User peer relationship identification method, readable storage medium and computer program product

Abstract

The embodiment of the invention discloses a user peer relationship identification method, a readable storage medium and a computer program product. The method comprises the following steps: determining a user path set associated with an entity store, grouping a plurality of user paths in the set in pairs, and respectively obtaining respective associated behavior fragment sets for a first user path and a second user path in each group; the behavior fragment set comprises associated behavior fragments of a single consumer user under different image acquisition devices and starting and ending time of the behavior fragments; performing time-space co-occurrence judgment on the behavior fragments associated with the first user path and the second user path under the same image acquisition equipment according to the starting time and the ending time, and establishing a co-occurrence connection relationship between the behavior fragments with the time-space co-occurrence; and according to the number of the behavior segments for establishing the co-occurrence connection relationship, obtaining a peer relationship identification result between the two user paths in the group so as to analyze shop customer group structure information according to the identification result and optimize a shop management strategy.

Description

Technical Field

[0001] This application relates to the field of commodity information service technology, and in particular to a method for identifying user peer relationships, a computer-readable storage medium, an electronic device, and a computer program product. Background Technology

[0002] In offline shopping scenarios, the relationship between consumers is an important dimension for store data analysis. Currently, most methods rely on images captured by cameras in the checkout area to identify shared checkout behavior and thus determine the relationship between multiple users.

[0003] However, in practical applications, when multiple people shop together, only one user may go to checkout. Furthermore, when there are queues at the checkout area, a single image captured by the camera may include multiple users, making it difficult to identify who is shopping together. All of these factors affect the accuracy of user relationship analysis.

[0004] How to accurately and effectively identify peer relationships among users has become a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0005] This application provides a method and apparatus for identifying user peer relationships, a method and apparatus for digital management of physical stores, a computer-readable storage medium, an electronic device, and a computer program product, which can accurately and effectively identify user peer relationships.

[0006] This application provides the following solution: A method for identifying peer relationships among users, the method comprising: Determine the set of user paths associated with physical stores, the set of user paths including multiple user paths generated for consumer users located in the store within a preset time period; The multiple user paths are grouped into pairs, and for the first user path and the second user path in each group, a set of associated behavior segments is obtained. The set of behavior segments includes the behavior segments associated with a single consumer user under different image acquisition devices and the start and end times of the behavior segments. The behavior segments are obtained by segmenting the consumer user's movement trajectory in the physical store. Based on the start and end times of the behavior segments, the spatiotemporal co-occurrence of the behavior segments associated with the first user path and the second user path under the same image acquisition device is determined, and a co-occurrence connection relationship is established between the behavior segments with spatiotemporal co-occurrence. Based on the number of behavioral fragments with co-occurrence connections, the peer relationship identification results between the first user path and the second user path in the group are obtained, so as to analyze the customer structure information of the physical store based on the peer relationship identification results, and optimize the store management strategy based on the customer structure information.

[0007] The method further includes: If multiple groups with peer relationships are identified based on the peer relationship identification results, then the multiple groups with peer relationships are aggregated, and at least two groups that include the same user path are merged into a multi-user group. If at least one group without peer relationship is identified based on the peer relationship identification result, the user paths in the at least one group without peer relationship are divided into different single customer groups.

[0008] The set of behavioral fragments associated with the user path is obtained in the following manner: Obtain cross-view behavior information of the consumer user in the physical store. The cross-view behavior information includes multiple user trajectories and a user path generated by merging the multiple user trajectories. The user trajectory is used to represent the sequence of movement trajectories formed by the continuous tracking of the consumer user within the field of view of a single image acquisition device. After sorting the multiple user trajectories in chronological order, the behavioral segments associated with the consumer users under different image acquisition devices are segmented, and the start and end times of the behavioral segments are obtained. The behavioral segments are generated by merging at least one user trajectory formed within the field of view of a single image acquisition device. The start and end times of the behavioral segments are determined according to the acquisition time of the at least one user trajectory. The acquisition time of the user trajectory includes the acquisition timestamp of the first trajectory point and the acquisition timestamp of the last trajectory point in the movement trajectory sequence.

[0009] The step of sorting the multiple user trajectories in chronological order includes: The multiple user trajectories are sorted chronologically based on the collection timestamp of the first trajectory point in the user trajectory.

[0010] The method further includes: After obtaining the temporal sorting results of the multiple user trajectories, if there is a time reversal and / or spatial disorder between two adjacent user trajectories, the abnormal user trajectories are filtered out.

[0011] The step of segmenting the consumer user's behavior segments associated with different image acquisition devices includes: Based on the detected device switching event of the image acquisition device, the user's at least one user trajectory under a single image acquisition device is segmented into a behavior segment.

[0012] The method further includes: Obtain the time window threshold and the collection timestamp associated with each trajectory point in the user trajectory; If the time difference between the acquisition timestamps of two adjacent trajectory points exceeds the time window threshold, then the behavior segmentation continues between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

[0013] The method further includes: Obtain the location jump threshold and the spatial location associated with each trajectory point in the user trajectory; If the spatial span between two adjacent trajectory points exceeds the position jump threshold, then the behavior segmentation continues between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

[0014] The acquisition of cross-view behavior information of consumer users in physical stores includes: The video streams captured by different image acquisition devices in the physical store are obtained as input to the multi-target tracking (MOT) system, which then performs target detection and tracking processing with the consumer user as the target. Obtain cross-view behavior information of different consumer users in the physical store, output by the MOT system.

[0015] The step of determining the spatiotemporal co-occurrence of behavioral segments associated with the first user path and the second user path under the same image acquisition device based on the start and end times of the behavioral segments includes: Determine the behavioral segments associated with the first user path and the second user path under the same image acquisition device. If the start and end times of the two behavioral segments overlap, and the start time difference between the two behavioral segments does not exceed a first time threshold and / or the end time difference between the two behavioral segments does not exceed a second time threshold, then determine that the two behavioral segments have spatiotemporal co-occurrence.

[0016] The step of obtaining the peer relationship identification result between the first user path and the second user path based on the number of behavioral segments with established co-occurrence connections includes: The path similarity S between the first user path and the second user path is calculated based on the number N of behavior segments with co-occurrence connections, the number T1 of behavior segments associated with the first user path, and the number T2 of behavior segments associated with the second user path.

[0017] If the path similarity is not lower than the similarity threshold, an identification result indicating that the first user path and the second user path have a peer relationship is obtained.

[0018] The method further includes: Obtain the co-occurrence threshold; The path similarity S is calculated when both the co-occurrence N / T1 of the first user path and the co-occurrence N / T2 of the second user path are not less than the co-occurrence threshold.

[0019] The method further includes: When it is necessary to improve the accuracy of peer relationship identification results, the following operation options are provided to adjust parameters for at least one of the following thresholds: similarity threshold, co-occurrence threshold, first time threshold, second time threshold, and the logical relationship between the start time difference and end time difference of two behavioral segments when performing spatiotemporal co-occurrence judgment; The adjusted parameter values ​​are obtained through the operation options, and the peer relationship is identified based on the adjusted parameter values.

[0020] A digital management method for physical stores, the method comprising: In the process of digital analysis of physical stores, the identification results of user peer relationship identification for the physical stores are obtained. The identification results are obtained by grouping the user paths of consumer users in the store in pairs within a preset time period, and then performing spatiotemporal co-occurrence judgment based on multiple behavioral segments associated with the first user path and the second user path in each group. The behavioral segments are obtained by segmenting the movement trajectory of consumer users in the physical store. The customer structure information of the physical store is determined based on the identification results, and the management strategy of the physical store is optimized based on the customer structure information.

[0021] A user peer relationship identification device, the device comprising: The user path set determination unit is used to determine the user path set associated with a physical store, wherein the user path set includes multiple user paths generated for consumer users located in the store within a preset time period. The behavior segment set acquisition unit is used to group the multiple user paths into pairs, and for the first user path and the second user path in each group, obtain their respective associated behavior segment sets; the behavior segment set includes the behavior segments associated with a single consumer user under different image acquisition devices and the start and end times of the behavior segments, and the behavior segments are obtained based on the consumer user's movement trajectory in the physical store. The spatiotemporal co-occurrence determination unit is used to determine the spatiotemporal co-occurrence of the first user path and the second user path associated with the same image acquisition device based on the start and end times of the behavior segments, and to establish a co-occurrence connection relationship between the behavior segments with spatiotemporal co-occurrence. The identification result acquisition unit is used to obtain the peer relationship identification result between the first user path and the second user path in the group based on the number of behavioral fragments with established co-occurrence connection relationship, so as to analyze the customer group structure information of the physical store based on the peer relationship identification result, and optimize the store management strategy based on the customer group structure information.

[0022] A digital management device for physical stores, the device comprising: The peer relationship identification result acquisition unit is used to obtain the identification result of the peer relationship identification of the physical store during the process of digital analysis of the physical store. The identification result is obtained by grouping the user paths of consumer users in the store in pairs within a preset time period, and then performing spatiotemporal co-occurrence judgment based on multiple behavioral segments associated with the first user path and the second user path in each group. The behavioral segments are obtained based on the segmentation of the consumer user's movement trajectory in the physical store. The management strategy optimization unit is used to determine the customer structure information of the physical store based on the identification results, and optimize the management strategy of the physical store based on the customer structure information.

[0023] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of any of the preceding methods.

[0024] An electronic device, comprising: One or more processors; and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of any of the preceding methods.

[0025] A computer program product includes a computer program / computer executable instructions that, when executed by a processor in an electronic device, implement the steps of any of the preceding methods.

[0026] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application embodiment is based on the user path generated by consumers during the shopping process in physical stores. The user path is grouped into pairs to identify whether there is a user companion relationship.

[0027] First, we can obtain the sets of behavioral segments associated with the first and second user paths in the group. Each set can include multiple behavioral segments, obtained by segmenting the user's movement trajectories under different cameras in the physical store, representing the user's dwell behavior under a single camera. Second, we can determine the spatiotemporal co-occurrence of behavioral segments under the same camera in both sets. For behavioral segments with spatiotemporal co-occurrence, we can establish co-occurrence connections, indicating that the two user paths overlap at this point. Finally, based on the number of behavioral segments with established co-occurrence connections, we can obtain the results of identifying the co-location relationship between the two paths.

[0028] This application converts users' shared shopping behavior into the frequency of spatiotemporal co-occurrence from the camera's perspective, thereby interpreting and identifying users' companion relationships. Compared to existing technologies that rely on shared checkout behavior for companion relationship identification, this application's solution provides a more complete description of the user's shopping process, thus achieving higher accuracy and interpretability in companion relationship identification.

[0029] Of course, any product implementing this application does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a schematic diagram of the user peer relationship identification system provided in the embodiments of this application; Figure 2 This is a flowchart of the user peer relationship identification method provided in the embodiments of this application; Figure 3 This is a flowchart of the digital management method for physical stores provided in the embodiments of this application; Figure 4This is a schematic diagram of the user peer relationship identification device provided in the embodiments of this application; Figure 5 This is a schematic diagram of the digital management device for physical stores provided in the embodiments of this application; Figure 6 This is a schematic diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0033] To improve the accuracy of identifying peer relationships in offline shopping scenarios, this application provides a peer relationship identification scheme based on the complete access route generated by a consumer user while browsing a physical store.

[0034] Specifically, from a system architecture perspective, the user peer relationship identification system in this application embodiment can be as follows: Figure 1 As shown, it includes an image acquisition device, a multi-object tracking (MOT) system, and a server.

[0035] The image acquisition equipment is deployed within the physical store to collect video streams from different areas and send them to the MOT system. The MOT system can perform user detection and tracking on multi-source video stream data. For different consumers, it outputs multiple discrete user trajectories generated during the user's shopping process, as well as a single user path generated by merging these trajectories. This information is then provided to the server as cross-view behavior information of the user in the physical store for identifying peer relationships.

[0036] The server in this application embodiment can be deployed in a physical store to identify peer relationships in the user paths generated by the store; or, the server can be deployed in the cloud, and this application does not limit it in this way.

[0037] The implementation process of the server-side user peer relationship identification method is explained below with specific examples. See [link to documentation]. Figure 2 The flowchart shown may include: S201: Determine the set of user paths associated with physical stores, wherein the set of user paths includes multiple user paths generated for consumer users located in the store within a preset time period.

[0038] S202: Divide the multiple user paths into pairs, and for the first user path and the second user path in each group, obtain their respective sets of associated behavior segments. The sets of behavior segments include the behavior segments associated with a single consumer user under different image acquisition devices and the start and end times of the behavior segments. The behavior segments are obtained based on the segmentation of the consumer user's movement trajectory in the physical store.

[0039] This application embodiment can identify the peer relationship between any two user paths generated by a physical store. A user path can correspond to the browsing path of a consumer user in a physical store.

[0040] Specifically, a set of user paths associated with the physical store can be obtained. This set may include multiple user paths generated for consumers located in the store within a preset time period. For example, the preset time period can be one day, and user paths can be generated separately for all users who visit the store on that day, obtaining the user path corresponding to each user and adding it to the user path set. In this embodiment, user paths for different users can be generated through the MOT system, and the specific implementation process can be found in the following description.

[0041] For multiple user paths in the user path set, they can be grouped in pairs, and peer relationship analysis can be performed on the two user paths in each group. The two paths in the group can be called the first user path and the second user path. The set of behavioral segments associated with each path can be obtained. Based on the matching of behavioral segments in the two sets, the overlap similarity between the two paths is determined, and the peer relationship identification result between the two consumer users corresponding to the two paths is obtained accordingly.

[0042] As an example, the set of behavioral fragments associated with a user path can be obtained in the following way.

[0043] First, it can obtain cross-view behavior information of consumers in physical stores.

[0044] Specifically, video streams captured by different image acquisition devices in a physical store can be used as input to the MOT system. Correspondingly, the MOT system can target consumers, performing target detection and continuous target tracking within the video stream sequences captured by multiple cameras, thereby outputting cross-view behavior information of different consumers in the physical store. The MOT system can achieve target detection based on user appearance features, thus obtaining cross-view behavior information of different users while protecting user privacy, and based on this, identifying user peer relationships.

[0045] In this application, cross-view behavior information may include: multiple discrete user tracks for the same consumer user, and a user path generated by merging multiple user tracks.

[0046] Among them, the user trajectory track refers to the sequence of movement trajectories formed by a consumer user being continuously tracked within the field of view of a single image acquisition device.

[0047] In one scenario, if a single user is continuously tracked by a camera, for example, when the MOT system tracks user A based on the video stream captured by camera 1, it continuously tracks user A from the moment user A enters the field of view of camera 1 until user A leaves the field of view of camera 1, thus obtaining a user trajectory track for user A under camera 1. 1A .

[0048] In another scenario, if a single user is not continuously tracked by a particular camera, a tracking interruption occurs. For example, when the MOT system is tracking user B based on the video stream captured by camera 1, user B may be obstructed by other users or objects such as store shelves, causing a tracking interruption. However, in subsequent target detection, user B is detected again based on the video stream captured by camera 1 and continues to be tracked until user B leaves the field of view of camera 1. This results in two user trajectories for user B under camera 1. 1B1 and track 1B2 .

[0049] In other words, based on the number of times a user is continuously tracked within the same camera's field of view, at least one user trajectory under that camera can be generated.

[0050] Each user path can include multiple trajectory points. Furthermore, under a unified clock system, the acquisition timestamp associated with each trajectory point can be obtained; and based on the camera spatial topology, the spatial location associated with each trajectory point can be obtained.

[0051] For user trajectory track 1A In this context, the first trajectory point in the sequence represents the first time user A enters the field of view of camera 1, and the timestamp of user A's first entry into the field of view and the spatial location of the first entry into the field of view can be obtained; the last trajectory point in the sequence represents the first time user A leaves the field of view of camera 1, and the timestamp of user A's first departure from the field of view and the spatial location of the first departure from the field of view can be obtained.

[0052] For user trajectory track 1B1In this context, the first trajectory point in the sequence represents the first time user B enters the field of view of camera 1, corresponding to the timestamp of user B's first entry into the field of view and the spatial location of that first entry. The last trajectory point in the sequence represents the first time user B leaves the field of view of camera 1, i.e., the last time user B was detected within the field of view of camera 1 before tracking was interrupted, corresponding to the timestamp of user B's first exit from the field of view and the spatial location of that first exit. For user trajectory tracking... 1B2 In this sequence, the first trajectory point indicates that user B re-enters the field of view of camera 1, and the timestamp of user B re-entering the field of view and the spatial location of re-entering the field of view can be obtained; the last trajectory point indicates that user B leaves the field of view of camera 1 again, and the timestamp of user B leaving the field of view and the spatial location of re-leaving the field of view can be obtained.

[0053] After the MOT system identifies multiple user trajectories of the same user under different cameras, it can merge and associate the user's cross-camera trajectories to reconstruct the user's complete access path within the physical store, i.e., the user path in this embodiment. Simultaneously, it can also associate multiple discrete user trajectories of the same user with the user path identifier path_code.

[0054] Secondly, after sorting the multiple user trajectories in chronological order, the behavioral segments associated with the consumer users under different image acquisition devices are segmented, and the start and end times of the behavioral segments are obtained. The behavioral segments are generated by merging at least one user trajectory formed within the field of view of a single image acquisition device. The start and end times of the behavioral segments are determined according to the acquisition time of the at least one user trajectory. The acquisition time of the user trajectory includes the acquisition timestamp of the first trajectory point and the acquisition timestamp of the last trajectory point in the movement trajectory sequence.

[0055] To improve the accuracy of behavior segmentation, this application can also sort all discrete user trajectories associated with the same path in chronological order under a unified clock system based on the path-trajectory relationship, construct a logically coherent cross-camera access sequence of users in the store, and then perform behavior segmentation.

[0056] To ensure the accuracy of the sorting results, multiple user trajectories can be sorted chronologically based on the timestamp of the first trajectory point in the user trajectory. However, sorting by the timestamps of other trajectory points, such as the last trajectory point, may lead to missorting in special cases. This is because, in practical applications, the images captured by cameras may overlap. For example, user C might first enter the field of view of camera 1 and later appear in the field of view of camera 2, meaning user C is currently located within the overlapping area of ​​the fields of view of cameras 1 and 2. If user C subsequently leaves the field of view of camera 2 before leaving the field of view of camera 1, then when sorting by the last trajectory point of user C within the camera's field of view, user C's trajectory at camera 2 would appear earlier in the chronological order than user C's trajectory at camera 1, which clearly does not match user C's actual movement path. According to the solution in this application, sorting by the first trajectory point of user C entering the camera's field of view yields a more accurate sorting result for the user trajectories.

[0057] Optionally, after obtaining the time-series ranking results of multiple user trajectories, user trajectories exhibiting the following anomalies can be filtered: 1. Time Reversal Anomaly Detection If there is a time reversal between two adjacent user trajectories, then the abnormal user trajectories are filtered out.

[0058] Typically, after sorting multiple user trajectories according to the scheme in this application, the timestamps of adjacent trajectories will increase sequentially. However, in practical applications, two temporally adjacent user trajectories may have a situation where the timestamp of the later-sorted user trajectory is earlier than the timestamp of the earlier-sorted user trajectory, resulting in a reverse time sequence anomaly. Correspondingly, abnormal user trajectories can be filtered out based on the cause of the anomaly.

[0059] As an example, when sorting based on a unified clock system, the occurrence of time reversal anomalies is mostly due to trajectory matching deviations in the MOT system. Specifically, when performing trajectory association across cameras, the MOT system may mistakenly stitch together the trajectories of different users, causing time reversal issues. Correspondingly, the erroneously stitched user trajectories of other users can be filtered out, and subsequent processing can be performed based on the filtered time-series sorting results.

[0060] 2. Spatial disorder anomaly detection If there is spatial disorder between two adjacent user trajectories, the abnormal user trajectories will be filtered out.

[0061] Typically, after chronologically sorting multiple user trajectories according to the proposed solution, the spatial positions of adjacent trajectories gradually change as the user's location within the store changes. However, in practical applications, if, based on the camera's spatial topology, two temporally adjacent user trajectories show a significant jump in their spatial position within the store—a situation of spatial disorder—abnormal user trajectories can be filtered out based on the cause of the anomaly.

[0062] As an example, spatial disorder could also be caused by trajectory matching errors in the MOT system. Specifically, when performing cross-camera trajectory association, the MOT system might mistakenly stitch together the trajectories of different users, leading to spatial disorder. Correspondingly, the erroneously stitched user trajectories of other users can be filtered out, and subsequent processing can be performed based on the filtered temporal ranking results.

[0063] As described above, after obtaining the time-series sorting results (or filtered time-series sorting results) of multiple user trajectories, the user can be divided into a behavior segment based on the device switching event of the image acquisition device, which is at least one user trajectory formed by the consumer user within the field of view of a single image acquisition device.

[0064] In other words, all user trajectories under a single camera can be merged to obtain a segment of the user's behavior under that camera. Furthermore, the start time of the behavior segment is determined by the timestamp of the user's first entry into the camera's field of view (i.e., the earliest associated user trajectory under that camera), and the timestamp of the first trajectory point in that sequence is used as the start time of the behavior segment. Conversely, the end time of the behavior segment is determined by the timestamp of the user's last exit from the camera's field of view (i.e., the latest associated user trajectory under that camera), and the timestamp of the last trajectory point in that sequence is used as the end time of the behavior segment. This yields the start and end times (TimeSpan) of the behavior segment.

[0065] For example, in the case of user A mentioned above, if it is detected that user A leaves the field of view of camera 1 and enters the field of view of camera 2, it can be determined that a camera switching event has occurred. The user trajectory formed from the time user A enters the field of view of camera 1 to the time user A leaves the field of view of camera 1 can then be tracked. 1A Divide into a behavior segment 1A and obtain the behavior segment. 1A Start and end times TimeSpan 1A For behavioral segments 1A User A's behavior while hovering over camera 1 can be represented using the following structured data: path A-camera1-TimeSpan 1A .

[0066] For example, in the user B example above, if it is detected that user B leaves the field of view of camera 1 and enters the field of view of camera 2, it can be determined that a camera switching event has occurred. The user trajectory formed from the time user B enters the field of view of camera 1 to the time he leaves the field of view of camera 1 can be recorded as a track. 1B1 and track 1B2 Divide into a behavior segment 1B and obtain the behavior segment. 1B Start and end times TimeSpan 1B For behavioral segments 1B User B's behavior while hovering over camera 1 can be represented using the following structured data: path B -camera1-TimeSpan 1B .

[0067] Understandably, if multiple camera switches are detected between two cameras installed in adjacent areas along a user's path, it can be determined that the user is repeatedly circling back and forth between these two areas. For example, if user E moves from the fruit area to the vegetable area and then returns to the fruit area, behavioral segments can be segmented based on time breakpoints and motion continuity analysis.

[0068] In this example, the following three behavioral segments can be obtained: When user E leaves the field of view of camera 3 in the fruit area and enters the field of view of camera 4 in the vegetable area, the behavioral segment formed by user E's first entry into the fruit area under camera 3 can be obtained. 3E1 When user E leaves the field of view of camera 4 in the vegetable area and then re-enters the field of view of camera 3 in the fruit area, the behavior segment formed by user E under camera 4 can be obtained. 4E When user E is detected leaving the fruit area again and entering the field of view of camera 3 and entering the field of view of other cameras, the behavior segment formed by user E entering the fruit area a second time under camera 3 can be obtained. 3E2 .

[0069] In practical applications, at least one user trajectory under a camera can be segmented into a behavior segment, as illustrated above. Optionally, this application can further combine a time window threshold and / or a location jump threshold to characterize the user's dwell behavior under the camera with more granular detail. That is, at least one user trajectory under a camera can be segmented into at least two behavior segments. Specific examples will be provided below for explanation.

[0070] In one example, a time window threshold can be obtained to perform more refined segmentation of behavior from a time dimension. Specifically, if the time difference between the acquisition timestamps of two adjacent trajectory points exceeds the time window threshold, behavior segmentation continues between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

[0071] For example, in the example of user A mentioned above, if the user's trajectory track... 1A If the timestamp interval between adjacent trajectory points A1 and A2 is relatively long, exceeding a preset time window threshold, then behavior segments can be obtained between trajectory points A1 and A2, resulting in two independent behavior segments. These two independent behavior segments represent user A's behavior at camera 1.

[0072] In addition, the time window threshold can also be used to identify long intervals of missing data between two user trajectories.

[0073] In practical applications, if the same user experiences a prolonged tracking interruption under a single camera, this abnormal target loss will result in a long gap between the two user trajectories before and after the interruption.

[0074] For example, during the process of tracking user D based on the video stream captured by camera 1, after a period of continuous tracking, the user is lost because camera 1 fails to capture the image of user D, or although it captures the image containing user D, it fails to successfully identify user D.

[0075] In the face of this target loss situation, if user D is detected again in the video stream captured by camera 1 and tracking continues within a preset time period after the user is lost (specifically, within a time window threshold), then this target loss can be considered a normal situation. The two user trajectories generated before and after the interruption can be merged and segmented into a single behavior segment. Simultaneously, the entry and exit times of the trajectories under the same camera can be shared; that is, the entry time of the previous user trajectory and the exit time of the next user trajectory can be used as the start and end times of this behavior segment.

[0076] If user D is not detected in the video stream captured by camera 1 within a preset time period after being lost, it can be determined that there is an anomaly in the target loss. A long interval of missing user trajectory exists between the two user trajectories generated before and after the interruption. The two trajectories before and after the interruption can be divided into independent behavioral segments, and their start and end times can be determined separately. This avoids the erroneous assumption that user D lingered under camera 1 for an extended period, ensuring the accuracy of representing lingering behavior through behavioral segments.

[0077] In another example, a location jump threshold can be obtained to perform more refined behavior segmentation from a spatial dimension. Specifically, if the spatial span between two adjacent trajectory points exceeds the location jump threshold, behavior segmentation continues between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

[0078] For example, in the example of user B mentioned above, if the user's trajectory track... 1B1 With track 1B2 Among the included trajectory points, track 1B1 The last trajectory point B1 and track 1B2 If the spatial range of the first trajectory point B2 is too large, exceeding the preset position jump threshold, then the behavior segment can be divided between trajectory points B1 and B2 to obtain two independent behavior segments. These two independent behavior segments are used to represent user B's stationary behavior at camera 1.

[0079] In this way, we can obtain the behavioral fragments associated with different user paths and generate corresponding sets of behavioral fragments. For example, user A's path... A If the path is divided into 10 behavior segments, then... A When a path is the first user path, its associated set of behavioral fragments includes 10 behavioral fragments. Similarly, if user B's path... B If the path is divided into 15 behavior segments, then... B When used as a second user path, its associated set of behavioral fragments includes 15 behavioral fragments.

[0080] S203: Based on the start and end times of the behavior segments, determine the spatiotemporal co-occurrence of the behavior segments associated with the first user path and the second user path under the same image acquisition device, and establish a co-occurrence connection relationship between the behavior segments with spatiotemporal co-occurrence.

[0081] In this embodiment, the spatiotemporal co-occurrence frequency of behavioral segments in any two paths can be used to analyze and identify peer relationships. Spatiotemporal co-occurrence refers to the phenomenon that different events, objects, or features (in this application, behavioral segments representing a user's dwell time under the camera) appear simultaneously and are related to each other within a specific time and space range, emphasizing the dynamic interaction of co-occurrence relationships in the spatiotemporal dimension.

[0082] Specifically, path pairs that satisfy the spatiotemporal co-occurrence constraint can be identified in the following way: If the start and end times of the two user paths overlap, and the start time difference between the two user paths does not exceed a first time threshold and / or the end time difference between the two user paths does not exceed a second time threshold, then it can be determined that the two user paths have spatiotemporal co-occurrence.

[0083] Using user path A with path B For example, the two are associated with a behavior segment under camera 1. 1A with segment 1B It can be based on TimeSpan 1A and TimeSpan 1B Perform a spatiotemporal co-occurrence determination. If TimeSpan 1A With TimeSpan 1B If there is a temporal overlap, and the start time difference does not exceed a first time threshold and / or the end time difference does not exceed a second time threshold, then a segment can be determined. 1A with segment 1B There is spatiotemporal co-occurrence between them. The values ​​of the first time threshold and the second time threshold can be the same or different, and can be set according to actual usage requirements. For example, both can be set to 15s.

[0084] Additionally, it should be noted that the logical relationship "AND / OR" between the start and end time differences of two behavioral segments can be flexibly adjusted according to actual usage requirements. If the logical relationship is "AND," then the start time difference must not exceed a first time threshold, and the end time difference must not exceed a second time threshold. This helps improve the accuracy of spatiotemporal co-occurrence determination. If the logical relationship is "OR," then either the start time difference must not exceed the first time threshold, or the end time difference must not exceed the second time threshold; satisfying either one is sufficient. This helps improve the recall rate of co-occurrence relationships, identifying more pairs of behavioral segments with spatiotemporal co-occurrence.

[0085] In this application, when it is determined that two behavioral segments have spatiotemporal co-occurrence, a bidirectional co-occurrence connection can be established between these two behavioral segments. That is, for a segment... 1A In other words, segment 1B It has a co-occurrence relationship with it; at the same time, for segment 1B In other words, segment 1A It also has a co-occurrence relationship with it.

[0086] As an example, this application can also generate a set of co-occurrence pairs based on the spatiotemporal co-occurrence judgment results for behavior segments that establish co-occurrence connection relationships, providing key data support for subsequent identification of user peer relationships.

[0087] S204: Based on the number of behavioral segments with co-occurrence connections, obtain the peer relationship identification results between the first user path and the second user path in the group, so as to analyze the customer structure information of the physical store based on the peer relationship identification results, and optimize the store management strategy based on the customer structure information.

[0088] The embodiments of this application can count the frequency of co-occurrence pairs in two paths. The higher the co-occurrence frequency, the higher the overlap and similarity between the two paths, and the greater the possibility that the users corresponding to the two paths are peers.

[0089] To improve the accuracy of peer relationship identification results, this application can calculate path similarity S by normalizing the number of times the first user path and the second user path appear in the global camera (i.e., the number of segments T1 and T2 in the set of behavioral segments associated with each of the two paths) and the co-occurrence frequency (i.e., the number of behavioral segments N with co-occurrence connection relationship).

[0090] As an example, path similarity S can be calculated as follows:

[0091] This example demonstrates how minimum co-occurrence frequency normalization can effectively suppress the dominance effect of high-frequency paths, reduce the bias caused by differences in individual activity levels, and obtain a relatively stable co-occurrence trend between the two paths.

[0092] As another example, the path similarity S can also be calculated preferably in the following manner:

[0093] This example uses 1 / 2 as the path weight and calculates the average similarity by weighting. This can also effectively suppress the high-frequency path dominance effect, weaken the bias caused by individual activity differences, and obtain a relatively stable co-occurrence trend between the two paths.

[0094] In this example, a co-occurrence threshold can also be obtained. If the co-occurrence N / T1 of the first user path and the co-occurrence N / T2 of the second user path are both not less than the preset co-occurrence threshold, the path similarity S is calculated. The co-occurrence threshold can be obtained through comparative experiments and adjusted according to the actual grouping effect. This application embodiment does not specifically limit the value of the co-occurrence threshold.

[0095] For example, in the example above, user C is a pedestrian who happens to pass by a physical store. If user C is detected and tracked in the video streams captured by two cameras, two behavioral segments are obtained. When performing peer relationship analysis based on the user paths of user C and user D, user D is segmented into 30 behavioral segments, and these segments overlap with user C's trajectory under the same two cameras, meaning their co-occurrence frequency is 2. To further reduce the impact of user C's high-frequency path on the accuracy of peer relationship identification, this application sets a co-occurrence threshold. If user D's co-occurrence frequency 2 / 30 is lower than this preset threshold, no overlap similarity calculation is performed; instead, it is directly determined that the two do not have a peer relationship.

[0096] In other words, a co-occurrence of a single user path is considered reasonable and normal only if its co-occurrence exceeds a preset co-occurrence threshold. In practical applications, a co-occurrence of 1 for a user path could be normal data generated from normal browsing, or it could be abnormal data generated under special circumstances, as shown in the example above. Identifying anomalies with a co-occurrence of 1 from a system perception perspective is quite difficult. Therefore, this application identifies abnormal data for user paths with a co-occurrence of 1 by setting a co-occurrence threshold to identify and limit the co-occurrence of another user path.

[0097] Furthermore, the embodiments of this application can also configure a similarity threshold. If the path similarity between the first user path and the second user path is not lower than a preset similarity threshold, then the identification result that the two paths have a peer relationship can be obtained; conversely, if the path similarity is lower than the preset similarity threshold, then the identification result that the two paths do not have a peer relationship can be obtained.

[0098] Taking a similarity threshold of 0.5 as an example, in the example of user A and user B above, if the co-occurrence frequency of the two is 6, then the similarity S = 1 / 2 * (6 / 10 + 6 / 15) = 0.5, which is not lower than the similarity threshold. It can be determined that there is a peer relationship between user A's first user path and user B's second user path.

[0099] Optionally, for two paths that are related as peers, they can be grouped into a single two-person customer group. For example, in the example above, a two-person customer group [AB] can be generated for users A and B. For two paths that are not related as peers, two independent single-person customer groups can be obtained. For example, in the example above, single-person customer group [C] and single-person customer group [D] can be generated for users C and D, respectively.

[0100] As described above, for user paths in physical stores, peer-to-peer relationship identification is performed in pairs. After obtaining the corresponding identification results, equivalent aggregation can be performed on the two-person customer groups. Specifically, if multiple groups with peer-to-peer relationships are identified based on the peer-to-peer relationship identification results, these groups can be aggregated, merging at least two groups that include the same user path into a multi-person customer group.

[0101] For example, given a two-person customer group [AB] consisting of users A and B, and another two-person customer group [BE] consisting of users B and E, these three groups can be dynamically merged into a multi-person customer group [ABE] based on the transitivity of peer relationships, achieving structured aggregation and redundancy removal of group relationships. As an example, a union-find algorithm or a graph connected component algorithm can be used to dynamically merge customer groups.

[0102] After aggregating and deduplicating customer groups, stable and semantically consistent customer grouping results can be obtained. Each customer group represents a group of highly collaborative individuals traveling together. For groups without travel companion relationships, the user paths within the group can be considered as independent user access behaviors and assigned to different single-person customer groups. In this way, a complete customer social graph covering both multi-person and single-person customer groups can be obtained, enabling the explicit expression of "implicit social relationships" in offline physical store scenarios.

[0103] Furthermore, this application can also perform a rationality analysis on the aggregated multi-person customer groups and adjust some parameters in the peer relationship identification process accordingly. Taking the rationality analysis of the number of associated users in a single multi-person customer group as an example, by statistically analyzing the distribution of the number of people shopping together in actual shopping scenarios, a corresponding preset user threshold can be configured. If the aggregated user peer relationship indicates that the number of peer users exceeds the preset user threshold, and such aggregated multi-person customer groups with abnormal user numbers occur multiple times, it can be determined that the accuracy of the user peer relationship identification results needs to be improved. Parameters of some thresholds in the identification process of this application can be adjusted to tighten the judgment threshold for two-person customer groups.

[0104] In addition, this application can also save the video stream captured by the store's cameras, i.e., the video stream used for peer relationship identification, for example, by saving it to a network video recorder (NVR). After obtaining the customer grouping results, the accuracy and reasonableness of the grouping results can be manually verified by combining the saved video stream. If the manual verification determines that the customer grouping includes "pseudo-peer users" with weak spatiotemporal correlation, some threshold parameters in the identification process of this application can be adjusted to improve the accuracy of the user peer relationship identification results.

[0105] Specifically, if it is determined that there is a need to improve the accuracy of peer relationship identification results, the following parameter adjustment options can be provided: Similarity threshold: You can increase the value of the similarity threshold to determine that two user paths with a higher degree of path overlap are related as peers; Co-occurrence threshold: The value of the co-occurrence threshold can be increased. Similarity calculation can only be performed when the co-occurrence of two user paths meets the higher requirement of the co-occurrence threshold. First time threshold and second time threshold: The values ​​of the first time threshold and / or the second time threshold can be reduced to identify two behavioral segments with overlapping start and end times and smaller time difference as having spatiotemporal co-occurrence, thereby increasing the identification threshold for co-occurrence pairs; When determining spatiotemporal co-occurrence, the logical relationship between the start time difference and the end time difference of two behavioral segments can be adjusted to an "AND" relationship. This requires that the start time difference does not exceed the first time threshold and the end time difference does not exceed the second time threshold, which can also increase the threshold for recognizing co-occurrence pairs.

[0106] In this way, the threshold to be adjusted can be determined through the operation options, and the adjusted parameter value corresponding to the threshold can be submitted. Then, peer relationship identification can be performed again based on the adjusted parameter value until it is determined that the obtained grouping result matches the real user peer relationship.

[0107] In summary, this application embodiment is based on the user path generated by consumers during the shopping process in physical stores, and the user path is grouped into pairs to identify whether there is a user companion relationship.

[0108] First, we can obtain the sets of behavioral segments associated with the first and second user paths in the group. Each set can include multiple behavioral segments, obtained by segmenting the user's movement trajectories under different cameras in the physical store, representing the user's dwell behavior under a single camera. Second, we can determine the spatiotemporal co-occurrence of behavioral segments under the same camera in both sets. For behavioral segments with spatiotemporal co-occurrence, we can establish co-occurrence connections, indicating that the two user paths overlap at this point. Finally, based on the number of behavioral segments with established co-occurrence connections, we can obtain the results of identifying the co-location relationship between the two paths.

[0109] This application converts users' shared shopping behavior into the frequency of spatiotemporal co-occurrence from the camera's perspective, thereby interpreting and identifying users' companion relationships. Compared to existing technologies that rely on shared checkout behavior for companion relationship identification, this application's solution provides a more complete description of the user's shopping process, thus achieving higher accuracy and interpretability in companion relationship identification.

[0110] In this embodiment of the application, after obtaining the peer relationship identification results between two user paths in different groups, the customer structure information of the store can be analyzed based on the identification results.

[0111] As an example, store customer structure information can be reflected in the size distribution of the store's customer base, such as the proportion of single-person, two-person, and multi-person customer groups. If a store has a high proportion of multi-person customer groups, it can be determined that the store's user base primarily engages in group activities, allowing for targeted optimization of store management strategies to better meet the needs of users in group shopping scenarios. Conversely, if a store has a high proportion of single-person customer groups, it can be determined that the store's user base primarily engages in individual activities, similarly allowing for targeted optimization of store management strategies to better meet the needs of users in individual shopping scenarios. These store management strategies can include product selection strategies, shelf display strategies, and so on.

[0112] As an example, store customer structure information can be reflected in customer attributes across different dimensions. These attributes can be categorized by social attributes, such as family groups, couples, and friends; or by consumption behavior, such as high-frequency, medium-frequency, and low-frequency consumers; or by value contribution, such as high-repurchase groups, potential customers, and general customers. Customer attributes can be determined based on user needs to obtain more refined store customer structure information.

[0113] Correspondingly, embodiments of this application can also provide a solution for store management based on peer relationship identification results. See also Figure 3 The flowchart shown illustrates that digital management methods for physical stores may include: S301: In the process of digital analysis of physical stores, the identification results of user peer relationship identification for the physical stores are obtained. The identification results are obtained by grouping the user paths of consumer users in the store in pairs within a preset time period, and then performing spatiotemporal co-occurrence judgment based on multiple behavioral segments associated with the first user path and the second user path in each group. The behavioral segments are obtained based on the segmentation of the consumer user's movement trajectory in the physical store. S302: Determine the customer structure information of the physical store based on the identification result, and optimize the management strategy of the physical store based on the customer structure information.

[0114] In other words, according to Figure 2The proposed solution identifies peer relationships within physical stores. After obtaining the identification results, these results can be used as a dimension for store digital analysis, enabling customer structure analysis and the acquisition of customer structure information. Based on this information, targeted store management strategies can be optimized—that is, store management strategies are improved based on customer structure information. As an example, store management strategies can be related to the store's products, such as adjusting the categories and display methods of available goods; and / or, they can be related to store operations, such as adjusting store opening hours and optimizing staffing; and / or, they can be related to store promotion, such as adjusting marketing plans and promotional activities.

[0115] For example, a digital analysis of a physical store might reveal that its customer base is primarily families. Optimized store management strategies could include at least: offering more family-friendly bundled packages, adjusting opening hours to coincide with peak family visit times, assigning staff to cater to children and the elderly, and organizing family-specific promotional events. Optimizing store management strategies based on customer demographics allows for more targeted adjustments, improving both the shopping experience for customers and conversion rates.

[0116] It should be noted that the embodiments of this application may involve the use of user data. In practical applications, user-specific personal data may be used in the scheme described herein within the scope permitted by applicable laws and regulations, provided that it complies with the applicable laws and regulations of the country (e.g., with the user's explicit consent, with the user being properly notified, etc.).

[0117] Corresponding to the foregoing method embodiments, this application also provides a user peer relationship identification device. See also Figure 4 The device may include: User path set determination unit 401 is used to determine the user path set associated with the physical store, wherein the user path set includes multiple user paths generated for consumer users located in the store within a preset time period. The behavior segment set acquisition unit 402 is used to group the multiple user paths into pairs, and for the first user path and the second user path in each group, obtain their respective associated behavior segment sets; the behavior segment set includes the behavior segments associated with a single consumer user under different image acquisition devices and the start and end times of the behavior segments, and the behavior segments are obtained based on the consumer user's movement trajectory in the physical store. The spatiotemporal co-occurrence determination unit 403 is used to determine the spatiotemporal co-occurrence of the first user path and the second user path associated with the same image acquisition device based on the start and end times of the behavior segments, and to establish a co-occurrence connection relationship between the behavior segments with spatiotemporal co-occurrence. The identification result acquisition unit 404 is used to obtain the peer relationship identification result between the first user path and the second user path in the group based on the number of behavioral fragments with established co-occurrence connection relationship, so as to analyze the customer group structure information of the physical store based on the peer relationship identification result, and optimize the store management strategy based on the customer group structure information.

[0118] The device further includes: The multi-user group merging unit is used to aggregate multiple groups with peer relationships when multiple groups with peer relationships are determined based on the peer relationship identification result, and merge at least two groups that include the same user path into a multi-user group. The single-person customer group segmentation unit is used to divide the user paths in the at least one group that does not have a peer relationship into different single-person customer groups when at least one group that does not have a peer relationship is determined based on the peer relationship identification result.

[0119] The device further includes: A cross-view behavior information acquisition unit is used to acquire cross-view behavior information of the consumer user in a physical store. The cross-view behavior information includes multiple user trajectories and a user path generated by merging the multiple user trajectories. The user trajectory is used to represent the sequence of movement trajectories formed by the continuous tracking of the consumer user within the field of view of a single image acquisition device. The behavior segmentation unit is used to sort the multiple user trajectories in chronological order, segment the behavior segments associated with the consumer user under different image acquisition devices, and obtain the start and end times of the behavior segments. The behavior segments are generated by merging at least one user trajectory formed within the field of view of a single image acquisition device. The start and end times of the behavior segments are determined according to the acquisition time of the at least one user trajectory. The acquisition time of the user trajectory includes the acquisition timestamp of the first trajectory point and the acquisition timestamp of the last trajectory point in the movement trajectory sequence.

[0120] Specifically, the behavior segmentation unit can be used to: sort the multiple user trajectories in chronological order based on the collection timestamp of the first trajectory point in the user trajectory.

[0121] Specifically, the behavior segmentation unit can also be used to: after obtaining the temporal sorting results of the multiple user trajectories, if there is a time reversal and / or spatial disorder between two adjacent user trajectories, then filter out the abnormal user trajectories.

[0122] Specifically, the behavior segmentation unit can be used to: based on the detected device switching event of the image acquisition device, segment at least one user trajectory of the consumer user under a single image acquisition device into a behavior segment.

[0123] Specifically, the behavior segmentation unit can also be used to: obtain a time window threshold and the collection timestamp associated with each trajectory point in the user trajectory; if the time difference between the collection timestamps of two adjacent trajectory points exceeds the time window threshold, then continue to perform behavior segmentation between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

[0124] Specifically, the behavior segmentation unit can also be used to: obtain a position jump threshold and the spatial position associated with each trajectory point in the user trajectory; if the spatial span between two adjacent trajectory points exceeds the position jump threshold, then continue to perform behavior segmentation between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

[0125] Specifically, the cross-view behavior information acquisition unit can be used to: acquire video streams acquired by different image acquisition devices in the physical store, and use them as input to the multi-target tracking (MOT) system, which then performs target detection and tracking processing with the consumer user as the target; and acquire cross-view behavior information of different consumer users in the physical store output by the MOT system.

[0126] Specifically, the spatiotemporal co-occurrence determination unit can be used to: determine the behavioral segments associated with the first user path and the second user path under the same image acquisition device; if the start and end times of the two behavioral segments overlap, and the start time difference of the two behavioral segments does not exceed a first time threshold and / or the end time difference of the two behavioral segments does not exceed a second time threshold, then determine that the two behavioral segments have spatiotemporal co-occurrence.

[0127] Specifically, the identification result acquisition unit can be used to: calculate the path similarity S between the first user path and the second user path based on the number N of behavior segments with co-occurrence connection relationships, the number T1 of behavior segments associated with the first user path, and the number T2 of behavior segments associated with the second user path.

[0128] If the path similarity is not lower than the similarity threshold, an identification result indicating that the first user path and the second user path have a peer relationship is obtained.

[0129] Specifically, the identification result acquisition unit can also be used to: obtain a co-occurrence threshold; and calculate the path similarity S when the co-occurrence N / T1 of the first user path and the co-occurrence N / T2 of the second user path are both not less than the co-occurrence threshold.

[0130] The device further includes: The parameter value adjustment unit is used to provide operation options for adjusting parameters for at least one of the following thresholds when it is necessary to improve the accuracy of peer relationship recognition results: similarity threshold, co-occurrence threshold, first time threshold, second time threshold, and the logical relationship between the start time difference and end time difference of two behavioral segments when performing spatiotemporal co-occurrence judgment; the adjusted parameter values ​​are obtained through the operation options, and peer relationship recognition is performed based on the adjusted parameter values.

[0131] Corresponding to the foregoing method embodiments, this application also provides a digital management device for physical stores. See also Figure 5 The device may include: The peer relationship identification result acquisition unit 501 is used to obtain the identification result of the peer relationship identification of the physical store during the process of digital analysis of the physical store. The identification result is obtained by grouping the user paths of consumer users in the store in pairs within a preset time period, and then performing spatiotemporal co-occurrence judgment based on multiple behavioral segments associated with the first user path and the second user path in each group. The behavioral segments are obtained by segmenting the movement trajectory of consumer users in the physical store. The management strategy optimization unit 502 is used to determine the customer structure information of the physical store based on the identification result, and optimize the management strategy of the physical store based on the customer structure information.

[0132] In addition, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in any of the foregoing method embodiments.

[0133] And an electronic device, comprising: One or more processors; and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any of the foregoing method embodiments.

[0134] A computer program product includes a computer program / computer executable instructions that, when executed by a processor in an electronic device, implement the steps of the method described in the foregoing method embodiments.

[0135] in, Figure 6 The architecture of an electronic device is illustrated by example. For instance, device 600 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, aircraft, etc.

[0136] Reference Figure 6 The device 600 may include one or more of the following components: a processing component 602, a memory 604, a power supply component 606, a multimedia component 608, an audio component 610, an input / output (I / O) interface 612, a sensor component 614, and a communication component 616.

[0137] Processing component 602 typically controls the overall operation of device 600, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 602 may include one or more processors 620 to execute instructions to perform all or part of the steps of the methods provided in this disclosure. Furthermore, processing component 602 may include one or more modules to facilitate interaction between processing component 602 and other components. For example, processing component 602 may include a multimedia module to facilitate interaction between multimedia component 608 and processing component 602.

[0138] Memory 604 is configured to store various types of data to support the operation of device 600. Examples of this data include instructions for any application or method operating on device 600, contact data, phonebook data, messages, pictures, videos, etc. Memory 604 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0139] Power supply component 606 provides power to various components of device 600. Power supply component 606 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 600.

[0140] Multimedia component 608 includes a screen that provides an output interface between device 600 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 608 includes a front-facing camera and / or a rear-facing camera. When device 600 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0141] Audio component 610 is configured to output and / or input audio signals. For example, audio component 610 includes a microphone (MIC) configured to receive external audio signals when device 600 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 604 or transmitted via communication component 616. In some embodiments, audio component 610 also includes a speaker for outputting audio signals.

[0142] I / O interface 612 provides an interface between processing component 602 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0143] Sensor assembly 614 includes one or more sensors for providing status assessments of various aspects of device 600. For example, sensor assembly 614 may detect the on / off state of device 600, the relative positioning of components such as the display and keypad of device 600, changes in the position of device 600 or a component of device 600, the presence or absence of user contact with device 600, the orientation or acceleration / deceleration of device 600, and temperature changes of device 600. Sensor assembly 614 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 614 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 614 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0144] Communication component 616 is configured to facilitate wired or wireless communication between device 600 and other devices. Device 600 can access wireless networks based on communication standards, such as WiFi, or mobile communication networks such as 2G, 6G, 4G / LTE, and 6G. In one exemplary embodiment, communication component 616 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 616 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0145] In an exemplary embodiment, device 600 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0146] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 604 including instructions, which can be executed by a processor 620 of device 600 to perform the method provided by the present disclosure. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0147] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0148] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0149] The solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for identifying user peer relationships, characterized in that, The method includes: Determine the set of user paths associated with physical stores, the set of user paths including multiple user paths generated for consumer users located in the store within a preset time period; The multiple user paths are grouped into pairs, and for the first user path and the second user path in each group, a set of associated behavior segments is obtained. The set of behavior segments includes the behavior segments associated with a single consumer user under different image acquisition devices and the start and end times of the behavior segments. The behavior segments are obtained by segmenting the consumer user's movement trajectory in the physical store. Based on the start and end times of the behavior segments, the spatiotemporal co-occurrence of the behavior segments associated with the first user path and the second user path under the same image acquisition device is determined, and a co-occurrence connection relationship is established between the behavior segments with spatiotemporal co-occurrence. Based on the number of behavioral fragments with co-occurrence connections, the peer relationship identification results between the first user path and the second user path in the group are obtained, so as to analyze the customer structure information of the physical store based on the peer relationship identification results, and optimize the store management strategy based on the customer structure information.

2. The method according to claim 1, characterized in that, The method further includes: If multiple groups with peer relationships are identified based on the peer relationship identification results, then the multiple groups with peer relationships are aggregated, and at least two groups that include the same user path are merged into a multi-user group. If at least one group without peer relationship is identified based on the peer relationship identification result, the user paths in the at least one group without peer relationship are divided into different single customer groups.

3. The method according to claim 1 or 2, characterized in that, The set of behavioral fragments associated with the user path can be obtained in the following way: Obtain cross-view behavior information of the consumer user in the physical store, the cross-view behavior information including multiple user trajectories and a user path generated by merging the multiple user trajectories; The user trajectory is used to represent the sequence of movement trajectories formed by the continuous tracking of the consumer user within the field of view of a single image acquisition device; After sorting the multiple user trajectories in chronological order, the behavioral segments associated with the consumer users under different image acquisition devices are segmented, and the start and end times of the behavioral segments are obtained. The behavioral segments are generated by merging at least one user trajectory formed within the field of view of a single image acquisition device. The start and end times of the behavioral segments are determined according to the acquisition time of the at least one user trajectory. The acquisition time of the user trajectory includes the acquisition timestamp of the first trajectory point and the acquisition timestamp of the last trajectory point in the movement trajectory sequence.

4. The method according to claim 3, characterized in that, The step of sorting the multiple user trajectories in chronological order includes: The multiple user trajectories are sorted chronologically based on the collection timestamp of the first trajectory point in the user trajectory.

5. The method according to claim 4, characterized in that, The method further includes: After obtaining the temporal sorting results of the multiple user trajectories, if there is a time reversal and / or spatial disorder between two adjacent user trajectories, the abnormal user trajectories are filtered out.

6. The method according to claim 3, characterized in that, The segmentation of the consumer user's behavior fragments associated with different image acquisition devices includes: Based on the detected device switching event of the image acquisition device, the user's at least one user trajectory under a single image acquisition device is segmented into a behavior segment.

7. The method according to claim 6, characterized in that, The method further includes: Obtain the time window threshold and the collection timestamp associated with each trajectory point in the user trajectory; If the time difference between the acquisition timestamps of two adjacent trajectory points exceeds the time window threshold, then the behavior segmentation continues between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

8. The method according to claim 6, characterized in that, The method further includes: Obtain the location jump threshold and the spatial location associated with each trajectory point in the user trajectory; If the spatial span between two adjacent trajectory points exceeds the position jump threshold, then the behavior segmentation continues between the two adjacent trajectory points to obtain at least two behavior segments of the consumer user under a single image acquisition device.

9. The method according to claim 3, characterized in that, The acquisition of cross-view behavior information of consumer users in physical stores includes: The video streams captured by different image acquisition devices in the physical store are obtained as input to the multi-target tracking (MOT) system, which then performs target detection and tracking processing with the consumer user as the target. Obtain cross-view behavior information of different consumer users in the physical store, output by the MOT system.

10. The method according to claim 1 or 2, characterized in that, The step of determining the spatiotemporal co-occurrence of behavioral segments associated with the first user path and the second user path under the same image acquisition device based on the start and end times of the behavioral segments includes: Determine the behavioral segments associated with the first user path and the second user path under the same image acquisition device. If the start and end times of the two behavioral segments overlap, and the start time difference between the two behavioral segments does not exceed a first time threshold and / or the end time difference between the two behavioral segments does not exceed a second time threshold, then determine that the two behavioral segments have spatiotemporal co-occurrence.

11. The method according to claim 1 or 2, characterized in that, The step of obtaining the peer relationship identification result between the first user path and the second user path based on the number of behavioral segments with established co-occurrence connections includes: The path similarity S between the first user path and the second user path is calculated based on the number N of behavior segments with co-occurrence connections, the number T1 of behavior segments associated with the first user path, and the number T2 of behavior segments associated with the second user path. If the path similarity is not lower than the similarity threshold, an identification result indicating that the first user path and the second user path have a peer relationship is obtained.

12. The method according to claim 11, characterized in that, The method further includes: Obtain the co-occurrence threshold; The path similarity S is calculated when both the co-occurrence N / T1 of the first user path and the co-occurrence N / T2 of the second user path are not less than the co-occurrence threshold.

13. The method according to claim 1 or 2, characterized in that, The method further includes: When it is necessary to improve the accuracy of peer relationship identification results, the following operation options are provided to adjust parameters for at least one of the following thresholds: similarity threshold, co-occurrence threshold, first time threshold, second time threshold, and the logical relationship between the start time difference and end time difference of two behavioral segments when performing spatiotemporal co-occurrence judgment; The adjusted parameter values ​​are obtained through the operation options, and the peer relationship is identified based on the adjusted parameter values.

14. A digital management method for physical stores, characterized in that, The method includes: In the process of digital analysis of physical stores, the identification results of user peer relationship identification for the physical stores are obtained. The identification results are obtained by grouping the user paths of consumer users in the store in pairs within a preset time period, and then performing spatiotemporal co-occurrence judgment based on multiple behavioral segments associated with the first user path and the second user path in each group. The behavioral segments are obtained by segmenting the movement trajectory of consumer users in the physical store. The customer structure information of the physical store is determined based on the identification results, and the management strategy of the physical store is optimized based on the customer structure information.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program performs the steps of the method described in any one of claims 1 to 14.

16. An electronic device, characterized in that, include: One or more processors; as well as A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1 to 14.

17. A computer program product comprising a computer program / computer executable instructions, characterized in that, When the computer program / computer executable instructions are executed by a processor in an electronic device, they implement the steps of the method according to any one of claims 1 to 14.

Images