Intention prediction method and device, electronic equipment, readable storage medium and chip
By cleaning and analyzing the store video customer flow data and processing customer intentions in batches, the inefficiency problem caused by traditional manual judgment is solved, and more efficient and accurate customer intention prediction is achieved.
Patent Information
- Application Number
- CN202411486659.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-23
- Publication Date
- 2025-02-28
AI Technical Summary
The method of human judgment and manual entry of customer levels leads to slow calculation and processing efficiency of customer wish prediction and insufficient business follow-up speed.
By obtaining the original customer flow video data, determining the frame image and frame generation time, filtering non-customer personnel data, calculating customer batches in batches, and analyzing customer behavior parameters, and determining customer intention prediction data using prediction algorithms.
It improves the efficiency and accuracy of customer flow data analysis, helps stores accurately classify customers, optimize resource allocation, and reduce personnel analysis costs.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image analysis technology, and in particular to an intention prediction method, device, electronic device, readable storage medium and chip. Background Art
[0002] At present, with the booming new energy vehicle industry, the sales and display model of automobiles is undergoing changes, gradually transforming from the traditional 4S store model to the shopping mall and supermarket model that is closer to consumers. With its superior location in a bustling shopping mall, the automobile supermarket attracts a huge flow of customers every day, but the store reception of customers is based on the business system, which predicts and calculates customer willingness through human judgment and manual entry of customer levels, resulting in low business gains and slow processing speed. Summary of the invention
[0003] The purpose of the embodiments of the present invention is to provide an intention prediction method, device, electronic device, readable storage medium and chip, which can solve the problems of slow processing efficiency and insufficient business follow-up speed caused by human judgment and manual entry of customer levels to perform customer intention prediction calculations.
[0004] In view of this, an embodiment of the first aspect of the present invention provides an intention prediction method.
[0005] An embodiment of the second aspect of the present invention provides an intention prediction device.
[0006] An embodiment of a third aspect of the present invention provides an electronic device.
[0007] An embodiment of a fourth aspect of the present invention provides a readable storage medium.
[0008] An embodiment of the fifth aspect of the present invention provides a chip.
[0009] In order to achieve the above-mentioned purpose, an embodiment of the first aspect of the present invention provides an intention prediction method, which includes: acquiring original passenger flow video data; determining frame images and frame generation time corresponding to the original passenger flow video data; determining first customer video data in the original passenger flow video data according to the frame image; determining at least one customer batch corresponding to the first customer video data according to the frame generation time and the frame image; determining a first behavior parameter corresponding to the customer batch according to the frame generation time and the frame image; and determining intention prediction data corresponding to the customer batch according to the first behavior parameter.
[0010] According to the intention prediction method proposed in the present invention, by filtering and cleaning the video passenger flow data collected by the store camera, after determining the video data of non-customer-related personnel other than store clerks, cleaners, takeaways, etc. in the video passenger flow data, based on the characteristics of the batches of car-buying customers in the automobile industry, the store video passenger flow data is calculated in batches, and the behavioral actions, business actions, etc. of different customer groups are analyzed in the form of customer batches, and combined with the prediction algorithm rules, the purchase intention degree of the main car-buying objects corresponding to each batch is determined as the intention prediction result. The traditional method in the automobile industry requires the business personnel of the automobile store to inquire and manually perform statistics when distinguishing batches. Not only can the customer batch data not be calculated on the same day, but also due to the high mobility of automobile stores, the business personnel cannot ask for opinions from every customer, resulting in poor accuracy of the batch data of automobile stores and low processing rate. Through the intention prediction method proposed in the present invention, after the original passenger flow video data is cleaned, invalid data of non-customer personnel in the original passenger flow video data is removed, and the customer personnel units in the video data are divided into batches. The customer groups in the automobile industry are mostly multiple people or family members who look at cars together. Therefore, according to the image data in the video data, analysis is performed to determine the batches corresponding to the customer personnel units, and behavioral actions, business actions, etc. of different customer groups are analyzed in the form of customer batches, thereby improving the efficiency and accuracy of passenger flow data analysis, helping stores to accurately classify and manage customers, optimize resource allocation, and reduce personnel analysis costs.
[0011] In the above technical solution, the first customer video data in the original passenger flow video data is determined according to the frame image, including: determining the employee image identification and the invalid personnel image identification; determining the non-customer unit identification according to the employee image identification and the invalid personnel image identification; filtering the non-customer unit identification in the frame image to obtain the first frame image; and determining the first customer video data according to the first frame image.
[0012] In this technical solution, after the store camera collects the video passenger flow data, the original passenger flow video data is cleaned and processed to remove the invalid data of non-customer personnel in the original passenger flow video data. Specifically, the store employees can be set with a special batch identity code (Identity document, id) as an employee image identifier through image recognition technology, and special work clothes marks can be set as invalid personnel image identifiers for other staff such as express delivery, takeout, and cleaning through image recognition technology. Through the data cleaning and filtering method, the employee image identifier and the invalid personnel image identifier are filtered and cleaned from the original passenger flow video data as invalid data of non-customer personnel, and the first customer video data of only the customer unit is obtained. The frame image in the first customer video data is the frame image from which the employee image identifier and the invalid personnel image identifier are filtered out. The invalid data of non-customer personnel in the original passenger flow video data is cleaned through image recognition technology and data cleaning and filtering methods, and the passenger flow of automobile supermarkets is screened to improve the accuracy of customer intention prediction.
[0013] In the above technical solution, at least one customer batch corresponding to the first customer video data is determined according to the frame generation time and the frame image, including: determining at least one customer unit in the first customer video data; determining the store entry time and store exit time of the customer unit according to the frame generation time; determining the motion trajectory parameters of the customer unit according to the first frame image; determining at least one customer batch according to the store entry time, store exit time and motion trajectory parameters of the customer unit.
[0014] In this technical solution, the image data in the video data is analyzed, the store video customer flow data is calculated in batches, and the batches corresponding to the customer personnel units are determined. Specifically, the frame image in the first customer video data is analyzed by image recognition technology to determine the customer unit contained in the frame image and the frame generation time corresponding to the customer unit. The store area range is set, and the time when the customer unit first appears in the store area range is determined as the entry time according to the frame generation time corresponding to the customer unit, and the time when the customer unit last appears in the store area range is the departure time. The location coordinates of the customer unit in the frame image are analyzed by image recognition technology to determine the in-store movement trajectory of the customer unit, and the customer batch to which the customer unit belongs is determined based on the in-store movement trajectory, entry time, and departure time of the customer unit.
[0015] It is understandable that the purpose of setting the store area range is to determine the entry and exit time of the customer unit. The store area range can be adjusted through the image acquisition settings of the store camera. There is no requirement for the division of the store area range, as long as the entry and exit time of the customer unit in the first customer video data can be determined.
[0016] In the above technical scheme, at least one customer batch is determined according to the entry time, exit time and motion trajectory parameters of the customer units, including: determining a time threshold and a trajectory overlap threshold; determining the difference between the entry time of the first customer unit and the entry time of the second customer unit as the first time parameter; determining the difference between the exit time of the first customer unit and the exit time of the second customer unit as the second time parameter; determining the difference between the motion trajectory parameters of the first customer unit and the motion trajectory parameters of the second customer unit as the first trajectory parameter; and determining that the first customer unit and the second customer unit are the same customer batch when the first time parameter is less than the time threshold, the second time parameter is less than the time threshold and the first trajectory parameter is greater than the trajectory overlap threshold.
[0017] In this technical solution, the store business operator analyzes the historical customer flow data of the store background, sets the business parameter adjustment algorithm, and determines the time threshold and trajectory overlap threshold according to the business parameter adjustment algorithm, that is, when the time for a customer unit to enter and leave the store is within the set threshold range, it is determined that the customer unit enters the store and leaves the store at the same time. When the overlap of the customer unit's in-store movement trajectory exceeds the threshold, it is determined that the customer unit's in-store movement trajectory, such as looking at cars, is the same. The customer batch corresponding to the customer unit is determined by the two aspects of the time of entering and leaving the store and the movement trajectory. Specifically, the store entry time of the first customer unit and the second customer unit is determined according to the frame generation time, and the store entry time difference of the two customer units is determined according to the store entry time of the first customer unit and the store entry time of the second customer unit; the store departure time of the first customer unit and the second customer unit is determined according to the frame generation time, and the store departure time difference of the two customer units is determined according to the store departure time of the first customer unit and the store departure time of the second customer unit. When the difference in the time of entering the store and the difference in the time of leaving the store are both less than the time threshold, it is determined that the two customer units enter and leave the store at the same time. When the difference in the time of entering the store is greater than the time threshold or the time of leaving the store is greater than the time threshold, the two customer units cannot be determined to enter and leave the store at the same time. The motion trajectory parameters of the first customer unit and the motion trajectory parameters of the second customer unit are obtained according to the frame image analysis. The motion trajectory parameter difference is determined by the motion trajectory parameters of the first customer unit and the motion trajectory parameters of the second customer unit. The motion trajectory parameter difference is the motion trajectory overlap. When the motion trajectory overlap exceeds the trajectory overlap threshold, it is determined that the two customer units are in the same in-store motion trajectory. When the first time parameter is less than the time threshold, the second time parameter is less than the time threshold, and the first trajectory parameter is greater than the trajectory overlap threshold, that is, the two customer units enter the store at the same time, leave the store at the same time, and have the same in-store motion trajectory, the two customer units are the same customer batch, and the customer units in the first customer video data are batched to obtain at least one customer batch.
[0018] In the above technical solution, the first behavior parameter corresponding to the customer batch is determined according to the frame generation time and the frame image, including: determining the behavior action data of at least one customer unit in the customer batch according to the first frame image and the frame generation time; determining the behavior parameter corresponding to the customer unit according to the behavior action data; and arranging at least one behavior parameter in the customer batch in descending order to determine the first behavior parameter corresponding to the customer batch.
[0019] In this technical solution, the customer batch in the first customer video data is analyzed by image recognition technology to determine the main car-buying customer units in a customer batch, and to determine the behavior parameters of the main car-buying customer units. Specifically, the behavior action data of all customer units in a customer batch are determined according to the first frame image and the frame generation time. The behavior action data include data corresponding to the in-store behavior actions such as whether to view the car, the length of time for viewing the car, the length of time for entering the car, the order of viewing the car, the length of time for negotiating with the salesperson, and the length of time for business negotiation. Store business operators analyze the behavior action data based on the historical customer flow data of the store background, determine the weight proportion corresponding to the behavior action data, and combine the business parameter adjustment algorithm to obtain the behavior parameters of all customer units in the same customer batch. The behavior parameters are the results of the weight proportion calculation of the behavior action data of all customer units in the same customer batch. The behavior parameters of all customer units in the same customer batch are arranged in descending order, and the first one in descending order is the first behavior parameter, that is, the behavior parameters of the main car-buying objects in the same customer batch are used as the behavior parameters of the customer batch.
[0020] In the above technical scheme, intention prediction data corresponding to a customer batch is determined according to the first behavior parameter, including: obtaining a customer intention degree prediction database; determining the customer intention level and transaction probability parameters corresponding to the first behavior parameter according to the customer intention degree prediction database; determining the intention prediction parameters corresponding to the customer batch according to the customer intention level and transaction probability parameters; determining a batch visualization analysis report according to the frame generation time and the customer batch; and determining the intention prediction data according to the batch visualization analysis report and the intention prediction parameters.
[0021] In this technical solution, store business operators obtain a customer intention prediction database based on the historical customer flow data of the store background, and determine the customer intention prediction algorithm based on the data in the customer intention prediction database, that is, the store time, car viewing time, car entry time, whether there is car viewing behavior, car entry behavior, negotiation time with sales staff, customer movement trajectory in the store, car viewing order, whether test drive, business negotiation key parameters and other business parameters. According to the customer intention prediction algorithm, the customer intention level and transaction probability parameters corresponding to the first behavior parameter are determined. And the customer batch visualization analysis report corresponding to the store is determined through frame generation, that is, the store customer batch entry and exit situation, car viewing situation, etc. corresponding to each time period, and the batch visualization analysis report and intention prediction parameters are used as intention prediction data to predict the customer intention level corresponding to the store, improve the efficiency of accurate classification management of customers in the store, optimize resource allocation, and reduce personnel analysis costs.
[0022] An embodiment of the second aspect of the present invention provides an intention prediction device, which includes: an acquisition module for acquiring original passenger flow video data; an image recognition module for determining frame images and frame generation time corresponding to the original passenger flow video data; a customer recognition module for determining the first customer video data in the original passenger flow video data based on the frame image; a batch determination module for determining at least one customer batch corresponding to the first customer video data based on the frame generation time and the frame image; a behavior determination module for determining a first behavior parameter corresponding to the customer batch based on the frame generation time and the frame image; and a prediction module for determining intention prediction data corresponding to the customer batch based on the first behavior parameter.
[0023] According to the intention prediction device provided by the embodiment of the present application, it mainly includes: an acquisition module, an image recognition module, a customer recognition module, a batch determination module, a behavior determination module and a prediction module. The original video passenger flow data collected by the store camera is acquired through the acquisition module; the frame image and frame generation time in the original passenger flow video data are determined through the image recognition module; the original video passenger flow data is filtered and cleaned through the customer recognition module to remove the video data of irrelevant personnel such as store clerks, cleaners, and deliverymen, and determine the first customer video data corresponding to the customer unit; through the batch determination module, the store video passenger flow data is settled in batches to determine the customer batch in the first customer video data; through the behavior determination module, the behavior actions, business actions, etc. of the customer units in the customer batch are analyzed to determine the behavior parameters of the main car purchasing objects in the customer batch as the first behavior parameters corresponding to the customer batch; the purchase intention degree of the main car purchasing objects corresponding to each customer batch is determined through the prediction module in combination with the prediction algorithm rules, and combined with the batch visualization analysis report as the intention prediction data.
[0024] An embodiment of the third aspect of the present invention provides an electronic device, including a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the steps of the intention prediction method in the first aspect are implemented.
[0025] An embodiment of the fourth aspect of the present invention provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the intention prediction method in the first aspect are implemented.
[0026] An embodiment of the fifth aspect of the present invention provides a chip, which includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the steps of the intention prediction method in the first aspect.
[0027] Additional aspects and advantages of the present invention will become apparent from the following description or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 A schematic diagram of a process flow of an intention prediction method according to an embodiment of the present application is shown; Figure 2 A schematic diagram of a process flow of an intention prediction method according to an embodiment of the present application is shown; Figure 3 A schematic diagram of a process flow of an intention prediction method according to an embodiment of the present application is shown; Figure 4 A schematic diagram of a process flow of an intention prediction method according to an embodiment of the present application is shown; Figure 5 A schematic diagram of a process flow of an intention prediction method according to an embodiment of the present application is shown; Figure 6 A schematic diagram of a process flow of an intention prediction method according to an embodiment of the present application is shown; Figure 7 A schematic block diagram of the structure of an intention prediction device according to an embodiment of the present application is shown; Figure 8 A block diagram of the structure of an electronic device according to an embodiment of the present application is shown.
[0029] in, Figure 7 and Figure 8 The corresponding relationship between the reference numerals and the component names is as follows: 900: intention prediction device; 902: acquisition module; 904: image recognition module; 906: customer recognition module; 908: batch determination module; 910: behavior determination module; 912: prediction module; 1000: electronic device; 1109: memory; 1110: processor. DETAILED DESCRIPTION
[0030] In order to more clearly understand the above-mentioned purposes, features and advantages of the embodiments of the present invention, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.
[0031] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the embodiments of the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present application is not limited to the specific embodiments disclosed below.
[0032] The following is combined with Figures 1 to 8 , the intention prediction method, device, electronic device, readable storage medium and chip provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios.
[0033] This embodiment provides a method for predicting intentions. Figure 1 As shown, including: Step S102: obtaining original passenger flow video data; Step S104: determining the frame image and frame generation time corresponding to the original passenger flow video data; Step S106: determining the first customer video data in the original customer flow video data according to the frame image; Step S108: determining at least one customer batch corresponding to the first customer video data according to the frame generation time and the frame image; Step S110: determining a first behavior parameter corresponding to a customer batch according to the frame generation time and the frame image; Step S112: Determine the intention prediction data corresponding to the customer batch according to the first behavior parameter.
[0034] According to the intention prediction method proposed in the present invention, by filtering and cleaning the video passenger flow data collected by the store camera, after determining the video data of non-customer-related personnel other than store clerks, cleaners, takeaways, etc. in the video passenger flow data, based on the characteristics of the batches of car-buying customers in the automobile industry, the store video passenger flow data is calculated in batches, and the behavioral actions, business actions, etc. of different customer groups are analyzed in the form of customer batches, and combined with the prediction algorithm rules, the purchase intention degree of the main car-buying objects corresponding to each batch is determined as the intention prediction result. The traditional method in the automobile industry requires the business personnel of the automobile store to inquire and manually perform statistics when distinguishing batches. Not only can the customer batch data not be calculated on the same day, but also due to the high mobility of automobile stores, the business personnel cannot ask for opinions from every customer, resulting in poor accuracy of the batch data of automobile stores and low processing rate. Through the intention prediction method proposed in the present invention, after the original passenger flow video data is cleaned, invalid data of non-customer personnel in the original passenger flow video data is removed, and the customer personnel units in the video data are divided into batches. The customer groups in the automobile industry are mostly multiple people or family members who look at cars together. Therefore, according to the image data in the video data, analysis is performed to determine the batches corresponding to the customer personnel units, and behavioral actions, business actions, etc. of different customer groups are analyzed in the form of customer batches, thereby improving the efficiency and accuracy of passenger flow data analysis, helping stores to accurately classify and manage customers, optimize resource allocation, and reduce personnel analysis costs.
[0035] In one embodiment, Figure 2 As shown, determining the first customer video data in the original customer flow video data according to the frame image includes: Step S1062: Determine employee image identification and invalid personnel image identification; Step S1064: determining the non-client unit identification according to the employee image identification and the invalid personnel image identification; Step S1066: filtering non-customer unit identifiers in the frame image to obtain a first frame image; Step S1068: Determine the first client video data according to the first frame image.
[0036] In this embodiment, after the store camera collects the video passenger flow data, the original passenger flow video data is cleaned and processed to remove the invalid data of non-customer personnel in the original passenger flow video data. Specifically, the store employees can be set with a special batch identity code (Identity document, id) as an employee image identifier through image recognition technology, and special work clothes marks can be set as invalid personnel image identifiers for other staff such as express delivery, takeout, and cleaning through image recognition technology. The employee image identifier and the invalid personnel image identifier are filtered and cleaned from the original passenger flow video data as invalid data of non-customer personnel through the data cleaning and filtering method, and the first customer video data of only the customer unit is obtained. The frame image in the first customer video data is the frame image from which the employee image identifier and the invalid personnel image identifier are filtered out. The invalid data of non-customer personnel in the original passenger flow video data is cleaned through image recognition technology and data cleaning and filtering methods, and the passenger flow of the automobile supermarket store is screened to improve the accuracy of customer intention prediction.
[0037] In one embodiment, Figure 3 As shown, determining at least one customer batch corresponding to the first customer video data according to the frame generation time and the frame image includes: Step S1082: determining at least one client unit in the first client video data; Step S1084: determining the customer unit's store entry time and store exit time according to the frame generation time; Step S1086: determining the motion trajectory parameters of the client unit according to the first frame image; Step S1088: Determine at least one customer batch based on the store entry time, store exit time and movement trajectory parameters of the customer units.
[0038] In this embodiment, the image data in the video data is analyzed, the store video customer flow data is calculated in batches, and the batches corresponding to the customer personnel units are determined. Specifically, the frame image in the first customer video data is analyzed by image recognition technology to determine the customer unit contained in the frame image and the frame generation time corresponding to the customer unit. The store area range is set, and the time when the customer unit first appears in the store area range is determined as the entry time according to the frame generation time corresponding to the customer unit, and the time when the customer unit last appears in the store area range is the departure time. The location coordinates of the customer unit in the frame image are analyzed by image recognition technology to determine the in-store movement trajectory of the customer unit, and the customer batch to which the customer unit belongs is determined based on the in-store movement trajectory, entry time, and departure time of the customer unit.
[0039] It is understandable that the purpose of setting the store area range is to determine the entry and exit time of the customer unit. The store area range can be adjusted through the image acquisition settings of the store camera. There is no requirement for the division of the store area range, as long as the entry and exit time of the customer unit in the first customer video data can be determined.
[0040] Specifically, based on the customer flow data collected by the existing general video surveillance, the multi-person batch division algorithm set by the store business operator is superimposed to identify whether the customers entering the store are from the same batch. According to each user's store entry time, store exit time, in-store movement trajectory, and in-store behavior, multiple people can be determined as one customer batch when the entry and exit time are within the set threshold range, the degree of overlap of in-store movement trajectories exceeds a certain threshold ratio, and there are conversations and other behaviors in the store. Through the multi-person batch division algorithm, the customer flow in shopping malls and supermarkets is divided, and multiple customer units are divided into the same customer batch to improve processing efficiency.
[0041] In one embodiment, Figure 4 As shown, at least one customer batch is determined according to the store entry time, store exit time and movement trajectory parameters of the customer units, including: Step S10882: Determine the time threshold and the trajectory overlap threshold; Step S10884: determining the difference between the store entry time of the first customer unit and the store entry time of the second customer unit as the first time parameter; Step S10886: determining that the difference between the check-out time of the first customer unit and the check-out time of the second customer unit is a second time parameter; Step S10888: determining that the difference between the motion trajectory parameter of the first customer unit and the motion trajectory parameter of the second customer unit is the first trajectory parameter; Step S10890: When the first time parameter is less than the time threshold, the second time parameter is less than the time threshold, and the first trajectory parameter is greater than the trajectory overlap threshold, it is determined that the first customer unit and the second customer unit are the same customer batch.
[0042] In this embodiment, the store business operator analyzes the historical customer flow data of the store background, sets the business parameter adjustment algorithm, and determines the time threshold and trajectory overlap threshold according to the business parameter adjustment algorithm, that is, when the time for a customer unit to enter and leave the store is within the set threshold range, it is determined that the customer unit enters the store and leaves the store at the same time. When the overlap of the customer unit's in-store movement trajectory exceeds the threshold, it is determined that the customer unit's in-store movement trajectory, such as looking at cars, is the same, and the customer batch corresponding to the customer unit is determined by the two aspects of the time of entering and leaving the store and the movement trajectory. Specifically, the store entry time of the first customer unit and the second customer unit is determined according to the frame generation time, and the store entry time difference of the two customer units is determined according to the store entry time of the first customer unit and the store entry time of the second customer unit; the store departure time of the first customer unit and the second customer unit is determined according to the frame generation time, and the store departure time difference of the two customer units is determined according to the store departure time of the first customer unit and the store departure time of the second customer unit. When the difference in the time of entering the store and the difference in the time of leaving the store are both less than the time threshold, it is determined that the two customer units enter and leave the store at the same time. When the difference in the time of entering the store is greater than the time threshold or the time of leaving the store is greater than the time threshold, the two customer units cannot be determined to enter and leave the store at the same time. The motion trajectory parameters of the first customer unit and the motion trajectory parameters of the second customer unit are obtained according to the frame image analysis. The motion trajectory parameter difference is determined by the motion trajectory parameters of the first customer unit and the motion trajectory parameters of the second customer unit. The motion trajectory parameter difference is the motion trajectory overlap. When the motion trajectory overlap exceeds the trajectory overlap threshold, it is determined that the two customer units are in the same in-store motion trajectory. When the first time parameter is less than the time threshold, the second time parameter is less than the time threshold, and the first trajectory parameter is greater than the trajectory overlap threshold, that is, the two customer units enter the store at the same time, leave the store at the same time, and have the same in-store motion trajectory, the two customer units are the same customer batch, and the customer units in the first customer video data are batched to obtain at least one customer batch.
[0043] In one embodiment, Figure 5 As shown, determining the first behavior parameter corresponding to the customer batch according to the frame generation time and the frame image includes: Step S1102: determining the behavior action data of at least one customer unit in the customer batch according to the first frame image and the frame generation time; Step S1104: determining the behavior parameters corresponding to the client unit according to the behavior action data; Step S1106: Arrange at least one behavior parameter in the customer batch in descending order to determine a first behavior parameter corresponding to the customer batch.
[0044] In this embodiment, the customer batch in the first customer video data is analyzed by image recognition technology to determine the main car-buying customer units in a customer batch, and to determine the behavior parameters of the main car-buying customer units. Specifically, the behavior action data of all customer units in a customer batch are determined according to the first frame image and the frame generation time. The behavior action data include data corresponding to the in-store behavior actions such as whether to view the car, the length of time for viewing the car, the length of time for entering the car, the order of viewing the car, the length of time for negotiating with the salesperson, and the length of time for business negotiation. Store business operators analyze the behavior action data according to the historical customer flow data of the store background, determine the weight proportion corresponding to the behavior action data, and combine the business parameter adjustment algorithm to obtain the behavior parameters of all customer units in the same customer batch. The behavior parameters are the results of the weight proportion calculation of the behavior action data of all customer units in the same customer batch. The behavior parameters of all customer units in the same customer batch are arranged in descending order, and the first one in descending order is the first behavior parameter, that is, the behavior parameters of the main car-buying objects in the same customer batch are used as the behavior parameters of the customer batch.
[0045] In one embodiment, Figure 6 As shown, according to the first behavior parameter, determining the intention prediction data corresponding to the customer batch includes: Step S1122: Acquire a customer intention prediction database; Step S1124: determining the customer intention level and transaction probability parameter corresponding to the first behavior parameter according to the customer intention degree prediction database; Step S1126: determining the intention prediction parameters corresponding to the customer batch according to the customer intention level and the transaction probability parameter; Step S1128: Determine a batch visualization analysis report according to the frame generation time and the customer batch; Step S1130: Determine the intention prediction data according to the batch visualization analysis report and the intention prediction parameters.
[0046] In this embodiment, store business operators obtain a customer intention prediction database based on the analysis of the store's back-end historical passenger flow data, and determine a customer intention prediction algorithm based on the data in the customer intention prediction database, i.e., the store time, car viewing time, car entry time, whether there is car viewing behavior, car entry behavior, negotiation time with sales staff, customer's in-store action trajectory, car viewing order, whether to test drive, business negotiation key parameters and other business parameters. According to the customer intention prediction algorithm, the customer intention level and transaction probability parameters corresponding to the first behavior parameter are determined. And the customer batch visualization analysis report corresponding to the store is determined by frame generation, i.e., the store customer batch entry and exit situation, car viewing situation, etc. corresponding to each time period, and the batch visualization analysis report and intention prediction parameters are used as intention prediction data to predict the customer intention level corresponding to the store, improve the efficiency of the store's accurate classification management of customers, optimize resource allocation, and reduce personnel analysis costs.
[0047] Specifically, the customer intention level prediction algorithm is based on a series of business parameters, such as the length of time batch customer members spend in the store, whether they look at cars, how they enter the car, the length of time they look at the car, the length of time they enter the car, the length of time they negotiate with the salesperson, the customer's behavior trajectory in the store, the order in which they look at cars, whether they take a test ride, whether the car model corresponds to the mainstream target group, key parameters for business negotiations (price, competing products, customer focus), etc., to comprehensively predict the customer intention level (H, A, B, C, N) and the probability of a transaction. Among them, when the customer intention level is predicted to be H, it means that the customer's willingness to buy is extremely high, and the customer is a predicted consumer user; when the customer intention level is predicted to be N, it indicates that the customer's willingness to buy is extremely low, and the customer is determined to have low consumption intention. Store staff can reduce labor costs for the customer batch corresponding to this user; customer intention level A, customer intention level B, and customer intention level C are descending intention levels, which are related to in-store business parameters and in-store car data. It is necessary to adjust business negotiation related parameters (price, competitive products, customer focus) to update the user intention level. Customer batches corresponding to customer intention level A, customer intention level B, and customer intention level C are all potential consumer customers. Secondly, when a customer batch enters the store for the second time, the predicted data corresponding to the customer batch can be obtained from the historical passenger flow data of the store background for update and calculation.
[0048] In one embodiment, store business operators can set an algorithm update period to update the customer intention prediction algorithm and multi-person batch division algorithm based on the store backend customer flow data within a time period to improve the prediction efficiency and accuracy of the intention prediction method.
[0049] like Figure 7As shown, an embodiment of the present application provides an intention prediction device 900, including: an acquisition module 902, used to acquire original passenger flow video data; an image recognition module 904, used to determine the frame image and frame generation time corresponding to the original passenger flow video data; a customer recognition module 906, used to determine the first customer video data in the original passenger flow video data according to the frame image; a batch determination module 908, used to determine at least one customer batch corresponding to the first customer video data according to the frame generation time and the frame image; a behavior determination module 910, used to determine the first behavior parameter corresponding to the customer batch according to the frame generation time and the frame image; a prediction module 912, used to determine the intention prediction data corresponding to the customer batch according to the first behavior parameter.
[0050] According to the intention prediction device 900 provided in this embodiment, it mainly includes: an acquisition module 902, an image recognition module 904, a customer recognition module 906, a batch determination module 908, a behavior determination module 910 and a prediction module 912. The acquisition module 902 acquires the original video passenger flow data collected by the store camera; the image recognition module 904 determines the frame image and frame generation time in the original passenger flow video data; the customer recognition module 906 filters and cleans the original video passenger flow data, removes the video data of irrelevant personnel such as store clerks, cleaners, and deliverymen, and determines the first customer video data corresponding to the customer unit; the batch determination module 908 calculates the store video passenger flow data in batches to determine the customer batch in the first customer video data; the behavior determination module 910 analyzes the behavior actions, business actions, etc. of the customer units in the customer batch, and determines the behavior parameters of the main car purchase objects in the customer batch as the first behavior parameters corresponding to the customer batch; the prediction module 912 combines the prediction algorithm rules to determine the purchase intention degree of the main car purchase objects corresponding to each customer batch, and combines the batch visualization analysis report as the intention prediction data.
[0051] Optionally, a batch determination module 908 is used to determine a batch visualization analysis report according to the frame generation time and the customer batch; Optionally, an acquisition module 902 is used to acquire a customer intention degree prediction database; Optionally, the behavior determination module 910 is used to determine a time threshold and a trajectory overlap threshold.
[0052] like Figure 8 As shown, an embodiment of the present application provides an electronic device 1000, including a processor 1110, a memory 1109, and a program or instruction stored in the memory 1109 and executable on the processor 1110. When the program or instruction is executed by the processor 1110, the various processes of the embodiment of the above-mentioned intention prediction method are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0053] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the embodiment of the above-mentioned intention prediction method are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0054] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the embodiment of the above-mentioned intention prediction method, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
[0055] According to the intention prediction method, device, electronic device, readable storage medium and chip provided by the present invention, the customer flow data collected by video is filtered, invalid data of store clerks and other non-customer personnel are removed, and customer batches are determined. The customer behavior parameters of different batches are analyzed and the intention degree is predicted, which speeds up business follow-up and improves customer analysis and processing efficiency.
[0056] In the present invention, the terms "first", "second", and "third" are used for descriptive purposes only and should not be understood as indicating or implying relative importance; the term "plurality" refers to two or more, unless otherwise clearly defined. The terms "installed", "connected", "connected", "fixed", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; "connected" can be a direct connection or an indirect connection through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0057] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by terms such as “upper”, “lower”, “left”, “right”, “front” and “back” are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or unit referred to must have a specific direction, be constructed and operated in a specific orientation, and therefore, cannot be understood as a limitation on the present invention.
[0058] In the description of this specification, the description of the terms "one embodiment", "some embodiments", "specific embodiments", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0059] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for predicting intention, characterized in that: include: Obtain original passenger flow video data; Determine a frame image and a frame generation time corresponding to the original passenger flow video data; Determine first customer video data in the original customer flow video data according to the frame image; determining at least one customer batch corresponding to the first customer video data according to the frame generation time and the frame image; determining a first behavior parameter corresponding to the customer batch according to the frame generation time and the frame image; Intention prediction data corresponding to the customer batch is determined according to the first behavior parameter.
2. The intention prediction method according to claim 1, characterized in that: The determining, according to the frame image, first customer video data in the original customer flow video data comprises: Determine employee image identification and invalid personnel image identification; Determine a non-client unit identification according to the employee image identification and the invalid personnel image identification; Filtering the non-customer unit identifier in the frame image to obtain a first frame image; First client video data is determined according to the first frame image.
3. The intention prediction method according to claim 2, characterized in that: The determining, according to the frame generation time and the frame image, at least one customer batch corresponding to the first customer video data comprises: determining at least one client unit in the first client video data; Determine the store entry time and store exit time of the customer unit according to the frame generation time; Determine the motion trajectory parameters of the client unit according to the first frame image; At least one customer batch is determined according to the store entry time, the store exit time and the movement trajectory parameters of the customer unit.
4. The intention prediction method according to claim 3, characterized in that: Determining at least one customer batch according to the store entry time, the store exit time and the movement trajectory parameters of the customer unit includes: Determine the time threshold and trajectory overlap threshold; Determine the difference between the store entry time of the first customer unit and the store entry time of the second customer unit as a first time parameter; Determine the difference between the check-out time of the first customer unit and the check-out time of the second customer unit as a second time parameter; Determine the difference between the motion trajectory parameter of the first customer unit and the motion trajectory parameter of the second customer unit as the first trajectory parameter; When the first time parameter is less than the time threshold, the second time parameter is less than the time threshold, and the first trajectory parameter is greater than a trajectory overlap threshold, it is determined that the first customer unit and the second customer unit are the same customer batch.
5. The intention prediction method according to claim 4, characterized in that: Determining the first behavior parameter corresponding to the customer batch according to the frame generation time and the frame image includes: Determine the behavior action data of at least one of the customer units in the customer batch according to the first frame image and the frame generation time; Determining a behavior parameter corresponding to the client unit according to the behavior action data; Arrange at least one of the behavior parameters in the customer batch in descending order to determine a first behavior parameter corresponding to the customer batch.
6. The intention prediction method according to any one of claims 1 to 5, characterized in that: The step of determining the intention prediction data corresponding to the customer batch according to the first behavior parameter includes: Obtain customer intention prediction database; Determine the customer intention level and transaction probability parameter corresponding to the first behavior parameter according to the customer intention degree prediction database; Determining an intention prediction parameter corresponding to the customer batch according to the customer intention level and the transaction probability parameter; Determine a batch visualization analysis report according to the frame generation time and the customer batch; The intention prediction data is determined according to the batch visualization analysis report and the intention prediction parameters.
7. An intention prediction device, characterized in that: include: An acquisition module is used to acquire original passenger flow video data; An image recognition module, used to determine a frame image and a frame generation time corresponding to the original passenger flow video data; A customer identification module, configured to determine first customer video data in the original customer flow video data according to the frame image; A batch determination module, configured to determine at least one customer batch corresponding to the first customer video data according to the frame generation time and the frame image; A behavior determination module, which determines a first behavior parameter corresponding to the customer batch according to the frame generation time and the frame image; A prediction module is used to determine the intention prediction data corresponding to the customer batch according to the first behavior parameter.
8. An electronic device, characterized in that: It comprises a processor, a memory and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the intention prediction method as described in any one of claims 1 to 6.
9. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by a processor, the steps of the intention prediction method according to any one of claims 1 to 6 are implemented.
10. A chip, characterized in that: The chip includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the steps of the intention prediction method as described in any one of claims 1 to 6.