detecting lost items and giving timely prompts, and avoiding additional expenses caused by the post-treatment of the owner of the lost item. The present disclosure also provides a lost item detection device, equipment, storage medium and program product.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence, and more particularly, to a lost item detection method, apparatus, device, medium, and program product. Background Art

[0002] It is possible that people's personal belongings are lost. Especially in public places, if the loss of personal belongings is not noticed and the time and location of the loss are uncertain, it is difficult to retrieve them. Usually, it depends on the staff in public places or people around to remind the owner on the spot. Or the lost items are sent to the relevant department and wait for the owner to claim them.

[0003] In the process of implementing the inventive concept of the present disclosure, the inventors found that there are at least the following problems in the related art: It is accidental that the staff or people around can find that someone has lost something, and omissions are inevitable. And the owner may not know which department to claim the lost item, and the process of searching is also troublesome. Therefore, the process of people searching for lost personal belongings is cumbersome, time-consuming, and the retrieval probability is low. Summary of the Invention

[0004] In view of the above problems, the present disclosure provides a lost item detection method, apparatus, device, medium, and program product.

[0005] One aspect of an embodiment of the present disclosure provides a lost item detection method, including: obtaining N images at at least two moments within a predetermined time period, where at least one image is obtained at each of the at least two moments, the N images include an item to be detected in a target area and M users, N is an integer greater than or equal to 2, and M is an integer greater than or equal to 1; determining a motion state of the item to be detected based on trajectory data of the item to be detected in the N images; if the motion state is a specific state, obtaining a matching degree between the item to be detected and each of the M users; if the matching degree between a first user and the item to be detected is greater than or equal to a first threshold, determining the first user as a candidate user for the lost item, where the M users include the first user.

[0006] According to an embodiment of the present disclosure, obtaining the matching degree between the item to be detected and each of the M users includes: obtaining displacement data of the item to be detected in a first direction according to the trajectory data, where the first direction is a vertical direction of a spatial coordinate system in the target area; obtaining first distance data between the item to be detected and each of the users; obtaining the matching degree between the item to be detected and each of the users based on the first distance data and the displacement data.

[0007] According to an embodiment of the present disclosure, M pieces of the first distance data are obtained in total, and there is a maximum distance data among the M pieces of the first distance data. The obtaining of the matching degree between the item to be detected and each user includes: setting the initial value of i to 1, where i is an integer greater than or equal to 1 and less than or equal to M, and cyclically performing the following operations: obtaining a first ratio of the i-th piece of the first distance data to the maximum distance data, where the i-th piece of the first distance data is the distance data between the item to be detected and the i-th user; obtaining the matching degree between the item to be detected and the i-th user based on the first ratio and the displacement data; and increasing the value of i by 1.

[0008] According to an embodiment of the present disclosure, the at least two moments include a second moment and a first moment, and the displacement data includes the moving distance of the item to be detected from the first moment to the second moment. The obtaining of the matching degree between the item to be detected and the i-th user based on the first ratio and the displacement data includes: obtaining a first change amount of the i-th piece of the first distance data from the first moment to the second moment; obtaining a second change amount of the maximum distance data from the first moment to the second moment; obtaining a second ratio of the first change amount to the second change amount; and obtaining the matching degree between the item to be detected and the i-th user based on the first ratio, the second ratio, and the displacement data.

[0009] According to an embodiment of the present disclosure, the at least two moments include a first moment and a second moment. The determining of the motion state of the item to be detected based on the trajectory data of the item to be detected in the N images includes: obtaining the coordinate data of the item to be detected at the first moment and the coordinate data at the second moment; calculating a motion state vector of the item to be detected according to the coordinate data at the first moment and the coordinate data at the second moment; and determining the motion state according to the motion state vector.

[0010] According to an embodiment of the present disclosure, the calculating of the motion state vector of the item to be detected according to the coordinate data at the first moment and the coordinate data at the second moment includes: obtaining the displacement data of the item to be detected in a first direction according to the coordinate data at the first moment and the coordinate data at the second moment, where the first direction is the vertical direction of the spatial coordinate system in the target area; and obtaining the motion state vector according to the displacement data.

[0011] According to an embodiment of the present disclosure, it includes: obtaining a time change amount from the first moment to the second moment; where the obtaining of the motion state vector according to the displacement data includes: obtaining the motion state vector according to the displacement data and the ratio of the displacement data to the time change amount.

[0012] According to an embodiment of the present disclosure, determining the motion state based on the motion state vector includes: determining the direction of the motion state vector according to the ratio; and determining that the motion state is the specific state when the direction of the motion state vector is a predetermined direction and / or when the displacement data is greater than or equal to a second threshold.

[0013] According to an embodiment of the present disclosure, it further includes: obtaining a predicted classification result of the item to be detected; wherein, determining the motion state of the item to be detected includes: determining the motion state of the item to be detected when the predicted classification result is a predetermined category.

[0014] According to an embodiment of the present disclosure, the at least two moments include a second moment, and the method further includes obtaining the trajectory data, specifically including: obtaining a first detection box of the item to be detected in the image at the second moment; processing the first detection box according to the predicted classification result; and performing target tracking prediction based on the data of the processed first detection box to obtain the trajectory data at the second moment.

[0015] According to an embodiment of the present disclosure, the at least two moments further include a first moment, the first moment being before the second moment, and it further includes: obtaining a second detection box of the item to be detected in the image at the first moment; obtaining a first screenshot based on the first detection box and a second screenshot based on the second detection box; obtaining the feature similarity between the first screenshot and the second screenshot; wherein, obtaining the trajectory data at the second moment includes: performing target tracking prediction based on the data of the processed first detection box and the feature similarity to obtain the trajectory data at the second moment.

[0016] Another aspect of the embodiments of the present disclosure provides a lost item detection device, including: an image acquisition module configured to obtain N images at at least two moments within a predetermined time period, wherein at least one image is obtained at each of the at least two moments, the N images include an item to be detected in a target area and M users, N is an integer greater than or equal to 2, and M is an integer greater than or equal to 1; a state determination module configured to determine the motion state of the item to be detected based on the trajectory data of the item to be detected in the N images; a matching degree module configured to obtain the matching degree between the item to be detected and each of the M users if the motion state is a specific state; and a candidate user module configured to determine the first user as a candidate user of the lost item if the matching degree between the first user and the item to be detected is greater than or equal to a first threshold, wherein the M users include the first user.

[0017] Another aspect of the embodiments of the present disclosure provides an electronic device, including: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the method as described above.

[0018] Another aspect of the embodiments of the present disclosure further provides a computer-readable storage medium, on which executable instructions are stored, and when the instructions are executed by a processor, the processor is caused to execute the method as described above.

[0019] Another aspect of the embodiments of the present disclosure further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method as described above is implemented.

[0020] The above one or more embodiments have the following beneficial effects: Using N images of the item to be detected and the user, target tracking can be performed to obtain the trajectory data of the item to be detected, and the motion state of the item to be detected can be judged in a timely manner within a predetermined time period. If the motion state is a specific state, the candidate users of the lost item can be determined using the matching degree, so as to confirm and remind in a timely manner, thereby realizing automatic lost item detection, timely prompting, and avoiding the additional cost caused by post-event processing, which can at least partially solve the problems that the process of people looking for lost personal items is cumbersome, time-consuming, and the recovery probability is low. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Through the following description of the embodiments of the present disclosure with reference to the drawings, the above content and other objects, features, and advantages of the present disclosure will become clearer. In the drawings:

[0022] Figure 1 Schematically shows an application scenario diagram of the lost item detection method according to an embodiment of the present disclosure;

[0023] Figure 2 Schematically shows a flowchart of the lost item detection method according to an embodiment of the present disclosure;

[0024] Figure 3 Schematically shows a flowchart of determining the motion state of the item to be detected according to an embodiment of the present disclosure;

[0025] Figure 4 Schematically shows a flowchart of obtaining trajectory data according to an embodiment of the present disclosure;

[0026] Figure 5 Schematically shows a flowchart of obtaining trajectory data according to another embodiment of the present disclosure;

[0027] Figure 6a Schematically shows a flowchart of determining the motion state according to another embodiment of the present disclosure;

[0028] Figure 6b Schematically shows a spatial coordinate system diagram according to an embodiment of the present disclosure;

[0029] Figure 7 Schematically shows a flowchart for calculating a motion state vector according to an embodiment of the present disclosure;

[0030] Figure 8 Schematically shows a flowchart for calculating a motion state vector according to another embodiment of the present disclosure;

[0031] Figure 9 Schematically shows a flowchart for determining a motion state according to another embodiment of the present disclosure;

[0032] Figure 10 Schematically shows a flowchart for obtaining a matching degree according to an embodiment of the present disclosure;

[0033] Figure 11 Schematically shows a flowchart for obtaining a matching degree according to another embodiment of the present disclosure;

[0034] Figure 12 Schematically shows a flowchart for obtaining a matching degree according to another embodiment of the present disclosure;

[0035] Figure 13 Schematically shows an architecture diagram for implementing a lost item detection method according to an embodiment of the present disclosure;

[0036] Figure 14 Schematically shows an architecture diagram of a feature extractor according to an embodiment of the present disclosure;

[0037] Figure 15 Schematically shows a structural block diagram of a lost item detection device according to an embodiment of the present disclosure;

[0038] Figure 16 Schematically shows a block diagram of an electronic device suitable for implementing a lost item detection method according to an embodiment of the present disclosure. Detailed implementation manners

[0039] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for the sake of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, obviously, one or more embodiments can also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present disclosure.

[0040] When people lose their personal belongings, it will cause property losses. If important items are lost, it will also cause psychological harm. For institutions with management responsibilities in public places, they may be required by the owner to retrieve the lost items, and the search process is quite laborious and costly. If the items are not retrieved, it may lead to a decline in service satisfaction.

[0041] In some embodiments, the function of automatically detecting lost items can be implemented based on hardware sensors. Such as light sensors, temperature sensors, sound sensors or vibration sensors, etc., which are used to detect items or pedestrians. In this embodiment, additional hardware sensors are added in public places. If the area is large, the number of added sensors is large and the cost is also high. In addition, it may not be possible to accurately detect whether it is a personal item dropped by people or other irrelevant items, resulting in false judgment and detection failure.

[0042] Embodiments of the present disclosure provide a lost item detection method, device, equipment, medium and program product. Specifically, N images of the item to be detected and the user can be obtained, and target tracking is performed to obtain the trajectory data of the item to be detected, so as to timely judge the motion state of the item to be detected within a predetermined time period. If the motion state is a specific state, the candidate user of the lost item can be determined by using the matching degree, so as to confirm and remind in time, thus realizing the automatic detection of lost items, timely prompting, and avoiding the additional expenses caused by post-processing. It can at least partially solve the problems that the process of people looking for lost personal belongings is cumbersome, time-consuming, and the retrieval probability is low.

[0043] In the technical solution of the present disclosure, before obtaining or collecting the user's personal information, the user's authorization or consent is obtained. The processing of the collection, storage, use, processing, transmission, provision, disclosure and application of the user's personal information involved all comply with the provisions of relevant laws and regulations, take necessary confidentiality measures, and do not violate public order and good customs.

[0044] Figure 1 An application scenario diagram of the lost item detection method according to an embodiment of the present disclosure is schematically shown.

[0045] As Figure 1 shown, the application scenario 100 according to this embodiment may include a first camera 111, a second camera 112, a network 140, and a server 150. Among them, the first user 131, the second user 132, the third user 133, and the fourth user 134 are users in public places. The network 140 is used to provide a medium for the communication link between the first camera 111, the second camera 112, and the server 150. The network 140 may include various connection types, such as wired, wireless communication links or fiber optic cables, etc. Among them, the first camera 111 and the second camera 112 may be respectively arranged at different positions in public places.

[0046] The first camera 111 and the second camera 112 can respectively capture videos within their corresponding areas. For example, when users 131 - 134 pass through the shooting area of any camera, their images can be collected. Then, the first camera 111 and the second camera 112 can interact with the server 150 via the network 140 to send videos or images, etc. The first camera 111 and the second camera 112 can be surveillance cameras or various electronic devices with a camera function. The various electronic devices include but are not limited to smartphones, tablets, laptop computers, desktop computers, and so on.

[0047] The server 150 can be a server that provides various services, such as processing videos or images, etc., and performing target tracking, motion state detection, and match degree acquisition, etc. In some embodiments, an electronic device can be used to interact with the server 150 to send user requests. The background management server can analyze and process data such as received user requests, etc., and feedback the processing results (such as web pages, information, or data, etc. obtained or generated according to user requests) to the electronic device.

[0048] As Figure 1 shown, the personal belongings (such as headphones) of the third user 133 are lost. The first camera 111 or the second camera 112 can respectively capture videos and send them to the server 150 to implement the lost item detection method.

[0049] It should be noted that the lost item detection method provided by the embodiments of the present disclosure can generally be executed by the server 150. Correspondingly, the lost item detection device provided by the embodiments of the present disclosure can generally be set in the server 150. The lost item detection method provided by the embodiments of the present disclosure can also be executed by a server or a server cluster different from the server 150 and capable of communicating with the above - mentioned camera and / or the server 150. Correspondingly, the lost item detection device provided by the embodiments of the present disclosure can also be set in a server or a server cluster different from the above - mentioned camera and / or the server 150 and capable of communicating with the server 150.

[0050] It should be understood that Figure 1 the numbers of cameras, users, networks, and servers in Figure 1 are merely illustrative. According to the implementation requirements, there can be any number of cameras, buildings, users, networks, and servers. Similarly, the installation positions of the cameras are also merely illustrative. Public places are not limited to Figure 1 the content shown in

[0051] The following will be based on Figure 1 the described scenario and will describe the lost item detection method of the embodiments of the present disclosure in detail through Figures 2 to 14 the following.

[0052] Figure 2 Schematically shows a flowchart of a lost item detection method according to an embodiment of the present disclosure.

[0053] As Figure 2 shown, the lost item detection method of this embodiment includes operations S210 to S260.

[0054] In operation S210, N images are obtained at at least two moments within a predetermined time period, wherein at least one image is obtained at each of the at least two moments, and the N images include the item to be detected and M users in the target area, N is an integer greater than or equal to 2, and M is an integer greater than or equal to 1.

[0055] Exemplarily, the predetermined time period can be determined according to the moving speed of the user, and the user does not leave the target area within this time period. For example, images are acquired on the first day and the second day respectively, then the user may not be in this area on the second day, and the lost item may also be picked up. For example, a video with a time period of 1 min (only for example) can be obtained, and images of each frame can be acquired from the video, and targets (items and users) can be tracked for multiple consecutive frames of images. Therefore, setting the predetermined time period can increase the probability of matching the owner of the lost item.

[0056] For example, multiple cameras can be set, and each camera acquires an image at each moment respectively. In some embodiments, RGB+Depth Map (RGBD) image technology can also be used to obtain three-dimensional images to obtain the depth information of the targets in the images.

[0057] The target area refers to a certain area in a public place, which can be determined according to the area that the arranged cameras can photograph. This area can include one or more items to be detected, one or more users, and at least one user carries the item to be detected.

[0058] In operation S220, based on the trajectory data of the item to be detected in the N images, the motion state of the item to be detected is determined.

[0059] Exemplarily, the trajectory data can characterize the continuous movement trajectory of the item to be detected within a predetermined time period in the video, and it can be obtained through the coordinate data of the item to be detected on each frame of the picture.

[0060] In operation S230, it is judged whether the motion state is a specific state. If so, operation S240 is executed. If not, it ends.

[0061] Exemplarily, the motion state may include an acceleration state or a deceleration state of the item to be detected, and may also include types such as a parallel movement state, a falling state, or a rising state relative to the ground. For example, if the item to be detected is a hydrogen balloon, it may fly upward without the user noticing. If the item to be detected is a mobile phone, wallet, or earphone, etc., it may fall downward.

[0062] In operation S240, if the motion state is a specific state, obtain the matching degree between the item to be detected and each of the M users.

[0063] Exemplarily, it may be that after the item to be detected is determined to be in a specific state in each of a series of consecutive frames of images, the matching degree is obtained. The specific state is one or more states in the motion state, such as a rising state or a falling state, etc.

[0064] In operation S250, determine whether the matching degree between the item to be detected and any one user is greater than or equal to the first threshold. If so, execute operation S260. If not, end.

[0065] In some embodiments, the matching degree can be obtained through a pre-trained classification model. For example, obtain the type, coordinates, and moving speed of the item to be detected, the information of the item and the nearest user captured for the first time within a predetermined time period, the distance information between the item and surrounding users, etc., to obtain the probability value (i.e., the matching degree) that the item belongs to a certain user.

[0066] In other embodiments, the matching degree can be determined according to the moving trajectory of the item and the moving trajectory of the user, and by judging the change in the distance between the item and the owner after the item is lost.

[0067] In operation S260, if the matching degree between the first user and the item to be detected is greater than or equal to the first threshold, determine the first user as a candidate user for the lost item, where the M users include the first user.

[0068] Exemplarily, the first threshold can be a predetermined fixed value, such as 0.8 (only for example). It can also be a dynamic value. For example, sort the M matching degrees, and the top 5 (only for example) corresponding users are considered candidate users. The matching degree ranked 5th is the first threshold.

[0069] In some embodiments, the lost item and the suspected owner information (i.e., candidate users) can be stored and relevant personnel or candidate users can be notified.

[0070] According to an embodiment of the present disclosure, N images of the item to be detected and the user can be obtained, and object tracking is performed to obtain the trajectory data of the item to be detected, and the motion state of the item to be detected is judged in a timely manner within a predetermined time period. If the motion state is a specific state, the candidate users of the lost item can be determined by using the matching degree, so as to confirm and remind in a timely manner, thereby realizing automatic detection of lost items, timely prompting, and avoiding the effect of additional expenses caused by post-processing, which can at least partially solve the problems that the process of people looking for lost personal items is cumbersome, time-consuming, and the recovery probability is low.

[0071] Figure 3 Schematically shows a flowchart for determining the motion state of an item to be detected according to an embodiment of the present disclosure.

[0072] As Figure 3 shown, the determination of the motion state of the item to be detected in this embodiment includes operations S310 to S320. Among them, operation S220 may include operation S320.

[0073] In operation S310, a predicted classification result of the item to be detected is obtained.

[0074] In operation S320, when the predicted classification result is a predetermined category, the motion state of the item to be detected is determined.

[0075] Exemplarily, the predicted classification result includes a prediction result of the type of the item to be detected, which can be realized by a classification model based on a neural network. The predetermined category may include personal item categories (such as balloon category, mobile phone category, wallet category or earphone category, etc.) or other categories (such as garbage category). Its function is to reduce the probability of misjudgment and reduce the amount of data processing. For example, in the process of a user throwing a mineral water bottle into the trash can, it can be determined that the mineral water bottle has a parabolic trajectory, but in fact it is not a lost item. When the predicted classification result is other category, the processing can be stopped at this time.

[0076] Figure 4 Schematically shows a flowchart for obtaining trajectory data according to an embodiment of the present disclosure.

[0077] As Figure 4 shown, the obtaining of the trajectory data in this embodiment includes operations S410 to S430. Among them, at least two moments include the second moment.

[0078] In operation S410, a first detection frame of the item to be detected in the image at the second moment is obtained.

[0079] Exemplarily, the second moment may be any moment within a predetermined time period. Object tracking can be performed on the image corresponding to this moment, and the area where the item to be detected is located is framed by the first detection frame.

[0080] In operation S420, process the first detection box according to the predicted classification result.

[0081] According to an embodiment of the present disclosure, the first detection box may be one or more. When there are multiple first detection boxes, it means that some are inaccurate. Therefore, the predicted classification result can be used to determine the same item on multiple images. The predicted classification result can also be used to apply non-maximum suppression processing to one or more first detection boxes to remove redundant detection boxes.

[0082] In operation S430, perform target tracking prediction based on the data of the processed first detection box to obtain the trajectory data at the second moment.

[0083] Exemplarily, the target tracking prediction can be implemented by technologies such as Kalman filter, pose information tracking, or Hungarian algorithm. Taking the Kalman filter as an example, it can process according to the processed first detection box and / or the detection box before the second moment to obtain a more accurate result to achieve target tracking.

[0084] Based on the Kalman filter, the trajectory data at the second moment can be predicted, that is, the coordinate data of the item to be detected on the image at the second moment. For example, the trajectory data at the second moment is (x, y, d, r, h, c, β), where (x, y, d) represents the center coordinates of the object to be detected in the spatial coordinate system in the target area, r represents the aspect ratio, h represents the height, c represents the predicted classification result, and β represents the mark of the detection box. Specifically, x is the x-axis coordinate of the spatial coordinate system, y is the y-axis coordinate of the spatial coordinate system, and d is the d-axis coordinate of the spatial coordinate system. The d-axis is used to represent the depth information in the image. For example, if there are two users in the target area, then the c of the two users is the same, while β is different, so as to perform target tracking on the two users respectively.

[0085] According to an embodiment of the present disclosure, after processing the first detection box according to the predicted classification result, a more accurate tracking effect can be obtained by using the processed first detection box.

[0086] Figure 5 Schematically shows a flowchart of obtaining trajectory data according to another embodiment of the present disclosure.

[0087] As Figure 5 shown, obtaining the trajectory data in this embodiment may include operations S510 to S540. Among them, at least two moments further include the first moment, and the first moment is before the second moment.

[0088] In operation S510, obtain the second detection box of the item to be detected in the image at the first moment.

[0089] Exemplarily, the first moment and the second moment can be the moments corresponding to two adjacent frames in the video captured by the same camera. The acquisition methods of the second detection box and the first detection box can be the same or similar. The item to be detected can remain moving between the first moment and the second moment.

[0090] In operation S520, a first screenshot is obtained based on the first detection box, and a second screenshot is obtained based on the second detection box.

[0091] Exemplarily, the content of the first screenshot is the area within the first detection box, and the content of the second screenshot is the area within the second detection box.

[0092] In operation S530, the feature similarity between the first screenshot and the second screenshot is obtained.

[0093] Exemplarily, the image features (such as texture, color, or edge features) of the first screenshot and the second screenshot can be extracted respectively, and then the cosine similarity (i.e., the feature similarity) can be calculated.

[0094] In operation S540, target tracking prediction is performed based on the data of the processed first detection box and the feature similarity, and the trajectory data at the second moment is obtained.

[0095] Exemplarily, in the case of performing Kalman filtering on the data of the processed first detection box, the feature similarity can be input as an auxiliary weight into the Kalman filter to combine the data at the first moment and obtain a more accurate target tracking result at the second moment.

[0096] Figure 6a Schematically shows a flowchart of determining a motion state according to another embodiment of the present disclosure. Figure 6b Schematically shows a spatial coordinate system diagram according to an embodiment of the present disclosure.

[0097] As Figure 6a shown, determining the motion state of the item to be detected based on the trajectory data of the item to be detected in N images in operation S220 includes operations S610 to S630.

[0098] In operation S610, the coordinate data of the item to be detected at the first moment and the coordinate data at the second moment are obtained.

[0099] Exemplarily, for example, the coordinate data at the second moment (assumed to be the t moment) includes (x t , y t , d t ), and the coordinate data at the first moment (assumed to be the t - 1 moment) includes (x t-1 , y t-1 , d t-1 ).

[0100] In operation S620, according to the coordinate data at the first moment and the coordinate data at the second moment, calculate the motion state vector of the item to be detected.

[0101] In operation S630, determine the motion state according to the motion state vector.

[0102] Refer to Figure 6b , this spatial coordinate system can be calibrated by the camera or by the server. Specifically, after obtaining N images, the target in the images can be transformed into Figure 6b the spatial coordinate system in. Figure 6b 1, 2, 3, 4 with dashed boxes in represent the positive quadrant areas of the spatial coordinate system (i.e., the positive direction of the y-axis), and 5, 6, 7, 8 with dashed boxes represent the negative quadrant areas of the spatial coordinate system (i.e., the negative direction of the y-axis). Figure 6b "0" in is the origin.

[0103] According to an embodiment of the present disclosure, according to the coordinate data of the item to be detected at the first moment and the coordinate data at the second moment, a motion state vector and the direction of this vector can be obtained. If it points to the positive quadrant area, it can be considered as the rising state. If it points to the negative quadrant area, it can be considered as the falling state.

[0104] Figure 7 Schematically shows a flowchart for calculating the motion state vector according to an embodiment of the present disclosure.

[0105] As Figure 7 shown, calculating the motion state vector of the item to be detected in operation S620 includes operations S710 to S720.

[0106] In operation S710, according to the coordinate data at the first moment and the coordinate data at the second moment, obtain the displacement data of the item to be detected in the first direction, where the first direction is the vertical direction of the spatial coordinate system in the target area.

[0107] Refer to Figure 6b , the first direction can be the direction along the y-axis. The y-axis can be perpendicular to the ground.

[0108] In operation S720, obtain the motion state vector according to the displacement data.

[0109] According to an embodiment of the present disclosure, the difference between y t and y t-1 can be obtained as the displacement data, and the motion state vector is calculated based on this. If the difference is positive, it can be considered that the motion state vector points to the positive quadrant area. If the difference is negative, it can be considered that the motion state vector points to the negative quadrant area.

[0110] Figure 8 Schematically shows a flowchart for calculating the motion state vector according to another embodiment of the present disclosure.

[0111] As Figure 8 shown, the calculation of the motion state vector in this embodiment includes operations S810 to S820.

[0112] In operation S810, the time variation from the first moment to the second moment is obtained.

[0113] In operation S820, based on the displacement data and the ratio of the displacement data to the time variation, the motion state vector is obtained.

[0114] Exemplarily, let Δt represent the time variation (e.g., the time interval between two frames of images). The motion state vector can be expressed in the following form:

[0115]

[0116] Figure 9 Schematically shows a flowchart for determining the motion state according to another embodiment of the present disclosure.

[0117] As Figure 9 shown, the lost object detection method in this embodiment includes operations S910 to S920.

[0118] In operation S910, the direction of the motion state vector is determined according to the ratio.

[0119] In operation S920, when the direction of the motion state vector is a predetermined direction, and / or when the displacement data is greater than or equal to the second threshold, the motion state is determined to be a specific state.

[0120] Exemplarily, the second threshold can be determined according to the type, volume or mass of the object, and the speed of the object when it falls or rises. Its function is to determine whether the object to be detected is in a normal moving state when being carried or a moving state after being lost, and can prevent misjudgment.

[0121] According to an embodiment of the present disclosure, in order to prevent misjudgment, the object to be detected is determined to be in a specific state only when it drops more than a certain distance within the time interval. For example, when a user walks normally with an item carried, there may be a certain displacement in the first direction, and it is not determined to be in a specific state. Or when a user picks up and puts down a mobile phone, there is a certain displacement in the first direction, and this displacement is not the same as the displacement of the mobile phone in free fall, and it is also not determined to be in a specific state.

[0122] In some other embodiments, the motion state vector can be represented by (x t , y t , d t ) and (x t-1 , yt-1 , d t-1 ), the three data in

[0123]

[0124] Referring to Equation (2), by judging the motion state vector Whether the directions of the first three velocity components are negative and whether they point to the negative quadrant of the space coordinate system, and combining whether the displacement of the object in the last term is greater than the first threshold, the motion state of the object is determined.

[0125] Figure 10 Schematically shows a flowchart of obtaining the matching degree according to an embodiment of the present disclosure.

[0126] As Figure 10 shown, obtaining the matching degree between the item to be detected and each of the M users in operation S240 includes operations S1010 to S1030.

[0127] In operation S1010, according to the trajectory data, the displacement data of the item to be detected in the first direction is obtained.

[0128] Exemplarily, the third moment may be the moment when the item to be detected is initially determined to be in a specific state. For example, N frames of images are obtained from a video with a predetermined time period length. When the item to be detected is initially determined to be in a specific state in the nth frame, the corresponding moment is the third moment. If it is determined to be in a specific state from the nth frame to the n + mth frame, the matching degree is calculated. Then the moment corresponding to the n + mth frame is the second moment. n and m are integers greater than or equal to 1. Among them, the displacement data in this embodiment may be the displacement data between the second moment and the third moment. It can be obtained by the y-axis coordinate y t at the second moment and the y-axis coordinate y0 at the third moment.

[0129] In operation S1020, the first distance data between the item to be detected and each user is obtained.

[0130] Exemplarily, the first distance data can be obtained by calculation methods such as Mahalanobis distance or Euclidean distance. Denote the coordinate data (x, y, d) of the item to be detected as D, and denote the coordinate data (x i , y i , d i ) of the i-th user as D i . Taking the first distance data as the Euclidean distance as an example, the Euclidean distance between the item to be detected and the i-th user is:

[0131]

[0132] In operation S1030, based on the first distance data and the displacement data, the matching degree between the item to be detected and each user is obtained.

[0133] Exemplarily, the moving speed of the user and the distances between each user and the item to be detected when the user has not lost the item can be considered, and the matching degree is obtained by combining the first distance data and the displacement data. For example, a classification model based on a neural network can be used to form a vector with the above-mentioned moving speed, distance, first distance data, and displacement data and input it into the classification model to obtain the predicted output result. The predicted result can be whether a certain user is a candidate user, and the matching degree can be the probability value output by the classification model.

[0134] In some other embodiments, the matching degree can also be obtained through a preset matching degree calculation rule, which will be further introduced below in conjunction with Figure 11 and Figure 12 for further introduction.

[0135] Figure 11 Schematically shows a flowchart of obtaining the matching degree according to another embodiment of the present disclosure.

[0136] As Figure 11 shown, obtaining the matching degree between the item to be detected and each user in operation S1030 includes: setting the initial value of i to 1, where i is an integer greater than or equal to 1 and less than or equal to M, and loop-executing operations S1110 to operation S1130. Among them, a total of M first distance data are obtained, and there is a maximum distance data among the M first distance data.

[0137] In operation S1110, obtain the first ratio of the i-th first distance data to the maximum distance data, where the i-th first distance data is the distance data between the item to be detected and the i-th user.

[0138] Exemplarily, the maximum distance data is the Euclidean distance between the item to be detected and the farthest user. The coordinate data of the farthest user can be represented by D max The maximum distance data is represented by Euclid(D, D max ). The first ratio is the ratio between Euclid(D, D i ) and Euclid(D, D max ).

[0139] In operation S1120, based on the first ratio and the displacement data, obtain the matching degree between the item to be detected and the i-th user.

[0140] In some embodiments, the first ratio and the displacement data can be input into a classification model to obtain a classification result. In some other embodiments, a matching degree calculation rule preset for the first ratio and the displacement data can be considered.

[0141] In operation S1130, it is determined whether i is less than M. If so, the value of i is incremented by 1, and operation S1110 is executed. If not, the process ends.

[0142] According to an embodiment of the present disclosure, not only the displacement data of the object to be detected itself is considered, but also the distance relationship between surrounding users is considered, which can improve the accuracy of determining candidate users.

[0143] Figure 12 A flowchart of obtaining a matching degree according to another embodiment of the present disclosure is schematically shown.

[0144] As Figure 12 shown, obtaining the matching degree between the item to be detected and the i-th user based on the first ratio and displacement data in operation S1120 includes operations S1210 to S1240. Among them, at least two moments include the first moment and the second moment, and the displacement data includes the moving distance of the item to be detected from the first moment to the second moment.

[0145] In operation S1210, a first change amount of the first distance data of the i-th user from the first moment to the second moment is obtained.

[0146] Exemplarily, the first distance data at the second moment is expressed as Euclid(D t , D i，t ), and the first distance data at the first moment is expressed as Euclid(D t-1 , D i，t-1 ). Then the first change amount can be the difference between Euclid(D t , D i，t ) and Euclid(D t-1 , D i，t-1 ).

[0147] In operation S1220, a second change amount of the maximum distance data from the first moment to the second moment is obtained.

[0148] Exemplarily, the first distance data at the second moment is expressed as Euclid(D t , D max，t ), and the first distance data at the first moment is expressed as Euclid(D t-1 , D max，t-1 ). Then the first change amount can include the difference between Euclid(D t , D max，t ) and Euclid(D t-1 , D max，t-1 ).

[0149] In operation S1230, a second ratio of the first change amount to the second change amount is obtained.

[0150] At operation S1240, based on the first ratio, the second ratio, and the displacement data, obtain the matching degree between the item to be detected and the \(i\)th user.

[0151] According to an embodiment of the present disclosure, the matching degree \(p\) is as follows.

[0152]

[0153] Where \(\tau\) is an adjustment coefficient, which can be valued according to the actual scenario, such as according to the falling speed of different objects to be detected or the moving speed of different users. \(y\) t is the y-axis coordinate at time \(t\). \(D\) t represents the predicted value of the coordinate of the object to be detected at the second moment. \(D\) i，t represents the predicted value when the classification result at the second moment is "person". \(y\) represents the vertical distance in the spatial coordinate system, where \(y_0\) represents the y-axis coordinate at the third moment, and \(y\) t represents the y-axis coordinate at the second moment.

[0154] Referring to Equation (4), the calculation of the matching degree \(p\) is divided into two sub-parts with "+" as the boundary. The first term is the matching degree between the falling object and the people around, and the second term is the matching degree between the movement trajectory of the falling object and the movement trajectories of the people around. It should be noted that the trajectory matching degree is proposed based on the tendency that lost items often move away from the owner. Therefore, the greater the deviation in movement between frame differences, the higher the matching degree, and the higher the weight as the object gets closer to the ground.

[0155] According to an embodiment of the present disclosure, after calculating the matching degrees between the falling object and all the "people" with the classification result of "person", take the first \(k\) set matching degrees corresponding to the "people" as the owners, store the data, send it to the staff or send a reminder message to the owner.

[0156] Figure 13 Schematically shows an architecture diagram of an implementation of a lost item detection method according to an embodiment of the present disclosure. Figure 14 Schematically shows an architecture diagram of a feature extractor according to an embodiment of the present disclosure.

[0157] Taking the business hall of a bank branch as an example in a public place, such as Figure 13 shown, first, the camera with depth information in the business hall transmits the image information to the backend processing system in real time. The processing system discretely samples the continuous image sequence at a preset interval, and the sampled image is converted into an RGBD format image. In some embodiments, a camera without depth information can be used.

[0158] Then, a detector is used to detect and classify a preset target to obtain the classification information and spatial positioning (i.e., coordinate data) of the target. The classification information is used to mark the target classification and apply maximum suppression to the detection boxes to remove redundant detection boxes. The detection boxes after post-processing (here refers to maximum suppression processing) are input into the Kalman filter for trajectory prediction.

[0159] In addition, in order to achieve re-identification of the detected object (the same target marking box in the upper and lower two frames of images), the detection box is used to intercept the current frame image. In some embodiments, the intercepted target image can be a two-dimensional image. The feature extractor extracts the feature expression of the target, and then calculates the cosine distance with the target feature of the previous frame to obtain the similarity metric. The metric information will also be input into the Kalman filter as an auxiliary weight to calculate the predicted trajectory data at this moment, for example, obtaining (x, y, r, h, d, c, β).

[0160] Exemplarily, referring to Figure 14 , the feature extractor 1400 is used to extract the features of the two-dimensional target original image intercepted based on the detection box, and is composed of a residual module stack composed of two-dimensional convolution (such as 3*3Conv), batch regularization (such as BN, Batch Normalization), and rectified linear unit (such as Relu activation function). For example, 5 (only for example) of the above residual modules are used to form the feature extractor.

[0161] Next, in the case where the item is of a specific category, an operation of detecting item drop is performed according to the predicted trajectory data. The movement direction of the target can be calculated through the coordinate information and motion vector of the object in the three-dimensional space in the current frame and the previous frame, and combined with the pre-calibrated camera image coordinates (i.e., the spatial coordinate system in the target area, referring to Figure 6b ) to determine whether the object belongs to the falling state.

[0162] Next, if the item is in the falling state in N consecutive frames, it is determined as a lost item, and lost item matching processing is performed. The lost item and the suspected owner are matched, and the matching degree is obtained through the Euclidean distance metric in the three-dimensional space.

[0163] Finally, the lost item information is saved to a storage system (such as a database system), and the notification system is triggered to notify the duty personnel, and the reserved image information is provided for manual verification, and the owner is notified in a timely manner. The lost item information includes candidate user images, lost item information (such as coordinate data and lost item category), and detection time, etc.

[0164] Based on the above lost item detection method, the present disclosure also provides a lost item detection device. The following will be combined with Figure 15 to describe this device in detail.

[0165] Figure 15 A structural block diagram of a lost item detection device according to an embodiment of the present disclosure is schematically shown.

[0166] As Figure 15 shown, the lost item detection device 1500 of this embodiment includes an image acquisition module 1510, a state determination module 1520, a matching degree module 1530, and a candidate user module 1540.

[0167] The image acquisition module 1510 can perform operation S210 to obtain N images at at least two moments within a predetermined time period, where at least one image is obtained at each of the at least two moments, and the N images include the item to be detected in the target area and M users. N is an integer greater than or equal to 2, and M is an integer greater than or equal to 1.

[0168] The state determination module 1520 can perform operation S220 to determine the motion state of the item to be detected based on the trajectory data of the item to be detected in the N images.

[0169] According to an embodiment of the present disclosure, the device 1400 may further include an item classification module, which can perform operation S310 to obtain a predicted classification result of the item to be detected. The state determination module 1520 can also perform operation S320 to determine the motion state of the item to be detected when the predicted classification result is a predetermined category.

[0170] The matching degree module 1530 can perform operation S240 to obtain the matching degree between the item to be detected and each of the M users if the motion state is a specific state.

[0171] The candidate user module 1540 can perform operation S250 to determine the first user as a candidate user for the lost item if the matching degree between the first user and the item to be detected is greater than or equal to a first threshold, where the M users include the first user.

[0172] According to an embodiment of the present disclosure, the device 1400 may further include a trajectory data module, which can perform operations S410 to S430, operations S510 to S540, which will not be elaborated here.

[0173] According to an embodiment of the present disclosure, the state determination module 1520 can also perform operations S610 to S630, operations S710 to S720, operations S810 to S820, which will not be elaborated here.

[0174] According to an embodiment of the present disclosure, the matching degree module 1540 can also perform operations S1010 to S1030, operations S1110 to S1130, operations S1210 to S1240, which will not be elaborated here.

[0175] It should be noted that the implementation manners, technical problems solved, functions achieved, and technical effects achieved by each module / unit / sub-unit, etc. in the device part of the embodiments are the same as or similar to those of the corresponding steps in the method part of the embodiments, and will not be elaborated herein.

[0176] According to an embodiment of the present disclosure, any plurality of modules among the image acquisition module 1510, the status determination module 1520, the matching degree module 1530, and the candidate user module 1540 can be combined and implemented in one module, or any one of them can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module.

[0177] According to an embodiment of the present disclosure, at least one of the image acquisition module 1510, the status determination module 1520, the matching degree module 1530, and the candidate user module 1540 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or any other reasonable way of integrating or packaging circuits, etc., implemented by hardware or firmware, or implemented in any one of the three implementation manners of software, hardware, and firmware, or in any suitable combination of several of them. Alternatively, at least one of the image acquisition module 1510, the status determination module 1520, the matching degree module 1530, and the candidate user module 1540 can be at least partially implemented as a computer program module, and when the computer program module runs, it can execute the corresponding functions.

[0178] Figure 16 A block diagram of an electronic device suitable for implementing the lost item detection method according to an embodiment of the present disclosure is schematically shown.

[0179] As Figure 16 shown, the electronic device 1600 according to an embodiment of the present disclosure includes a processor 1601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1602 or a program loaded from a storage part 1608 into a random access memory (RAM) 1603. The processor 1601 can include, for example, a general microprocessor (such as a CPU), an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application specific integrated circuit (ASIC)), etc. The processor 1601 can also include on-board memory for caching purposes. The processor 1601 can include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0180] In the RAM 1603, various programs and data required for the operation of the electronic device 1600 are stored. The processor 1601, the ROM 1602, and the RAM 1603 are connected to each other via a bus 1604. The processor 1601 performs various operations of the method flow according to the embodiments of the present disclosure by executing programs in the ROM 1602 and / or the RAM 1603. It should be noted that the program can also be stored in one or more memories other than the ROM 1602 and the RAM 1603. The processor 1601 can also perform various operations of the method flow according to the embodiments of the present disclosure by executing programs stored in one or more memories.

[0181] According to an embodiment of the present disclosure, the electronic device 1600 may further include an input / output (I / O) interface 1605, and the input / output (I / O) interface 1605 is also connected to the bus 1604. The electronic device 1600 may further include one or more of the following components connected to the I / O interface 1605: an input part 1606 including a keyboard, a mouse, etc. An output part 1607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc. A storage part 1608 including a hard disk, etc. And a communication part 1609 including a network interface card such as a LAN card, a modem, etc. The communication part 1609 performs communication processing via a network such as the Internet. A drive 1610 is also connected to the I / O interface 1605 as needed. A removable medium 1611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1610 as needed so that a computer program read from it can be installed into the storage part 1608 as needed.

[0182] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments. It may also exist separately and not be assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of the present disclosure is implemented.

[0183] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include one or more memories other than the above-described ROM 1602 and / or RAM 1603 and / or ROM 1602 and RAM 1603.

[0184] An embodiment of the present disclosure further includes a computer program product, which includes a computer program that contains program code for executing the method shown in the flowchart. When the computer program product runs in a computer system, the program code is used to enable the computer system to implement the method provided by the embodiment of the present disclosure.

[0185] When the computer program is executed by the processor 1601, it executes the above functions defined in the system / apparatus of the embodiment of the present disclosure. According to an embodiment of the present disclosure, the above-described systems, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0186] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices and magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and be downloaded and installed through the communication part 1609, and / or be installed from the removable medium 1611. The program code contained in the computer program can be transmitted by any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0187] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1609, and / or be installed from the removable medium 1611. When the computer program is executed by the processor 1601, it executes the above functions defined in the system of the embodiment of the present disclosure. According to an embodiment of the present disclosure, the above-described systems, devices, apparatuses, modules, units, etc. can be implemented by computer program modules.

[0188] According to embodiments of the present disclosure, program code for executing the computer programs provided by the embodiments of the present disclosure may be written in any combination of one or more programming languages. Specifically, these computing programs may be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. The programming languages include, but are not limited to, such as Java, C++, Python, "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., by connecting through the Internet using an Internet service provider).

[0189] The above describes embodiments of the present disclosure. However, these embodiments are merely for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described separately above, this does not mean that the measures in each embodiment cannot be used advantageously in combination. The scope of the present disclosure is defined by the appended claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art can make various substitutions and modifications, and these substitutions and modifications should fall within the scope of the present disclosure.

Claims

1. A lost property detection method, comprising: Obtaining N images at at least two moments in a predetermined time period, wherein at least one image is obtained at each of the at least two moments in time, and the N images include an object to be detected and M users in a target area, where N is an integer greater than or equal to 2, and M is an integer greater than or equal to 1; determining a motion state of the object to be detected based on trajectory data of the object to be detected in the N images, wherein the trajectory data represents a continuous movement trajectory of the object to be detected within a predetermined time period and is obtained by using coordinate data of the object to be detected in the N images; If the motion state is a specific state, obtaining a matching degree between the object to be detected and each of the M users; If the matching degree between the first user and the item to be detected is greater than or equal to a first threshold, the first user is determined to be a candidate user of the lost item, wherein the M users include the first user.

2. The method according to claim 1, wherein Obtaining the matching degree between the object to be detected and each of the M users includes: Obtaining displacement data of the object to be detected in a first direction according to the trajectory data, wherein the first direction is a vertical direction of a spatial coordinate system in the target area; Obtaining first distance data between the object to be detected and each user; A matching degree between the object to be detected and each user is obtained based on the first distance data and the displacement data.

3. The method according to claim 2, wherein: A total of M first distance data are obtained, wherein the M first distance data have the maximum distance data, and obtaining the matching degree between the object to be detected and each user includes: Let the initial value of i be 1, i be an integer greater than or equal to 1, i be less than or equal to M, and loop to perform the following operations: Obtaining a first ratio of the i-th first distance data to the maximum distance data, wherein the i-th first distance data is the distance data between the object to be detected and the i-th user; Obtaining a matching degree between the object to be detected and the i-th user based on the first ratio and the displacement data; Let the value of i be increased by 1.

4. The method according to claim 3, wherein: The at least two moments include a second moment and a first moment, the displacement data includes a movement distance of the object to be detected from the first moment to the second moment, and obtaining a matching degree between the object to be detected and the i-th user based on the first ratio and the displacement data includes: Obtaining a first change in the i-th first distance data from the first moment to the second moment; Obtaining a second change in the maximum distance data from the first moment to the second moment; Obtaining a second ratio of the first change to the second change; A matching degree between the object to be detected and the i-th user is obtained based on the first ratio, the second ratio, and the displacement data.

5. The method according to claim 1, wherein The at least two moments include a first moment and a second moment, and determining the motion state of the object to be detected based on the trajectory data of the object to be detected in the N images includes: Obtaining coordinate data of the object to be detected at the first moment and coordinate data of the object at the second moment; Calculating a motion state vector of the object to be detected based on the coordinate data at the first moment and the coordinate data at the second moment; The motion state is determined according to the motion state vector.

6. The method according to claim 5, wherein: Calculating the motion state vector of the object to be detected based on the coordinate data at the first moment and the coordinate data at the second moment includes: Obtaining displacement data of the object to be detected in a first direction according to the coordinate data at the first moment and the coordinate data at the second moment, wherein the first direction is a vertical direction of the spatial coordinate system in the target area; The motion state vector is obtained according to the displacement data.

7. The method according to claim 6, wherein: include: Obtaining a time change from the first moment to the second moment; Wherein, obtaining the motion state vector according to the displacement data includes: The motion state vector is obtained according to the displacement data and the ratio of the displacement data to the time variation.

8. The method according to claim 7, wherein: Determining the motion state according to the motion state vector includes: determining the direction of the motion state vector according to the ratio; When the direction of the motion state vector is a predetermined direction, and / or when the displacement data is greater than or equal to a second threshold, the motion state is determined to be the specific state.

9. The method according to claim 1, wherein Also includes: Obtaining a predicted classification result of the object to be detected; Wherein, determining the motion state of the object to be detected includes: When the predicted classification result is a predetermined category, the motion state of the object to be detected is determined.

10. The method according to claim 9, wherein: The at least two moments include a second moment, and the method further includes obtaining the trajectory data, specifically including: Obtaining a first detection frame of the object to be detected in the image at the second moment; Processing the first detection frame according to the predicted classification result; Target tracking prediction is performed based on the processed data of the first detection frame to obtain trajectory data at the second moment.

11. The method according to claim 10, wherein: The at least two moments further include a first moment, the first moment being before the second moment, and further include: Obtaining a second detection frame of the object to be detected in the image at the first moment; Obtaining a first screenshot based on the first detection frame, and obtaining a second screenshot based on the second detection frame; Obtaining feature similarity between the first screenshot and the second screenshot; Wherein, obtaining the trajectory data at the second moment includes: Target tracking prediction is performed based on the processed data of the first detection frame and the feature similarity to obtain trajectory data at the second moment.

12. A lost property detection device, comprising: an image acquisition module, configured to acquire N images at at least two moments in a predetermined time period, wherein at least one image is acquired at each of the at least two moments in time, and the N images include an object to be detected and M users in a target area, where N is an integer greater than or equal to 2, and M is an integer greater than or equal to 1; a state determination module, configured to determine a motion state of the object to be detected based on trajectory data of the object to be detected in the N images, wherein the trajectory data represents a continuous movement trajectory of the object to be detected within a predetermined time period and is obtained by using coordinate data of the object to be detected in the N images; a matching degree module, configured to obtain a matching degree between the object to be detected and each of the M users if the motion state is a specific state; The candidate user module is configured to determine that the first user is a candidate user of the lost item if the matching degree between the first user and the item to be detected is greater than or equal to a first threshold, wherein the M users include the first user.

13. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors are enabled to execute the method according to any one of claims 1 to 11.

14. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to perform the method according to any one of claims 1 to 11.

15. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.

Images