A passenger flow data monitoring method and system

Abstract

The present application relates to the technical field of data processing, and particularly relates to a passenger flow data monitoring method and system, comprising: dividing users into standing users and non-standing users, wherein the standing users include standing-in-store users and non-standing-in-store users; in each preset period within each preset cycle, obtaining a standing passenger flow direct conversion rate according to the number of standing-in-store users and standing users; determining standing conversion passenger flow in each preset period in the next cycle in the future by combining the comparison result of the number of non-standing-in-store users and non-standing-in-store users in the same preset period within different preset cycles; and determining predicted in-store passenger flow in each preset period in the next cycle in the future. The present application can effectively improve the interpretability of predicted passenger flow and guarantee the accuracy of in-store passenger flow prediction.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for monitoring passenger flow data. Background Technology

[0002] Customer flow monitoring is a crucial step in supporting business operations, public safety, and urban management by statistically analyzing data such as pedestrian traffic, density, and movement patterns within a given area. For business operations, monitoring customer flow data can optimize operational strategies and resource allocation, improve marketing efficiency and customer experience, enhance security management and risk warning, and promote the development of smart shopping malls. In short, customer flow monitoring in shopping malls is both a tool for improving short-term operational efficiency and an important basis for long-term strategic optimization.

[0003] Existing problems: Current methods typically only consider the changing trends of in-store foot traffic to predict future foot traffic. This approach ignores the impact of foot traffic that can be converted into actual in-store traffic. For example, some historical foot traffic may enter the store in the future, thus affecting current foot traffic, similar to the impact of "repeat customers"—customers who linger and observe at the store entrance are potential customers. Traditional methods based solely on in-store foot traffic monitoring ignore the conversion impact of foot traffic, resulting in low accuracy in foot traffic monitoring and prediction, and poor interpretability of foot traffic composition. Summary of the Invention

[0004] This invention provides a method and system for monitoring passenger flow data to solve existing problems.

[0005] The present invention provides a method and system for monitoring passenger flow data, which adopts the following technical solution: One embodiment of the present invention provides a method for monitoring passenger flow data, the method comprising the following steps: Obtain the time sequence of walking trajectory points and walking speed sequence of each user in the monitoring video within each preset period of each preset cycle. The monitoring video is divided into the in-store area and the out-of-store area. Based on the time sequence of the walking trajectory points and the walking speed sequence, combined with the distribution of the walking trajectory points in the store area and the store area, users are divided into users who stop and users who do not stop and enter the store. Users who stop include users who stop and then enter the store and users who stop but do not enter the store. In each preset period within each preset cycle, the direct conversion rate of stopping traffic is obtained based on the difference between the number of users who stop and then enter the store and the number of users who stop. Based on the comparison of the number of users who did not stop and entered the store within the same preset time period in different preset cycles and the number of users who stopped but did not enter the store, the predicted historical conversion customer traffic in each preset time period in the next cycle is determined; the number of users who stopped within the preset time period is adjusted according to the direct conversion rate of the stopped customer traffic, and combined with the predicted historical conversion customer traffic, the stopped conversion customer traffic in each preset time period in the next cycle is determined. Based on the size of the customer traffic converted from stopping at the store, the predicted customer traffic for each preset time period in the next cycle is determined.

[0006] Furthermore, the process of dividing users into those who stop at the store and those who enter the store but do not stop at the store includes users who stop at the store and those who stop at the store but do not enter the store. The specific steps involved are as follows: The degree of each user's dwell time is determined based on the time sequence of each user's walking trajectory points and walking speed sequence; Users whose dwell time exceeds a preset threshold are recorded as dwelling users; Users whose dwell time is less than or equal to the preset judgment threshold are recorded as users who do not dwell. For users who stop, when the first walking trajectory point in the time sequence is in the area outside the store and the last walking trajectory point is in the area inside the store, they are recorded as users who stop and then enter the store. When both the first and last walking trajectory points in the time sequence are in the area outside the store, they are recorded as users who stop but do not enter the store. For users who do not stop, if the first walking trajectory point in the time sequence is located outside the store and the last walking trajectory point is located inside the store, they are recorded as users who do not stop and enter the store.

[0007] Furthermore, the specific steps for determining the degree of pausing for each user based on the time sequence of each user's walking trajectory points and walking speed sequence are as follows: For each user, the time interval between the first and last walking trajectory points in the time sequence is obtained, and the normalized value of the difference between the maximum walking speed and the minimum walking speed in the walking speed sequence is obtained and recorded as the degree of discontinuity. The normalized value of the product of the degree of discontinuity and the time interval is recorded as the degree of lingering for each user.

[0008] Furthermore, the specific steps for obtaining the direct conversion rate of stopped-at-home customers are as follows: Within each preset period, the ratio of the number of customers who stop and then enter the store to the total number of customers who stop is recorded as the direct conversion rate of customer traffic within each preset period.

[0009] Furthermore, the specific steps for determining the predicted historical conversion passenger flow for each preset time period in the next future cycle are as follows: Preset quantity threshold The current preset period is compared with the nearest previous preset period. Each preset period is denoted as the target period; According to the target period The number of customers who did not stop to enter the store within a preset time period is used to obtain the conversion customer traffic sequence; Starting from the current preset period, proceed in reverse chronological order, using the preset step size sequence... Iterate through the previous steps, and then iterate through the previous steps. The preset period is denoted as the nth preset step size sequence. The reference period corresponding to each step size; Count the numbers in the preset step size sequence in chronological order. Within all reference periods corresponding to the step size, the first... The number of users who stopped but did not enter the store within a preset time period constitutes a sequence of the number of users who stopped but did not enter the store; Based on the converted customer traffic sequence and the sequence of the number of users who stopped but did not enter the store, determine the first step in the preset step sequence. The indirect conversion rate of customers who stop but do not enter the store to customers who enter the store within a certain step length; In the preset step size sequence, the first... At each step length, the product of the mean of the sequence of users who stopped but did not enter the store and the indirect conversion rate of stopped-but-not-entered customers into in-store customers is denoted as the second product. The sum of the second products at all step lengths in the preset step length sequence is denoted as the number of users in the next future period. Predicted historical conversion rate of customers within a preset time period.

[0010] Furthermore, the statement based on the first [period] within the target period The specific steps involved in determining the number of customers who did not stop to enter the store within a preset time period and obtaining the conversion customer traffic sequence are as follows: Statistics on the first period of all target periods The minimum number of customers who did not stop to enter the store within a preset time period is taken as the floor value of the product of the minimum value and the preset proportion threshold. Within each target period, the first The difference between the number of users who do not stop to enter the store within a preset time period and the regular, stable number of customers entering the store is recorded as the converted customer traffic. Statistically analyze the first [number]th ... The converted customer traffic corresponding to each preset time period constitutes a converted customer traffic sequence.

[0011] Furthermore, the step of determining the first step in the preset step sequence based on the converted customer flow sequence and the sequence of the number of users who stopped but did not enter the store The indirect conversion rate of customers who stop but do not enter the store to customers who do enter the store at a certain step length includes the following specific steps: Obtain the conversion customer traffic sequence and the preset step size sequence. The inversely proportional normalized value of the DTW distance of the sequence of the number of users who stopped but did not enter the store corresponding to the nth step length is denoted as the nth step length sequence. The trends of customer traffic that stops but does not enter the store and customer traffic that converts are consistent across different steps; In the preset step size sequence, the first... At each step length, the product of the mean of the sequence of users who stopped but did not enter the store and the consistency of the changing trends of the stopped-but-not-entered and converted customer flows is recorded as the first product. The sum of the first products at all step lengths in the preset step length sequence is recorded as the first sum. The ratio of the mean of the converted customer flow sequence to the first sum is calculated and recorded as the first ratio. The first ratio is then compared with the first step length sequence at each step length. The normalized value of the product of the consistent trends of the remaining but not-entering customer flow and the converted customer flow under the preset step length is denoted as the i-th step length sequence. The indirect conversion rate of customers who stop but do not enter the store into the store within a certain step length.

[0012] Furthermore, the specific steps for determining the number of customers who stop and convert within each preset time period in the next cycle are as follows: Based on the first of all target periods The number of users staying within a preset time period is used to obtain the number of users in the next future period using an LSTM network. The number of visitors staying within a preset time period; Get the first [number] within all target periods The average direct conversion rate of foot traffic within a preset time period is used to compare the average direct conversion rate of foot traffic with the average direct conversion rate of the foot traffic within the next period. The product of the number of visitors staying within a preset time period is recorded as the product of the number of visitors staying within the next period. The number of people stopping within a preset time period directly converts into customer traffic; The next cycle The sum of the predicted historical conversion rate and the direct conversion rate within a preset time period, rounded up, is recorded as the integer value of the sum of the predicted historical conversion rate and the direct conversion rate within a given time period for the next period. The number of customers who stop and convert within a preset time period.

[0013] Furthermore, the specific steps for determining the predicted customer traffic in each preset time period within the next cycle are as follows: Compare the regular and stable customer traffic with the next cycle. The sum of the number of customers who stop and convert within a preset time period is recorded as the value of the number of customers who stop and convert within the next period. Predicted customer traffic within a preset time period.

[0014] The present invention also proposes a passenger flow data monitoring system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned passenger flow data monitoring method.

[0015] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, users are divided into users who stop and users who enter the store but do not stop. Users who stop include users who enter the store after stopping and users who do not enter the store after stopping. By classifying users, the impact of useless user data is reduced, and the accuracy of subsequent customer traffic prediction is ensured. Within each preset period, the direct conversion rate of foot traffic is obtained based on the difference between the number of users who stopped and entered the store and the number of users who stopped but did not enter the store. By comparing the number of users who did not stop and entered the store within the same preset period in different preset periods with the number of users who stopped but did not enter the store, the predicted historical conversion foot traffic for each preset period in the next period is determined. Then, the number of users who stopped within the preset period is adjusted based on the direct conversion rate of foot traffic to obtain the direct conversion foot traffic for each preset period in the next period. This determines the conversion foot traffic for each preset period in the next period. Based on the predicted historical conversion foot traffic of users who stopped but did not enter the store and subsequently converted to customers, and the direct conversion foot traffic of users who stopped and directly entered the store, the total conversion foot traffic is obtained. This analysis of the conversion of foot traffic to customers further ensures the accuracy of subsequent foot traffic prediction. The predicted foot traffic for each preset period in the next period is obtained. Thus, this invention effectively improves the interpretability of predicted foot traffic and ensures the accuracy of foot traffic prediction. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the steps of a passenger flow data monitoring method according to the present invention. Figure 2 A diagram illustrating user segmentation. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a passenger flow data monitoring method and system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following description, in conjunction with the accompanying drawings, details the specific scheme of a passenger flow data monitoring method and system provided by the present invention.

[0021] Please see Figure 1 The diagram illustrates a flowchart of a passenger flow data monitoring method according to an embodiment of the present invention, which includes the following steps: Step S001: Obtain the time sequence of walking trajectory points and walking speed sequence of each user in the monitoring video of each preset period within each preset cycle. The monitoring video is divided into the in-store area and the out-of-store area.

[0022] Cameras are installed at fixed locations outside any store to collect real-time video footage of user movement. The capture frequency is 24 frames per second, with each frame corresponding to a timestamp. Since the shooting position and angle are fixed, the monitoring range is also fixed. This allows for pre-marking and delineation of the store's interior and exterior areas within the video feed. The interior area is limited to the store entrance; once a user enters the store, they are outside the monitoring range. A video recognition model is then built to identify human figures within the video. This model uses a pre-trained YOLO network, which is then trained on manually labeled human recognition video footage. The result is a bounding box representing the human figure in each video frame. The center coordinates of the bounding box represent the user's coordinates within the monitored video frame. Matching is performed on the same user across consecutive video frames using human features within the bounding box, employing the SIFT (Scale Invariant Feature Transform) algorithm. This yields a sequence of the center coordinates of the bounding boxes for the same user across consecutive video frames, representing the user's walking trajectory point time sequence. A planar coordinate system is constructed with the bottom left corner of each video frame as the origin, the horizontal axis pointing to the right as the positive x-axis, and the vertical axis pointing upwards as the positive y-axis. The center coordinates of the bounding boxes within the video frames are then obtained within this planar coordinate system. The Euclidean distance between two adjacent walking trajectory points in the time sequence is used as the walking speed corresponding to the preceding walking trajectory point. The walking speed corresponding to the last walking trajectory point is then set to the walking speed corresponding to the second-to-last walking trajectory point, thus obtaining the user's walking speed sequence.

[0023] The YOLO network model and SIFT (Scale Invariant Feature Transform) are both well-known techniques, and their specific methods will not be described here.

[0024] In this embodiment, the preset cycle is one day and the preset time period is 3 hours, which will be used as an example for description.

[0025] This allows us to obtain the time sequence of each user's walking trajectory points and walking speed sequence in the monitoring video within each preset period and preset time period of each preset cycle.

[0026] It should be noted that if the time sequence of a user's walking trajectory points and walking speed sequence in the surveillance video within a certain preset time period is incomplete, meaning that some of the user's walking trajectory points are within the time period before or after the preset time period, then the complete time sequence of the user's walking trajectory points and walking speed sequence will be assigned to that preset time period. For the same user, the entire process from entering to leaving the monitoring range constitutes one complete monitoring session. When the same user re-enters the monitoring range, it is treated as a new monitoring session, i.e., analyzed as a new user.

[0027] Step S002: Based on the time sequence of the walking trajectory points and the walking speed sequence, and combined with the distribution of the walking trajectory points in the store area and the store area, users are divided into users who stop and users who do not stop and enter the store. Users who stop include users who stop and then enter the store and users who stop but do not enter the store. In each preset period within each preset cycle, the direct conversion rate of stopping traffic is obtained based on the difference between the number of users who stop and then enter the store and the number of users who stop.

[0028] It's important to note that since the basic unit of customer traffic is the user, and user behavior reflects their potential to enter the store, generally, the longer a user lingers outside the store, the higher their potential interest, and the more likely they are to become a customer upon passing by again. Therefore, from an overall customer flow perspective, lingering customers at a specific time can be divided into those who linger and then enter the store, and those who linger but do not enter. Those who linger but do not enter are more likely to become customers in the future; that is, the more lingering but non-entering customers there are, the more likely they are to become customers in the future. Thus, customer traffic at a given time includes not only long-term, stable customer traffic, but also customers who linger and then enter, and those who lingered but did not enter in the past and subsequently became customers. Long-term, stable customer traffic is generally considered to be similar across historical time periods; therefore, differences in customer traffic at different times are due to historical lingering but non-entering customer traffic. Therefore, by obtaining the consistency between the historical trends of foot traffic (those who stopped but did not enter the store) and the long-term stable trends of customer traffic over different periods, the higher the consistency of these trends over a certain period, the higher the conversion rate of foot traffic (those who stopped but did not enter the store) to in-store traffic. Furthermore, in-store traffic for a given period can be divided into regular traffic and foot traffic converted to in-store traffic. Foot traffic converted to in-store traffic includes in-store traffic that stopped but did not enter during the current period, and in-store traffic that stopped but did not enter in the past but later converted to in-store traffic. Corresponding conversion rate weights are calculated for foot traffic that stopped but did not enter the store over multiple periods. Therefore, in this embodiment, the time period is first divided, and historical in-store traffic and foot traffic are componentized to obtain the direct customer traffic conversion rate. Secondly, the corresponding indirect customer traffic conversion rate is calculated based on the historical componentized customer traffic, ultimately obtaining the predicted conversion customer traffic for future periods.

[0029] It's important to further clarify that store traffic is typically counted based on the number of people entering the store within a given period. Current store traffic can be broadly categorized into two parts based on whether the customer lingers outside the store: regular inbound traffic and lingering-and-converting traffic. Regular inbound traffic indicates customers with a clear goal who enter the store without lingering outside. Lingering-and-converting traffic indicates customers who, during the current period, exhibit certain observational behaviors outside the store before entering. Common lingering-and-converting characteristics include the duration of time a user spends in front of the store and changes in their speed. For example, a longer duration and intermittent pauses in speed indicate a higher potential interest in the store, suggesting the user is likely lingering. When a user's first walking point is outside the store and their last walking point is inside the store, they are considered to have entered. When their first walking point is inside the store and their last walking point is outside, they are considered to have left. When their first walking point is outside the store and their last walking point is outside, they are considered to have not entered the store. Since the store's interior area only includes the area near the entrance, once a user enters, they are outside the monitoring range. This means that a single walking trajectory sequence does not include the process of a user entering and leaving the store; it only includes the entry or exit process. Therefore, by analyzing the pedestrian flow information in the surveillance video outside the store, corresponding customer flow data for different time periods can be segmented. In other words, users in the surveillance video are categorized based on characteristics, primarily dividing the real-time monitored customer flow data into those who are stopping and those who are entering the store.

[0030] Preferably, in one embodiment of the present invention, the method for obtaining the direct conversion rate of foot traffic includes: For the For each user, the time interval between the first and last walking trajectory points in the time sequence is obtained. Then, the normalized value of the difference between the maximum and minimum walking speeds in the walking speed sequence is obtained and denoted as the discontinuity level. The normalized value of the product of the discontinuity level and this time interval is denoted as the [number of]th [users]. The level of user engagement.

[0031] It should be noted that when a user shows interest in a store, they will stop and observe. When this happens, the user typically appears in the frame for a relatively long time, and their walking speed is usually intermittent. Therefore, the degree to which a user stops, as detected by the camera outside the store, is directly proportional to the length of time they appear in the frame and the degree of discontinuity in their walking speed sequence. This embodiment uses... The linear normalization function normalizes the difference between the maximum walking speed and the minimum walking speed, as well as the product of the degree of discontinuity and the time interval, to between 0 and 1.

[0032] The preset judgment threshold is 0.4, and this will be used as an example for explanation.

[0033] Users whose dwell time exceeds a preset threshold are recorded as dwelling users. Users whose dwell time is less than or equal to the preset threshold are recorded as non-dwelling users.

[0034] For users who stop, if the first walking trajectory point in the time sequence is located outside the store and the last walking trajectory point is located inside the store, they are recorded as users who stop and then enter the store. If both the first and last walking trajectory points in the time sequence are located outside the store, they are recorded as users who stop but do not enter the store.

[0035] For users who do not stop, when the first walking trajectory point in the time sequence is in the area outside the store and the last walking trajectory point is in the area inside the store, they are recorded as users who do not stop and enter the store (regular in-store traffic). When both the first and last walking trajectory points in the time sequence are in the area outside the store, they are recorded as users who do not stop and do not enter the store.

[0036] It should be noted that the user segmentation diagram is as follows: Figure 2 As shown, other user scenarios are not analyzed.

[0037] Within each preset period, the ratio of the number of customers who stop and then enter the store to the total number of customers who stop is recorded as the direct conversion rate of customer traffic within each preset period.

[0038] Step S003: Based on the comparison results of the number of users who did not stop and entered the store and the number of users who stopped but did not enter the store within the same preset time period in different preset cycles, determine the predicted historical conversion customer traffic in each preset time period in the next cycle; adjust the number of users who stopped within the preset time period according to the direct conversion rate of the stopped customer traffic, and combine it with the predicted historical conversion customer traffic to determine the stopped conversion customer traffic in each preset time period in the next cycle.

[0039] It's important to note that the customer traffic that doesn't stop to enter the store includes not only the regular, stable customer flow but also indirect conversions from historical customers who stopped but didn't enter. Therefore, it's necessary to analyze the impact of historical customer traffic from different time periods on current customer traffic. For an individual user, passing by a store typically exhibits a certain cyclical pattern, such as daily or weekly cycles. Thus, a potential user (stopping outside the store but not entering) will only enter the store within a specific cycle. Furthermore, the increase in customer traffic within a given time period compared to the previous period is primarily due to changes in customer traffic from those who stopped. Therefore, if we exclude current-time customer traffic from those who stopped but didn't enter, the remaining customer traffic represents historical customer traffic from those who stopped but didn't. Since historical conversions exist at different cyclical levels, the closer the trend of customer traffic from those who stopped but didn't enter during a given period is to the trend of historical customer traffic from those who entered (excluding current-time conversions), the higher the indirect conversion rate of those who stopped during that period. Therefore, in this embodiment, different period intervals are set, and the consistency between the number of customers who stop but do not enter the store and the historical customer flow that enters the store is calculated under different period intervals, so as to obtain the conversion rate of the number of customers who stop but do not enter the store under different period intervals.

[0040] Preferably, in one embodiment of the present invention, the method for obtaining the number of customers who stop and convert within each preset time period in the next cycle includes: Preset quantity threshold The value is 10, and the preset ratio threshold is 90%. This will be used as an example for explanation.

[0041] Compare the current preset period with the nearest previous preset period. Each preset period is denoted as the target period.

[0042] Specifically: The current preset period is denoted as the [number]. The first preset cycle will be the first The preset cycle to the first All preset periods between the first preset periods are denoted as the target period (including the first preset period). The preset cycle and the first (One preset period). The target number of periods is... .

[0043] Statistics on the first [number]th [period] within each target period The number of users who did not stop to enter the store within a preset time period, and then the number of users within all target periods. The minimum number of customers who did not stop to enter the store within a preset time period is taken as the floor value of the product of the minimum number and the preset proportion threshold.

[0044] Within each target period, the first The difference between the number of users who do not stop to enter the store within a preset time period and the regular, stable number of customers entering the store is recorded as the converted customer traffic.

[0045] Statistically analyze the first [number]th ... The converted customer traffic corresponding to each preset time period constitutes a converted customer traffic sequence.

[0046] The preset step size sequence is {1, 2, 3, 4, 5}, and this will be used as an example for explanation.

[0047] Starting from the current preset period, proceed in reverse chronological order, using the preset step size sequence... Iterate through the previous steps, and then iterate through the previous steps. The preset period is denoted as the nth preset step size sequence. The reference period corresponding to each step size.

[0048] For example: the When the step size is 1, the reference period is the th One to the first The preset period between preset periods. When the step size is 2, the reference period is the th The, the The, the The, ..., the first A preset cycle. When the step size is 3, the reference period is the th The, the The, the The, ..., the first A preset cycle.

[0049] For the first step in the preset step size sequence Each step size is used to count the first step within each reference period. The number of users who stopped at the store but did not enter within a preset time period was counted sequentially for each reference period. The number of users who stopped at the store but did not enter the store within a preset time period constitutes a sequence of the number of users who stopped at the store but did not enter the store.

[0050] Using the DTW algorithm, obtain the first step of the conversion customer flow sequence and the second step of the preset step size sequence. DTW distance of the sequence of the number of users who stopped but did not enter the store corresponding to each step length. DTW distance The inverse proportional normalized value is denoted as the i-th value in the preset step size sequence. The trends of customer traffic that stops but does not enter the store and customer traffic that converts are consistent across different steps.

[0051] It should be noted that the DTW (Dynamic Time Warping) algorithm is a well-known technique, and its specific method will not be described here. The smaller the DTW distance, the more similar the two sequences are. As DTW distance The inverse proportional normalized value, This is a normalization function used to normalize data values ​​to a range of 0 to 1, with the sum of all normalized values ​​being 1. Specifically, the sum of the consistent trends between the number of customers who stopped but did not enter the store and the converted customer flow across all step lengths in the preset step length sequence is 1. In other words, the more consistent the trends between the sequence of users who stopped but did not enter the store and the sequence of converted customer flow under that step length, the higher the indirect conversion rate of historical customers who stopped but did not enter the store under that step length.

[0052] In the preset step size sequence, the first... At each step length, the product of the mean of the sequence of users who stopped but did not enter the store and the consistency of the changing trends of the stopped-but-not-entered and converted customer flows is recorded as the first product. The sum of the first products at all step lengths in the preset step length sequence is recorded as the first sum. The ratio of the mean of the converted customer flow sequence to the first sum is calculated and recorded as the first ratio. The first ratio is then compared with the first step length sequence at each step length. The product of the consistent trends between the number of customers who linger but do not enter the store and the number of customers who convert at each step length. The normalized value is denoted as the th value in the preset step size sequence. The indirect conversion rate of customers who stop but do not enter the store into the store within a certain step length.

[0053] It should be noted that: with As product The normalized value, This is a normalization function used to normalize data values ​​to a range of 0 to 1, with the sum of all normalized values ​​being 1. The consistency across all step sizes represents the proportion of customers who remain in the store but do not enter, which is then converted into customers entering the store. This consistency is used to determine the indirect conversion rate of customers who remain in the store but do not enter the store at each step size.

[0054] In the preset step size sequence, the first... At each step length, the product of the mean of the sequence of users who stopped but did not enter the store and the indirect conversion rate of stopped-but-not-entered customers into in-store customers is denoted as the second product. The sum of the second products at all step lengths in the preset step length sequence is denoted as the number of users in the next future period. Predicted historical conversion rate of customers within a preset time period.

[0055] It should be noted that: since the indirect conversion rate and direct conversion rate of historical foot traffic remain consistent in the short term, the indirect conversion rate under different time periods is used to calculate the weighted average of the series of number of users who stopped but did not enter the store under different time periods to obtain the predicted historical conversion foot traffic.

[0056] Statistics on the first [number]th ... The number of users who remain within a preset time period, based on the number of users in all target periods. The number of users staying within a preset time period is used to obtain the number of users in the next future period using an LSTM network. The number of visitors stopping within a preset time period.

[0057] Among them, LSTM (Long Short-Term Memory) network is a well-known technology, and the specific method will not be introduced here.

[0058] Get the first [number] within all target periods The average direct conversion rate of visitor traffic within a preset time period is used to compare this average with the average conversion rate of the next period. The product of the number of visitors staying within a preset time period is recorded as the product of the number of visitors staying within the next period. The number of people stopping within a preset time period directly translates into customer traffic.

[0059] The next cycle The sum of the predicted historical conversion rate and the direct conversion rate within a preset time period, rounded up, is recorded as the integer value of the sum of the predicted historical conversion rate and the direct conversion rate within a given time period for the next period. The number of customers who stop and convert within a preset time period.

[0060] It should be noted that the predicted historical conversion traffic is obtained from the indirect conversion rate of customers who stop but do not enter the store to customers who enter the store, and the direct conversion traffic is obtained from the direct conversion rate of customers who stop to enter the store. This determines the conversion traffic within the same time period in the future cycle.

[0061] Step S004: Based on the size of the customer traffic converted from stopping to entering the store, determine the predicted customer traffic for each preset time period in the next cycle.

[0062] It should be noted that the proportion of customers who stop and then enter the store during the current period is usually related to the store environment, special store events, special services, etc. These factors are usually stable in the short term. The average local historical conversion rate of the nearest neighbor can be used as the predicted conversion rate value. Then, based on the shopping mall's stopping traffic, the predicted stopping traffic and then entering traffic can be obtained.

[0063] Preferably, in one embodiment of the present invention, the method for obtaining the predicted customer traffic in each preset time period within the next cycle includes: Compare the regular and stable customer traffic with the next cycle. The sum of the number of customers who stop and convert within a preset time period is recorded as the value of the number of customers who stop and convert within the next period. Predicted customer traffic within a preset time period.

[0064] It should be noted that: based on the next cycle, the... The predicted in-store customer traffic is obtained by combining the customer traffic conversion rate within a preset time period with the actual, stable in-store customer traffic. In this embodiment, the predicted customer traffic is predicted separately from multiple components, which improves the accuracy of traditional predictions that only predict in-store customer traffic.

[0065] The present invention also provides a passenger flow data monitoring system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned passenger flow data monitoring method.

[0066] This invention is now complete.

[0067] In summary, in this embodiment of the invention, users are divided into those who stop and those who enter the store but do not. Those who stop include both those who stop and then enter the store and those who stop but do not enter. Within each preset period, the direct conversion rate of stopped customer traffic is obtained based on the difference between the number of users who enter after stopping and those who stop. Based on a comparison of the number of users who enter without stopping and those who stop but do not enter within the same preset period in different preset periods, the predicted historical conversion customer traffic for each preset period in the next period is determined. The number of stopped users within the preset period is adjusted based on the direct conversion rate of stopped customer traffic to determine the total converted customer traffic for each preset period in the next period, thereby determining the predicted in-store customer traffic for each preset period in the next period. This invention effectively improves the interpretability of predicted customer traffic and ensures the accuracy of customer traffic prediction.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring passenger flow data, characterized in that, The method includes the following steps: Obtain the time sequence of walking trajectory points and walking speed sequence of each user in the monitoring video within each preset period of each preset cycle. The monitoring video is divided into the in-store area and the out-of-store area. Based on the time sequence of the walking trajectory points and the walking speed sequence, combined with the distribution of the walking trajectory points in the store area and the store area, users are divided into users who stop and users who do not stop and enter the store. Users who stop include users who stop and then enter the store and users who stop but do not enter the store. In each preset period within each preset cycle, the direct conversion rate of stopping traffic is obtained based on the difference between the number of users who stop and then enter the store and the number of users who stop. Based on the comparison of the number of users who did not stop and entered the store within the same preset time period in different preset cycles and the number of users who stopped but did not enter the store, the predicted historical conversion customer traffic in each preset time period in the next cycle is determined; the number of users who stopped within the preset time period is adjusted according to the direct conversion rate of the stopped customer traffic, and combined with the predicted historical conversion customer traffic, the stopped conversion customer traffic in each preset time period in the next cycle is determined. Based on the size of the customer traffic converted from stopping at the store, the predicted customer traffic for each preset time period in the next cycle is determined.

2. The method for monitoring passenger flow data according to claim 1, characterized in that, The process of categorizing users into those who stop at the store and those who enter the store but do not stop at the store includes users who stop at the store and those who stop at the store but do not enter the store. The specific steps involved are as follows: The degree of each user's dwell time is determined based on the time sequence of each user's walking trajectory points and walking speed sequence; Users whose dwell time exceeds a preset threshold are recorded as dwelling users; Users whose dwell time is less than or equal to the preset judgment threshold are recorded as users who do not dwell. For users who stop, when the first walking trajectory point in the time sequence is in the area outside the store and the last walking trajectory point is in the area inside the store, they are recorded as users who stop and then enter the store. When both the first and last walking trajectory points in the time sequence are in the area outside the store, they are recorded as users who stop but do not enter the store. For users who do not stop, if the first walking trajectory point in the time sequence is located outside the store and the last walking trajectory point is located inside the store, they are recorded as users who do not stop and enter the store.

3. The method for monitoring passenger flow data according to claim 2, characterized in that, The specific steps for determining the degree of pausing for each user based on the time sequence of each user's walking trajectory points and walking speed sequence are as follows: For each user, the time interval between the first and last walking trajectory points in the time sequence is obtained, and the normalized value of the difference between the maximum walking speed and the minimum walking speed in the walking speed sequence is obtained and recorded as the degree of discontinuity. The normalized value of the product of the degree of discontinuity and the time interval is recorded as the degree of lingering for each user.

4. The method for monitoring passenger flow data according to claim 1, characterized in that, The specific steps involved in obtaining the direct conversion rate of foot traffic are as follows: Within each preset period, the ratio of the number of customers who stop and then enter the store to the total number of customers who stop is recorded as the direct conversion rate of customer traffic within each preset period.

5. The method for monitoring passenger flow data according to claim 1, characterized in that, The specific steps involved in determining the predicted historical conversion passenger flow for each preset time period in the next cycle are as follows: Preset quantity threshold The current preset period is compared with the nearest previous preset period. Each preset period is denoted as the target period; According to the target period The number of customers who did not stop to enter the store within a preset time period is used to obtain the conversion customer traffic sequence; Starting from the current preset period, proceed in reverse chronological order, using the preset step size sequence... Iterate through the previous steps, and then iterate through the previous steps. The preset period is denoted as the nth preset step size sequence. The reference period corresponding to each step size; Count the numbers in the preset step size sequence in chronological order. Within all reference periods corresponding to the step size, the first... The number of users who stopped but did not enter the store within a preset time period constitutes a sequence of the number of users who stopped but did not enter the store; Based on the converted customer traffic sequence and the sequence of the number of users who stopped but did not enter the store, determine the first step in the preset step sequence. The indirect conversion rate of customers who stop but do not enter the store to customers who enter the store within a certain step length; In the preset step size sequence, the first... At each step length, the product of the mean of the sequence of users who stopped but did not enter the store and the indirect conversion rate of stopped-but-not-entered customers into in-store customers is denoted as the second product. The sum of the second products at all step lengths in the preset step length sequence is denoted as the number of users in the next future period. Predicted historical conversion rate of customers within a preset time period.

6. The passenger flow data monitoring method according to claim 5, characterized in that, According to the target period, the first The specific steps involved in determining the number of customers who did not stop to enter the store within a preset time period and obtaining the conversion customer traffic sequence are as follows: Statistics on the first period of all target periods The minimum number of customers who did not stop to enter the store within a preset time period is taken as the floor value of the product of the minimum value and the preset proportion threshold. Within each target period, the first The difference between the number of users who do not stop to enter the store within a preset time period and the regular, stable number of customers entering the store is recorded as the converted customer traffic. Statistically analyze the first [number]th ... The converted customer traffic corresponding to each preset time period constitutes a converted customer traffic sequence.

7. The method for monitoring passenger flow data according to claim 5, characterized in that, The step is determined based on the converted customer traffic sequence and the sequence of the number of users who stopped but did not enter the store, in the preset step length sequence. The indirect conversion rate of customers who stop but do not enter the store to customers who do enter the store at a certain step length includes the following specific steps: Obtain the conversion customer traffic sequence and the preset step size sequence. The inversely proportional normalized value of the DTW distance of the sequence of the number of users who stopped but did not enter the store corresponding to the nth step length is denoted as the nth step length sequence. The trends of customer traffic that stops but does not enter the store and customer traffic that converts are consistent across different steps; In the preset step size sequence, the first... At each step length, the product of the mean of the sequence of users who stopped but did not enter the store and the consistency of the changing trends of the stopped-but-not-entered and converted customer flows is recorded as the first product. The sum of the first products at all step lengths in the preset step length sequence is recorded as the first sum. The ratio of the mean of the converted customer flow sequence to the first sum is calculated and recorded as the first ratio. The first ratio is then compared with the first step length sequence at each step length. The normalized value of the product of the consistent trends of the remaining but not-entering customer flow and the converted customer flow under the preset step length is denoted as the i-th step length sequence. The indirect conversion rate of customers who stop but do not enter the store into the store within a certain step length.

8. The method for monitoring passenger flow data according to claim 5, characterized in that, The specific steps involved in determining the conversion rate of visitors within each preset time period in the next cycle are as follows: Based on the first of all target periods The number of users staying within a preset time period is used to obtain the number of users in the next future period using an LSTM network. The number of visitors staying within a preset time period; Get the first [number] within all target periods The average direct conversion rate of foot traffic within a preset time period is used to compare the average direct conversion rate of foot traffic with the average direct conversion rate of the foot traffic within the next period. The product of the number of visitors staying within a preset time period is recorded as the product of the number of visitors staying within the next period. The number of people stopping within a preset time period directly converts into customer traffic; The next cycle The sum of the predicted historical conversion rate and the direct conversion rate within a preset time period, rounded up, is recorded as the integer value of the sum of the predicted historical conversion rate and the direct conversion rate within a given time period for the next period. The number of customers who stop and convert within a preset time period.

9. The passenger flow data monitoring method according to claim 6, characterized in that, The specific steps involved in determining the predicted customer traffic for each preset time period in the next cycle are as follows: Compare the regular and stable customer traffic with the next cycle. The sum of the number of customers who stop and convert within a preset time period is recorded as the value of the number of customers who stop and convert within the next period. Predicted customer traffic within a preset time period.

10. A passenger flow data monitoring system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of a passenger flow data monitoring method as described in any one of claims 1-9.

Images