A method and system for predicting store customer flow

Abstract

The embodiment of the present application discloses a method for predicting store customer flow. The method for predicting store customer flow includes: obtaining relevant information of multiple stores within a preset range; processing relevant information and associations of multiple stores based on a customer flow prediction model to determine the customer flow of a target store in a target time period; the customer flow prediction model includes a customer flow embedding layer and a prediction layer; wherein the customer flow embedding layer determines the customer flow characteristics of multiple stores; and the prediction layer determines the customer flow of a target store in a target time period based on the flow characteristics of the multiple stores.

Description

Technical Field

[0001] The present invention relates to the field of computer information processing, and in particular to a method and system for predicting store customer flow. Background Art

[0002] Accurately predicting the store's customer flow in a certain period of time in the future is of great significance to store operations. For example, store managers can more reasonably arrange employee shifts and business hours, prepare the number of goods for sale, and guide retailers' purchase and inventory plans based on the predicted customer flow. However, the store's customer flow is affected by many factors, and it is difficult to accurately predict the customer flow in a certain period of time in the future.

[0003] Therefore, it is necessary to provide a method and system for predicting store customer flow, so as to obtain a more accurate store customer flow prediction value with higher efficiency. Summary of the invention

[0004] One of the embodiments of the present specification provides a method for predicting store customer flow, the method comprising: obtaining relevant information of multiple stores within a preset range, the multiple stores comprising a target store and multiple associated stores having an associated relationship with the target store, the relevant information comprising information related to customer flow of the multiple stores in a historical target time period before a target time point; processing the relevant information of the multiple stores and the associated relationship based on a customer flow prediction model to determine the customer flow of the target store in a target time period after the target time point; the customer flow prediction model is a machine learning model, the customer flow prediction model comprises a customer flow embedding layer and a prediction layer; wherein the customer flow embedding layer determines the customer flow characteristics of the multiple stores; the customer flow embedding layer is a GNN model, with the multiple stores as nodes of a graph and the associated relationship as edges of the graph; the prediction layer determines the customer flow of the target store in the target time period based on the customer flow characteristics of the multiple stores, and the prediction layer is a fully connected layer; the customer flow prediction model is trained by an end-to-end learning method.

[0005] One of the embodiments of the present specification provides a store customer flow prediction system, the system comprising an acquisition module and a determination module; the acquisition module is used to acquire relevant information of multiple stores within a preset range, the multiple stores comprising a target store and multiple associated stores having an associated relationship with the target store, the relevant information comprising information related to the customer flow of the multiple stores in a historical target time period before a target time point; the determination module is used to process the relevant information of the multiple stores and the associated relationship based on a customer flow prediction model, and determine the customer flow of the target store in a target time period after the target time point; the customer flow prediction model is a machine learning model, and the customer flow prediction model comprises a customer flow embedding layer and a prediction layer; wherein the customer flow embedding layer determines the customer flow characteristics of the multiple stores; the customer flow embedding layer is a GNN model, with the multiple stores as nodes of a graph and the associated relationship as edges of the graph; the prediction layer determines the customer flow of the target store in the target time period based on the flow characteristics of the multiple stores, and the prediction layer is a fully connected layer; the customer flow prediction model is trained by an end-to-end learning method.

[0006] One of the embodiments of the present specification provides a store customer flow prediction device, including a processor, wherein the processor is used to execute the store customer flow prediction method.

[0007] One of the embodiments of the present specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the method for predicting store customer flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0009] Figure 1 is a schematic diagram of an application scenario of a store customer flow prediction system 100 according to some embodiments of this specification;

[0010] Figure 2 is an exemplary structural flow chart of using a customer flow prediction model to predict the customer flow of a target store in a target time period according to some embodiments of this specification;

[0011] Figure 3 is an exemplary schematic diagram of a store flow diagram according to some embodiments of this specification;

[0012] Figure 4 is an exemplary flow chart of a method for training a passenger flow prediction model according to some embodiments of this specification;

[0013] Figure 5 It is an exemplary flow chart of a method for obtaining current store customer flow / historical target time period customer flow as shown in some embodiments of this specification. DETAILED DESCRIPTION

[0014] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0015] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0016] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0017] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0018] Figure 1 It is a schematic diagram of an application scenario of a store customer flow prediction system 100 according to some embodiments of the present specification.

[0019] The application scenario may include a server 110 , a network 120 , at least two store terminals 130 , a database 140 and other data sources 150 . The server 110 may include a processing device 112 .

[0020] In some embodiments, the store customer flow prediction system 100 can obtain a customer flow prediction value for a target store in a target time period after a target time point by implementing the methods and / or processes disclosed in this specification.

[0021] In some application scenarios, users using at least two store terminals 130 may include store owners or service personnel within the store.

[0022] In some embodiments, relevant information about the customer flow of multiple stores can be obtained through at least two store terminals 130 and / or other data sources 150, and the predicted customer flow value of the target store in the target time period after the target time point is determined after processing by the server 110. The server 110 can obtain data on the database 140 or save the data to the database 140 during processing. The operations in this specification can be performed by executing program instructions by the processing device 112. The above method is only for convenience of understanding, and the system can also implement the method in this specification in other feasible operation modes.

[0023] In some instances, different functions may be performed on different devices, such as data screening, querying, preprocessing, model training, model execution, etc., which are not limited here.

[0024] In some embodiments, the server 110 , at least two store terminals 130 , and other possible system components may include a database 140 .

[0025] In some embodiments, the server 110 , at least two store terminals 130 , and other possible system components may include a processing device 112 .

[0026] The server 110 and at least two store terminals 130 may be connected via a network 120, and the database 140 may be connected to the server 110 via the network 120, or may be directly connected to the server 110 or located inside the server 110. The database 140 and other data sources 150 may be connected to the network 120 to communicate with one or more components of the store customer flow prediction system 100. One or more components of the store customer flow prediction system 100 may access information or instructions stored in the database 140 and other data sources 150 via the network 120.

[0027] Server 110 can be used to manage resources and process data and / or information from at least one component of the system or an external data source (e.g., a cloud data center). In some embodiments, server 110 can be a single server or a server group. The server group can be centralized or distributed (e.g., server 110 can be a distributed system), can be dedicated, and can also be provided by other devices or systems at the same time. In some embodiments, server 110 can be regional or remote. In some embodiments, server 110 can be implemented on a cloud platform or provided in a virtual manner. As an example only, the cloud platform can include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc., or any combination thereof.

[0028] In some embodiments, the server 110 may include a processing device 112. The processing device 112 may process data and / or information obtained from other devices or system components. The processor may execute program instructions based on these data, information and / or processing results to perform one or more functions described in the present application. In some embodiments, the processing device 112 may include one or more sub-processing devices (e.g., a single-core processing device or a multi-core multi-core processing device). By way of example only, the processing device 112 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic circuit (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), a microprocessor, etc., or any combination thereof.

[0029] The network 120 can connect the components of the system and / or connect the system with external resources. The network 120 allows the components to communicate with each other and with other parts outside the system, and promotes the exchange of data and / or information. In some embodiments, the network 120 can be any one or more of a wired network or a wireless network. For example, the network 120 can include a cable network, an optical fiber network, a telecommunications network, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network (ZigBee), a near field communication (NFC), a bus within a device, a line within a device, a cable connection, etc. or any combination thereof. The network connection between the various parts can be one of the above methods, or multiple methods can be adopted. In some embodiments, the network can be a point-to-point, shared, central, or other topological structure or a combination of multiple topological structures. In some embodiments, the network 120 can include one or more network access points. For example, the network 120 may include wired or wireless network access points, such as base stations and / or network exchange points 120-1, 120-2, ..., through which one or more components of the store customer traffic prediction system 100 can be connected to the network 120 to exchange data and / or information.

[0030] The store terminal 130 refers to one or more terminal devices or software in the store. In some embodiments, the store terminal 130 may be used by one or more users, including the store owner, or other relevant personnel. In some embodiments, the store terminal 130 may be a mobile device, a tablet computer, a laptop computer, a desktop computer, or any combination thereof, and may be any other device with input and / or output functions.

[0031] The database 140 may be used to store data and / or instructions. The database 140 is implemented in a single central server, multiple servers connected by a communication link, or multiple personal devices. In some embodiments, the database 140 may include a large-capacity memory, a removable memory, a volatile read-write memory (e.g., a random access memory RAM), a read-only memory (ROM), etc., or any combination thereof. Exemplarily, the large-capacity storage may include a magnetic disk, an optical disk, a solid-state disk, etc. In some embodiments, the database 140 may be implemented on a cloud platform.

[0032] Other data sources 150 can be used to provide one or more sources of other information to the system. Other data sources 150 can be one or more devices, one or more application programming interfaces, one or more database query interfaces, one or more protocol-based information acquisition interfaces, other ways of acquiring information, or a combination of two or more of the above methods. The information provided by the information source may already exist when the information is extracted, or may be temporarily generated when the information is extracted, or may be a combination of the above methods. In some embodiments, other data sources 150 can be used to provide the system with store review information, order information for historical target time periods, weather information, and traffic conditions around stores, etc.

[0033] Figure 2 FIG. 1 is an exemplary structural flow chart of using a customer flow prediction model to predict the customer flow of a target store in a target time period according to some embodiments of this specification. Figure 2 As shown, the process 200 includes the following steps. In some embodiments, the process 200 may be performed by a processing device (eg, the processing device 112).

[0034] Step 210, obtaining relevant information of multiple stores within a preset range, wherein the multiple stores include a target store and multiple associated stores that are associated with the target store. In some embodiments, this step may be performed by an acquisition module.

[0035] In some embodiments, a store may be a physical sales service place that delivers goods or services to end consumers or institutions. In some embodiments, a target store may be a store whose customer flow in a target time period needs to be predicted, and an associated store may be a store that has an associated relationship with the target store. For example, in a shopping mall, there are bakeries, cafes, clothing stores, fast food restaurants, and movie theaters, among which the fast food restaurant can be used as the target store, and other stores in the mall can be used as associated stores. In some embodiments, the preset range may be a range set based on the prediction needs. For example, the preset range may be within a shopping mall. For another example, the preset range may be a commercial area of ​​a certain area centered on the target store.

[0036] In some embodiments, the relevant information may be data and / or information that directly or indirectly reflects the store's passenger flow in the target time period. For example, the relevant information may include the evaluated information of the store. In some embodiments, the relevant information may include information related to passenger flow of multiple stores in the historical target time period before the target time point. The target time point may be a certain time point in the present or future, the target time period may be a certain time period after the target time point, and the historical target time period may be a certain time period before the target time point. For example, the current time is 9:00 am Beijing time, the target time period may be 14:00-15:00 or 14:30-15:30 Beijing time on the same day, and the historical time period may be 10:00-11:00, 11:00-12:00, etc. Beijing time on the same day. In some embodiments, the target time period and the historical time period may be for a specific period. For example, the target time period and the historical time period may be weekends, Wednesdays, etc. The target time period and the historical time period may be specific time periods in a day. For example, the historical time period may be 15:00-17:00 in the afternoon, and the target time period may be the dinner time period in the evening, such as 17:00-18:00, 18:00-19:00, 19:00-20:00, etc.

[0037] In some embodiments, the target time period and the historical time period may be for a specific period. For example, the target time period and the historical time period may be weekends, Wednesdays, etc. The target time period and the historical time period may be a specific time period in a day. For example, the historical time period may be 15:00-17:00 in the afternoon, and the target time period may be a dinner time period in the evening, such as 17:00-18:00, 18:00-19:00, 19:00-20:00, etc.

[0038] In some embodiments, the relevant information may include at least one of store type, evaluation information, historical target time period order information, historical target time period customer flow, and current customer flow.

[0039] In some embodiments, the store type may be determined according to the types of goods and / or services the store operates. For example, the store type may include a shopping mall, a convenience store, a supermarket, a coffee shop, a restaurant, a library, a cinema, etc.

[0040] In some embodiments, the evaluated information may be a review record of a store by a consumer. For example, the evaluated information may be a review record related to a store displayed on an online platform such as a review website or a food delivery app. The evaluated information may reflect the normal customer flow of the store. For example, the more evaluated information there is, the more people have visited the store to consume. For another example, if the proportion of positive reviews in the evaluated information is relatively large, it means that more people may visit the store to consume in the future.

[0041] In some embodiments, the historical target time period may be a time period that has passed and corresponds to the target time period. For example, the current time is 9:00 a.m. Beijing time on Tuesday, the target time period is 5:00-6:00 p.m. Beijing time on Tuesday, and the historical target time period may be 5:00-6:00 p.m. Beijing time on Monday. Historical target time period orders may refer to orders received and / or completed during the historical target time period. In some embodiments, historical target time period order information may include the number of orders within the historical target time period, order completion status (e.g., completed or canceled, etc.), order evaluation information (e.g., good or bad reviews, etc.), the type of ordered goods, the quantity of ordered goods, and the amount of ordered goods, etc., or any combination thereof.

[0042] In some embodiments, the customer flow may be the number of people in the store. In some embodiments, the current customer flow may be the customer flow at the current time point. In some embodiments, the customer flow in the historical target time period may be the customer flow in the historical target time period. The current customer flow and the customer flow in the historical target time period can largely reflect the customer flow of the store in a certain time period in the future. For example, if the target time period is the peak dining period of Saturday this week, and the peak dining period of Saturday last week is taken as the historical target time period, the customer flow in the historical target time period can largely reflect the customer flow in the target time period.

[0043] The processing device may obtain the relevant information of the store in various ways. For example, the processing device may obtain the type of the store through the registration information of the store obtained from other data sources 150. For another example, the processing device may obtain the store's evaluation information and historical target time period order information by querying the relevant information of the online platform.

[0044] For an example of a processing device obtaining the current passenger flow and the passenger flow in the historical target time period, see Figure 5 and related instructions.

[0045] In some embodiments, the relevant information also includes information related to wireless signals within the store. Information related to wireless signals may include, but is not limited to: whether the store provides wireless signals, the method of accessing wireless signals within the store, and the number of stores within a certain range where the signal strength is greater than a threshold. Information related to wireless signals can affect the flow of customers within the store to a certain extent. For example, under the same conditions, consumers may stay longer in stores that provide wireless signals because they can use the wireless signals in the store for entertainment or work. For another example, consumers who need to use the Internet in stores (e.g., cafes) are more willing to choose stores with stronger wireless signals.

[0046] In some embodiments, the association relationship may include multiple relationships related to customer flow between stores (e.g., between a target store and an associated store, and between an associated store and an associated store). For example, the correlation of business between the target store and the associated store.

[0047] In some embodiments, the association relationship includes a distance relationship, a location relationship and / or a business relationship between multiple stores.

[0048] In some embodiments, the distance relationship may represent the distance between stores. The distance relationship may reflect the mutual influence of customer flow between stores to a certain extent. For example, if the customer flow of an amusement park increases, the customer flow of the adjacent toy store will also increase accordingly.

[0049] In some embodiments, the location relationship includes at least one of whether the target store and the associated store are in the same building, whether they are across the street, and whether they belong to the same business district. The location relationship can reflect the mutual influence of customer flow between stores to a certain extent. For example, the customer flow of an associated store in the same building as the target store has a greater impact on the target store than that of an associated store that needs to cross the street.

[0050] In some embodiments, the business relationship includes business similarity and / or business succession.

[0051] In some embodiments, business similarity can be the degree of similarity of business between stores. For example, the business similarity between stores selling mobile phone accessories is high, while the business similarity between stores selling mobile phone accessories and catering stores is low. Business similarity can reflect the mutual influence of customer flow between stores to a certain extent. For example, if the target store is a catering store, the customer flow of associated stores that are also catering stores will have a greater impact on the customer flow of the target store.

[0052] In some embodiments, business inheritance can be the possibility that when a consumer consumes at one store, he or she will simultaneously or subsequently consume at other stores. For example, since a large proportion of consumers choose to have lunch near the cinema after watching a movie, the inheritance between the cinema and the restaurants near the cinema is relatively high.

[0053] Step 220, based on the customer flow prediction model, the relevant information and association relationships of the multiple stores are processed to determine the customer flow of the target store in the target time period after the target time point. In some embodiments, this step can be performed by a customer flow prediction module.

[0054] In some embodiments, the customer flow prediction model may be a machine learning model that has been trained to predict the store customer flow in a certain time period in the future. The customer flow prediction model may be constructed based on a variety of machine learning models, such as a Long Short-Term Memory (LSTM) model, a Gate Recurrent Unit (GRU) model, and a Graph Neural Networks (GNN) model.

[0055] In some embodiments, the customer flow prediction model may include a customer flow embedding layer and a prediction layer.

[0056] In some embodiments, the customer flow embedding layer can be used to process relevant information and associations, and extract customer flow characteristics of multiple stores. In some embodiments, the input of the customer flow embedding layer can be feature data obtained by preprocessing relevant information and associations (for example, normalization, discretization, etc.), and the output of the customer flow embedding layer can be customer flow characteristics of multiple stores in vector form. The customer flow embedding layer can be constructed based on a variety of machine learning models. In some embodiments, the customer flow embedding layer may include but is not limited to: Long Short-Term Memory (LSTM) model, Gate Recurrent Unit (GRU) model and Graph Neural Networks (GNN) model, etc.

[0057] In some embodiments, the passenger flow embedding layer is a GNN model. For more description of the GNN model, see Figure 3 and related instructions.

[0058] In some embodiments, the prediction layer can determine the customer flow of the target store in the target time period after the target time point based on the customer flow characteristics of the target store extracted by the customer flow embedding layer. In some embodiments, the input of the prediction layer can be the customer flow characteristics of the target store, and the output of the prediction layer can be the customer flow of the target store in the target time period after the target time point. The customer flow output by the prediction layer can be represented in a variety of ways. For example, the customer flow level can be used to represent the customer flow. The customer flow level can be represented by a number, for example, 1 means that the customer flow is less than 3 people, and 10 means that the number of people in the store has reached saturation. The prediction layer can be built based on a variety of machine learning models. In some embodiments, the prediction layer may include, but is not limited to: decision tree (Decision Tree), random forest (Random Forest, RF) and support vector machine (Support Vector Machine, SVM), etc.

[0059] In some embodiments, the prediction layer may be a fully connected layer. The fully connected layer may predict the customer flow of the target store in a target time period after a target time point based on the customer flow features of the target store extracted by the customer flow embedding layer.

[0060] In some embodiments, the passenger flow prediction model is trained by end-to-end learning. For more information about the training method of the passenger flow prediction model, please refer to Figure 4 and related instructions.

[0061] The methods described in some embodiments of this specification not only consider the relevant information of the target store itself when predicting the customer flow of the target store, but also consider the mutual influence of the customer flow between stores, thereby making the prediction result more accurate. In some embodiments, a machine learning model is used to process the relevant information of multiple stores and the association relationship between multiple stores, so that more accurate test results can be obtained with higher processing efficiency.

[0062] It should be noted that the above description of process 200 is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to process 200 under the guidance of this specification. However, these modifications and changes are still within the scope of this specification. For example, the input of the prediction layer of the customer flow prediction model can be the customer flow characteristics of multiple stores, and the output can be the customer flow of multiple stores in the target time period.

[0063] Figure 3 It is an exemplary schematic diagram of a store traffic map shown in some embodiments of this specification.

[0064] In some embodiments, the customer flow embedding layer in the customer flow prediction model is a GNN model, which uses multiple stores as nodes of a graph and association relationships as edges of the graph. For ease of description, in one or more embodiments of this specification, the graph corresponding to the GNN model is called a store customer flow graph.

[0065] like Figure 3 As shown, the store customer flow graph includes multiple nodes (e.g., store 1, store 2, etc.), where store 1 is the target store and other stores are associated stores. The store customer flow graph also includes multiple edges (e.g., edge 1, edge 2, etc.), where the edges are used to represent the association relationship between multiple stores.

[0066] In some embodiments, the node features of the store customer flow graph may include relevant information and environmental information. In some embodiments, environmental information may include but is not limited to: weather information, traffic conditions around the store, etc. Environmental information can affect the store's customer flow to a certain extent. For example, the store's customer flow may be relatively less on rainy days than on sunny days. For another example, if there is traffic congestion around the store, consumers may choose other stores with more convenient transportation to consume, which will also affect the store's customer flow. For more descriptions of relevant information, see Figure 2 and related instructions.

[0067] In some embodiments, the processing device may pre-process the relevant information and environmental information as node features of the store customer flow graph. For example, the processor may use one-hot encoding to digitize the store type, evaluated information, etc. For another example, the processor may bin the order quantity to discretize it.

[0068] In some embodiments, the edge feature of the store customer flow graph may be an association relationship. For more information about the association relationship, see Figure 2 and related instructions.

[0069] In some embodiments, the processing device may pre-process the association relationship and use it as the edge feature of the store customer flow graph. For example, the processor may use one-hot encoding to digitize the location relationship, business relationship, etc. For another example, the processor may bin the distance between stores to discretize it.

[0070] The methods described in some embodiments of this specification use a GNN model to process relevant information of a target store and relevant information of associated stores, so that the customer flow features extracted by the customer flow embedding layer fully integrate the impact of the customer flow of associated stores on the customer flow of the target store, thereby making the customer flow prediction results of the target store more accurate.

[0071] Figure 4 It is an exemplary flow chart of the training method of the passenger flow prediction model shown in some embodiments of this specification.

[0072] Step 410, obtaining training samples. In some embodiments, this step may be performed by a training module.

[0073] In some embodiments, the training sample may include sample-related information and sample environment information of multiple sample stores, and sample association relationships between multiple sample stores, wherein the multiple sample stores include sample target stores and sample associated stores. The label of the training sample is the customer flow value of the sample target store in the sample historical target time period. The training sample can be historical real data, and the processing device can obtain the training sample in a common manner, such as network download, database extraction, etc.

[0074] Step 420: Perform end-to-end training on the passenger flow prediction model based on the training samples. In some embodiments, this step may be performed by a training module.

[0075] In some embodiments, the passenger flow prediction model can be trained end-to-end based on the training samples. In the process of end-to-end training of the passenger flow prediction model, the processing device can input the training sample data into the passenger flow embedding layer of the passenger flow prediction model, and then use the loss function to obtain the error value based on the passenger flow prediction value output by the prediction layer of the passenger flow prediction model, wherein the loss function is constructed based on the prediction value and the label of the sample data. Then the processing device can use the backpropagation algorithm (BP algorithm) according to the error value to calculate the gradient corresponding to all the parameters that need to be learned in the passenger flow prediction model (including the parameters of the passenger flow embedding layer, the parameters of the prediction layer, etc.), and adjust the parameters of the passenger flow prediction model (including the parameters of the passenger flow embedding layer, the parameters of the prediction layer, etc.) in combination with the stochastic gradient descent algorithm until the model converges.

[0076] The methods described in some embodiments of this specification, in the process of training the customer flow prediction model, eliminate the data labeling process of the customer flow features output by the customer flow embedding layer through end-to-end learning, saving a lot of data labeling workload and reducing the time cycle of model training. At the same time, because the end-to-end training method can fully consider the correlation between the parameters of each processing layer, the customer flow prediction model obtained by end-to-end learning can more accurately predict the customer flow of the store.

[0077] It should be noted that the above description of process 400 is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to process 400 under the guidance of this specification. However, these modifications and changes are still within the scope of this specification. For example, step 420 can be divided into two steps 420-1 and 420-2, in which a loss function is constructed in step 420-1, and the model is trained in step 420-2.

[0078] Figure 5It is an exemplary flow chart of a method for obtaining current store customer flow / historical target time period customer flow as shown in some embodiments of this specification.

[0079] Step 510: extract at least one frame of image data from the video data of the store in the current time period / historical target time period. In some embodiments, this step may be performed by a customer flow acquisition module.

[0080] In some embodiments, the processing device may obtain video data of the store in the current time period and / or the historical target time period through the store terminal 130, wherein the video data may be obtained by shooting by the monitoring device. The processing device may also obtain video data in other ways, which are not limited here. The format of the video data may include but is not limited to: MP4 format, MKV format, M4A format, MP3 format, FLV format and 3GP format, etc.

[0081] In some embodiments, the processing device may extract at least one frame of image data from the video data for subsequent processing. For example, the processing device may extract one or more frames of image data with higher resolution from the video data.

[0082] In some embodiments, in order to obtain image data with higher picture quality, the processing device may extract image data with a code stream value greater than a preset threshold from the video data. The preset threshold may be a code stream value preset based on experience.

[0083] Step 520: Process at least one frame of image data based on the trained customer flow determination model to determine the current customer flow / historical target time period customer flow of the store. In some embodiments, this step can be performed by a customer flow acquisition module.

[0084] In some embodiments, the customer flow determination model can process the image data to obtain the current customer flow of the store and / or the customer flow of the historical target time period. For example, the image data corresponding to the current time period is input into the customer flow determination model to obtain the current customer flow of the store. For another example, the image data corresponding to the historical target time period is input into the customer flow determination model to obtain the customer flow of the store in the historical target time period.

[0085] In some embodiments, the input of the passenger flow determination model may be image data. In some embodiments, the output of the passenger flow determination model may be passenger flow. The passenger flow output by the passenger flow determination model may be represented in a variety of ways. For example, the passenger flow level may be used to represent the passenger flow. For another example, the number of people may be used to represent the passenger flow.

[0086] In some embodiments, the passenger flow determination model may include but is not limited to: a convolutional neural network model, a deep neural network model, etc.

[0087] In some embodiments, the passenger flow determination model is a convolutional neural network model, and the parameters of the convolution kernel of the passenger flow determination model are obtained based on the migration of the convolution kernel in the pre-trained image recognition model. For example, multiple sets of sample data can be used to train the image recognition model, wherein the sample data includes image data and labels as model input, the label is the number of people in the image, and the convolution kernel of the trained image recognition model is used as the convolution kernel of the passenger flow determination model. In some embodiments, the convolution kernel in the image recognition model can include multiple network layers, and the following operations can be performed on the input image data: convolution operation (Convolution, Conv), normalization operation (Normalization), activation operation (e.g., ReLU function) and dimensionality reduction operation (e.g., maximum pooling), etc.

[0088] The methods described in some embodiments of this specification use a convolutional neural network model to process image data, and can obtain more accurate current passenger flow / historical target time period passenger flow.

[0089] It should be noted that the above description of process 500 is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to process 500 under the guidance of this specification. However, these modifications and changes are still within the scope of this specification. For example, the passenger flow judgment model can be constructed based on a deep neural network model.

[0090] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0091] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0092] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0093] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0094] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0095] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0096] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A method for predicting store customer flow, characterized in that: The method comprises: Obtaining relevant information of multiple stores within a preset range, the multiple stores including a target store and multiple associated stores associated with the target store, the relevant information including at least one of store types, evaluated information, order information during the historical target time period, customer flow during the historical target time period, current customer flow, and wireless signal related information of the multiple stores during the historical target time period before the target time point; Processing the relevant information and the association relationship of the multiple stores based on the customer flow prediction model to determine the customer flow of the target store in a target time period after the target time point; The passenger flow prediction model is a machine learning model, and the passenger flow prediction model includes a passenger flow embedding layer and a prediction layer; wherein, The customer flow embedding layer determines the customer flow characteristics of the multiple stores; the customer flow embedding layer is a GNN model, with the multiple stores as nodes of the graph and the association relationships as edges of the graph; The prediction layer determines the customer flow of the target store in the target time period based on the customer flow characteristics of the multiple stores, and the prediction layer is a fully connected layer; The passenger flow prediction model is trained by end-to-end learning; The current passenger flow or the passenger flow in the historical target time period is obtained by: For each of the multiple stores, extract at least one frame of image data from the video data of the store in the current time period or the historical target time period, wherein the at least one frame of image data is an image in the video data whose bit rate value is greater than a preset threshold; The at least one image data is processed based on a trained customer flow judgment model to determine the current customer flow of the store or the customer flow in the historical target time period. The customer flow judgment model is a convolutional neural network model, and the parameters of the convolution kernel of the customer flow judgment model are obtained based on the convolution kernel migration in a pre-trained image recognition model.

2. A store customer flow prediction system, characterized in that: The system includes an acquisition module and a determination module; The acquisition module is used to acquire relevant information of multiple stores within a preset range, the multiple stores including a target store and multiple associated stores associated with the target store, the relevant information including at least one of store types, evaluated information, order information during the historical target time period, customer flow during the historical target time period, current customer flow, and wireless signal related information of the multiple stores during the historical target time period before the target time point; The determination module is used to process the relevant information of the multiple stores and the association relationship based on the customer flow prediction model to determine the customer flow of the target store in a target time period after the target time point; The passenger flow prediction model is a machine learning model, and the passenger flow prediction model includes a passenger flow embedding layer and a prediction layer; wherein, The customer flow embedding layer determines the customer flow characteristics of the multiple stores; the customer flow embedding layer is a GNN model, with the multiple stores as nodes of the graph and the association relationships as edges of the graph; The prediction layer determines the customer flow of the target store in the target time period based on the customer flow characteristics of the multiple stores, and the prediction layer is a fully connected layer; The passenger flow prediction model is trained by end-to-end learning; The current passenger flow or the passenger flow in the historical target time period is obtained by: For each of the multiple stores, extract at least one frame of image data from the video data of the store in the current time period or the historical target time period, wherein the at least one frame of image data is an image in the video data whose bit rate value is greater than a preset threshold; The at least one image data is processed based on a trained customer flow judgment model to determine the current customer flow of the store or the customer flow in the historical target time period. The customer flow judgment model is a convolutional neural network model, and the parameters of the convolution kernel of the customer flow judgment model are obtained based on the convolution kernel migration in a pre-trained image recognition model.

3. A store customer flow prediction device, comprising a processor, characterized in that: The processor is used to execute the store customer flow prediction method described in claim 1.

4. A computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the store customer flow prediction method as claimed in claim 1.

Images