Monitoring and early warning method and system for enterprise cashier system

Abstract

The invention discloses a monitoring and early warning method and system for an enterprise cashier system, and relates to the technical field of data processing, and the method comprises the steps: obtaining a goods taking timestamp, goods taking category information, a store leaving timestamp and store shopping track information through an uploaded in-store video stream; carrying out portrait combination evaluation based on the timestamp and the shopping track information; a goods category combination scheme meeting the portrait combination scheme is extracted; counting a first receivable item list based on a discount constraint condition, and matching a first collection theoretical time list; counting a second receivable item list, and matching with a second collection theoretical time list; order analysis is carried out, and abnormal order identification information is generated and sent to the management end for early warning. The technical problem that the system has supervision vulnerabilities due to insufficient related elements of a current cashier system monitoring scheme of a retail enterprise is solved, and the technical effects of improving the supervision quality of the cashier system, guaranteeing the healthy development of the enterprise and protecting the rights and interests of consumers are achieved.

Description

Technical Field

[0001] This application relates to the field of data processing technologies, and particularly to a monitoring and early warning method and system for an enterprise cash register system. Background Art

[0002] Driven by the digital wave, traditional enterprise cash register systems are facing unprecedented challenges and opportunities. Retail enterprises are faced with a complex sales environment brought about by various promotional activities at different sales points. Although the current enterprise cash register system can trace and verify whether the accounts are incorrect through order information, some malicious personnel take advantage of system loopholes to conduct illegal operations. They sell goods without going through the cash register equipment and then earn price differences by supplementing orders that meet the discount conditions. The order information of such behavior is correct. Therefore, how to monitor such behaviors that seriously damage the rights and interests of consumers and the healthy development of enterprises and give early warnings has become an urgent problem for current retail enterprises.

[0003] Therefore, at the current stage, the enterprise cash register system has a technical problem that the current monitoring scheme of the retail enterprise cash register system has insufficient involved elements, resulting in regulatory loopholes in the system. Summary of the Invention

[0004] By providing a monitoring and early warning method and system for an enterprise cash register system, this application uses technical means such as image acquisition, portrait combination evaluation, and goods category extraction to solve the technical problem that the current monitoring scheme of the retail enterprise cash register system has insufficient involved elements, resulting in regulatory loopholes in the system, and achieves the technical effects of improving the regulatory quality of the cash register system, ensuring the healthy development of enterprises, and protecting the rights and interests of consumers.

[0005] The present application provides a monitoring and early warning method for an enterprise cash register system. The method includes: obtaining a first pick-up timestamp, first pick-up category information, first store exit timestamp, and first shopping trajectory information of a first person image in a preset time zone from the in-store video stream uploaded by an image acquisition device, until the Nth pick-up timestamp, Nth pick-up category information, Nth store exit timestamp, and Nth shopping trajectory information of the Nth person image; performing a person image combination evaluation based on the first store exit timestamp until the Nth store exit timestamp, and the first shopping trajectory information until the Nth shopping trajectory information, to obtain a person image combination plan; extracting a goods category combination plan that meets the person image combination plan from the first pick-up category information until the Nth pick-up category information; traversing the goods category combination plan, and based on discount constraint conditions, counting a first list of receivables, and matching a first theoretical collection time list for the first receivables information based on the first pick-up timestamp until the Nth pick-up timestamp, and the first store exit timestamp until the Nth store exit timestamp; based on the first pick-up category information until the Nth pick-up category information, and based on the discount constraint conditions, counting a second list of receivables, and matching a second theoretical collection time list for the second receivables information based on the first pick-up timestamp until the Nth pick-up timestamp, and the first store exit timestamp until the Nth store exit timestamp; performing order analysis based on the first list of receivables and the first theoretical collection time list, and the second list of receivables and the second theoretical collection time list, to generate abnormal order identification information and send it to the management terminal for early warning.

[0006] In a possible implementation manner, to obtain the first pick-up timestamp, first pick-up category information, first store exit timestamp, and first shopping trajectory information of the first person image in the preset time zone, the following processing is performed: performing semantic segmentation on the in-store video stream to obtain a first person image video identification sequence; performing pick-up analysis based on the first person image video identification sequence to obtain the first pick-up timestamp and the first pick-up category information; performing motion position time series extraction based on the first person image video identification sequence to obtain the first shopping trajectory information; performing end time extraction based on the first person image video identification sequence, and setting it as the first store exit timestamp.

[0007] In a possible implementation manner, to perform pick-up analysis based on the first person image video identification sequence to obtain the first pick-up timestamp and the first pick-up category information, the following processing is performed: when a first item is taken out by the first person image and not put back into the goods location, recording a first item pick-up timestamp, first item category information, and first item pick-up quantity information; adding the first item pick-up timestamp to the first pick-up timestamp; adding the first item category information and the first item pick-up quantity information to the first pick-up category information.

[0008] In a possible implementation, based on the first store exit timestamp up to the Nth store exit timestamp, and the first shopping trajectory information up to the Nth shopping trajectory information, a portrait combination evaluation is performed to obtain a portrait combination plan, and the following processing is also executed: pairwise trajectory deviation analysis is performed according to the first shopping trajectory information up to the Nth shopping trajectory information to generate a set of shopping trajectory deviation coefficients; according to the set of shopping trajectory deviation coefficients, clustering analysis is performed on the first portrait up to the Nth portrait to generate multi-cluster portrait information; the multi-cluster portrait information is traversed, and k-item timestamp verification is performed based on the first store exit timestamp up to the Nth store exit timestamp to generate the portrait combination plan.

[0009] In a possible implementation, pairwise trajectory deviation analysis is performed according to the first shopping trajectory information up to the Nth shopping trajectory information to generate a set of shopping trajectory deviation coefficients, and the following processing is also executed: according to the three-dimensional virtual space, the ith shopping trajectory information is located to generate a first trajectory curve, and the jth shopping trajectory information is located to generate a second trajectory curve, where i and j are any two different trajectories of the first shopping trajectory information up to the Nth shopping trajectory information, N≥i≥1, N≥j≥1, i≠j; the starting positions of the first trajectory curve and the second trajectory curve are connected, and the ending positions of the first trajectory curve and the second trajectory curve are connected to obtain the enclosed graphic area information, which is set as the first shopping trajectory deviation coefficient and added to the set of shopping trajectory deviation coefficients.

[0010] In a possible implementation, the multi-cluster portrait information is traversed, and k-item timestamp verification is performed based on the first store exit timestamp up to the Nth store exit timestamp to generate the portrait combination plan, and the following processing is also executed: the first cluster of portraits in the multi-cluster portrait information is obtained, where the first cluster of portraits has a portrait quantity M; an integer k value is set, and the initial value is set to 2, and k-item enumeration combination is performed on the first cluster of portraits to generate a first set of k-item portrait combination plans; based on the first store exit timestamp up to the Nth store exit timestamp, the first set of k-item portrait combination plans is traversed to perform pairwise portrait timestamp deviation verification: when any timestamp deviation is greater than or equal to the timestamp deviation threshold, the corresponding first set of k-item portrait combination plans is eliminated; when each timestamp deviation is less than the timestamp deviation threshold, the corresponding first set of k-item portrait combination plans is added to the portrait combination plan; when the traversal of the first set of k-item portrait combination plans is completed, it is judged whether k is greater than or equal to M; if k is less than M, k is incremented by one, and the k-item timestamp verification is looped; if k is greater than or equal to M, the portrait combination plan is output.

[0011] In a possible implementation manner, order analysis is performed based on the first list of receivables and the first list of theoretical collection times, as well as the second list of receivables and the second list of theoretical collection times to generate abnormal order identification information, and the following processing is also performed: Through the cash register device, upload order information in the preset time zone is obtained, where the upload order information includes the total amount of recorded funds and the number of actual payment accounts; according to the first list of receivables, the total amount of the first receivables and the number of the first payment accounts are obtained, and according to the second list of receivables, the total amount of the second receivables and the number of the second payment accounts are obtained; according to the total amount of the first receivables and the total amount of the second receivables, a theoretical collection interval is constructed; according to the number of the first payment accounts and the number of the second payment accounts, a theoretical account number interval is constructed; when the total amount of the recorded funds does not belong to the theoretical collection interval, and / or the number of the actual payment accounts does not belong to the theoretical account number interval, if the upload order information does not meet the first list of receivables and the first list of theoretical collection times, and does not meet the second list of receivables and the second list of theoretical collection times, the abnormal order identification information is generated.

[0012] The present application also provides a monitoring and warning system for an enterprise cash register system, including: A video stream information obtaining module, which is used to obtain the first pick-up timestamp, the first pick-up category information, the first store exit timestamp, and the first shopping trajectory information of the first person image in the preset time zone from the in-store video stream uploaded by the image acquisition device, until the Nth pick-up timestamp, the Nth pick-up category information, the Nth store exit timestamp, and the Nth shopping trajectory information of the Nth person image; A person image combination evaluation module, which is used to perform person image combination evaluation based on the first store exit timestamp until the Nth store exit timestamp, and the first shopping trajectory information until the Nth shopping trajectory information to obtain a person image combination plan; A goods category combination plan extraction module, which is used to extract a goods category combination plan that meets the person image combination plan from the first pick-up category information until the Nth pick-up category information; A first list of receivables statistics module, which is used to traverse the goods category combination plan, and based on the discount constraint conditions, count the first list of receivables, and match the first list of theoretical collection times for the first receivables information based on the first pick-up timestamp until the Nth pick-up timestamp, and the first store exit timestamp until the Nth store exit timestamp; The second accounts receivable list statistics module is used to count the second accounts receivable list based on the first pick-up category information to the Nth pick-up category information and based on the discount constraint condition, and match the second collection theoretical time list for the second accounts receivable information based on the first pick-up timestamp to the Nth pick-up timestamp and the first store exit timestamp to the Nth store exit timestamp; The abnormal order identification information generation module is used to perform order analysis according to the first accounts receivable list and the first collection theoretical time list, and the second accounts receivable list and the second collection theoretical time list, and generate abnormal order identification information to be sent to the management terminal for warning.

[0013] A monitoring and warning method and system for an enterprise cash register system proposed by this application obtain a pick-up timestamp, pick-up category information, store exit timestamp, and shopping trajectory information through an in-store video stream uploaded by an image acquisition device; perform portrait combination evaluation based on the timestamp and shopping trajectory information; extract a goods category combination plan that meets the portrait combination plan; count the first accounts receivable list based on the discount constraint condition and match the first collection theoretical time list; count the second accounts receivable list and match the second collection theoretical time list; perform order analysis, and generate abnormal order identification information to be sent to the management terminal for warning. It solves the technical problem that the current cash register system monitoring solution of retail enterprises has regulatory loopholes due to insufficient involved elements, and achieves the technical effects of improving the supervision quality of the cash register system, ensuring the healthy development of the enterprise, and protecting the rights and interests of consumers. Brief Description of the Drawings

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations in the front or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0015] Figure 1 It is a schematic flowchart of a monitoring and warning method for an enterprise cash register system provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of a monitoring and warning system for an enterprise cash register system provided by an embodiment of the present application.

[0016] Description of the drawing reference numerals: Video stream information acquisition module 10, Portrait combination evaluation module 20, Goods category combination scheme extraction module 30, First receivables list statistics module 40, Second receivables list statistics module 50, Abnormal order identification information generation module 60. Detailed implementation manners

[0017] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other objects, features and advantages of the present application more obvious and understandable, the following specifically gives the detailed implementation manners of the present application.

[0018] In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings. The described embodiments should not be regarded as limitations of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0019] In the following description, "some embodiments" are involved, which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first\second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0020] The embodiment of the present application provides a monitoring and early warning method for an enterprise cash register system, as Figure 1 shown, the method includes: Step S100, obtain the first pick-up timestamp, the first pick-up category information, the first store exit timestamp, and the first in-store shopping trajectory information of the first person image in the preset time zone from the in-store video stream uploaded by the image acquisition device, until the Nth pick-up timestamp, the Nth pick-up category information, the Nth store exit timestamp, and the Nth in-store shopping trajectory information of the Nth person image. The in-store video stream refers to the real-time video data captured by the surveillance cameras installed inside the store. By obtaining the in-store video stream uploaded by the image acquisition device, detailed behavioral information of customers in the store can be obtained, including pick-up timestamp, pick-up category information, store exit timestamp, and in-store shopping trajectory information. Specifically, by installing high-definition cameras and intelligent recognition software, the in-store goods transaction process is monitored in real time to ensure that each piece of goods is scanned and recorded normally by the cashier device. The first person image refers to the image information of the first customer identified in the video stream. The first pick-up timestamp refers to the specific time when this customer picks up goods in the store, usually recorded in the time format of year, month, day, hour, minute, and second; the first pick-up category information refers to the category of the goods selected by this customer when picking up goods. For example, a customer may pick up a piece of clothing from the clothing area or a bag of snacks from the food area; the first store exit timestamp is the time when this customer leaves the store; the first in-store shopping trajectory information refers to the movement path or trajectory of this customer in the store, tracking the walking route of the customer in the store to understand which areas they have browsed, which goods they are interested in, and which areas in the store may be more popular; N refers to the number of customers in the store within the preset time zone, and each customer corresponds to obtaining pick-up timestamp, pick-up category information, store exit timestamp, and in-store shopping trajectory information.

[0021] In a possible implementation, step S100 further includes step S110 of performing semantic segmentation on the in-store video stream to obtain a first sequence of portrait video identifiers. Semantic segmentation can divide each pixel point in an image or video into different semantic categories. Specifically, for the in-store video stream, semantic segmentation can distinguish different objects in the video, such as people, shelves, the ground, etc. Through semantic segmentation, relevant information about people can be extracted from the in-store video stream, which may include the position, shape, movement trajectory, etc. of the people. The first sequence of portrait video identifiers is the organization and identification of this people information, including a series of frames or segments related to people, and each frame or segment contains the specific position and state of the people in the video. It also includes step S120 of performing pick-up analysis based on the first sequence of portrait video identifiers to obtain the first pick-up timestamp and the first pick-up category information. It also includes step S130 of performing extraction of the movement position time series based on the first sequence of portrait video identifiers to obtain the first store-shopping trajectory information. The extraction of the movement position time series refers to further extracting the movement trajectory information of the customer from the first sequence of portrait video identifiers, that is, tracking and recording the position of the people in each frame or segment to form a series of time series data, and obtaining the complete movement trajectory of the customer in the store, that is, the first store-shopping trajectory information, including the starting position, movement path, staying area, and final leaving position of the customer in the store, etc. It also includes step S140 of performing extraction of the end time based on the first sequence of portrait video identifiers and setting it as the first store-exiting timestamp. The extraction of the end time refers to identifying the specific time when the customer leaves the store from the first sequence of portrait video identifiers. Specifically, analyzing the movement trajectory of the customer to determine when they leave the monitoring range of the store. In actual operation, some rules or thresholds may need to be set to determine whether the customer has left the store, such as the customer walking out of the store door, not appearing in the monitoring range for a long time, etc. Once the end time of the customer, that is, the time when they leave the store, is determined, it is recorded as a specific timestamp, called the first store-exiting timestamp.

[0022] In a possible implementation, step S120 further includes step S121. When the first item is taken out by the first person and not put back into the storage location, record the pickup timestamp of the first item, the category information of the first item, and the quantity information of the first item picked up. Through the video, it can be seen whether the customer has taken a certain item. For example, if the initial quantity is 0 and one item is taken out, the quantity increases by one; if one item is put back on the store shelf, the quantity decreases by one. Check how many items are finally taken and record the corresponding pickup timestamp, item category information, and quantity information of the item picked up. It further includes step S122, adding the pickup timestamp of the first item to the first pickup timestamp. Integrate the pickup timestamp of the first item into the first pickup timestamp. It further includes step S123, adding the category information of the first item and the quantity information of the first item picked up to the first pickup category information. Integrate the category information of the first item and the quantity information of the first item picked up into the corresponding first pickup category information.

[0023] Step S200: Based on the first store exit timestamp to the Nth store exit timestamp, and the first shopping trajectory information to the Nth shopping trajectory information, conduct a portrait combination evaluation to obtain a portrait combination plan. Conduct a portrait combination evaluation based on the store exit timestamp and the shopping trajectory information to obtain a portrait combination plan. Specifically, by integrating and analyzing the shopping behavior data of in-store customers, a more comprehensive customer behavior portrait can be formed, which helps to identify different types of customer groups. Finally, by integrating the analysis results of the timestamp and the shopping trajectory, the store can form a comprehensive portrait combination plan, including formulating marketing strategies based on the behavior portrait, such as providing different promotional activities or recommending different products for different customer groups. It also includes that through in-store trajectory analysis, the store can adjust the product display and layout and optimize the customer shopping path at the same time.

[0024] In a possible implementation, step S200 further includes step S210 of performing pairwise trajectory deviation analysis based on the first store-shopping trajectory information up to the Nth store-shopping trajectory information to generate a set of store-shopping trajectory deviation coefficients. Comparing and analyzing the store-shopping trajectories of multiple customers to quantify the deviation degree between different trajectories. Specifically, pairwise trajectory deviation analysis refers to comparing the store-shopping trajectories of any two customers to evaluate their similarity or difference, calculating a deviation coefficient for the trajectories of each customer pair, quantifying the deviation degree between the two trajectories, and the deviation coefficients between all customer pairs form a set, that is, the set of store-shopping trajectory deviation coefficients. It further includes step S220 of performing clustering analysis on the first portrait up to the Nth portrait according to the set of store-shopping trajectory deviation coefficients to generate multi-cluster portrait information. Clustering analysis groups the first portrait to the Nth portrait according to the set of store-shopping trajectory deviation coefficients, clustering customers with similar store-shopping trajectories together to form different clusters, each cluster representing a group of customers with similar store-shopping trajectories, forming multi-cluster portrait information, and the portraits in each cluster have a certain common shopping behavior pattern or preference. It further includes step S230 of traversing the multi-cluster portrait information and performing k-item timestamp verification based on the first store-exit timestamp up to the Nth store-exit timestamp to generate the portrait combination scheme. k-item timestamp refers to performing a certain form of comparison or analysis on timestamps to determine whether there is a certain pattern or rule. Specifically, traversing the multi-cluster portrait information, searching for close or overlapping store-exit timestamps to determine which customers may leave the store within the same time period, and based on the timestamp verification result, identifying the customer combinations that may leave the store within the same time period, and the portrait combination scheme is generated according to these identified customer combinations, and each scheme may include a specific group of customers and their store-exit time information.

[0025] In a possible implementation, step S210 further includes step S211 of positioning the i-th store-shopping trajectory information according to the three-dimensional virtual space to generate a first trajectory curve, and positioning the j-th store-shopping trajectory information to generate a second trajectory curve, where i and j are any two different trajectories from the first store-shopping trajectory information to the N-th store-shopping trajectory information, N≥i≥1, N≥j≥1, and i≠j. The three-dimensional virtual space is a computer-generated three-dimensional environment for simulating or displaying real-world scenes or objects, representing a virtual model of a store, which contains three-dimensional representations of elements such as the internal layout of the store, shelves, and aisles. Mapping the store-shopping trajectory information into the three-dimensional virtual space enables each trajectory to find a corresponding position and path in the virtual environment, generating corresponding trajectory curves. The first trajectory curve corresponds to the i-th store-shopping trajectory information, and the second trajectory curve corresponds to the j-th store-shopping trajectory information, reflecting the store-shopping behavior patterns of different customers. Here, i and j represent different store-shopping trajectory information, and i and j are any numbers between 1 and N, but they cannot be equal to ensure that we are comparing two different trajectories. It further includes step S212 of connecting the starting positions of the first trajectory curve and the second trajectory curve, connecting the ending positions of the first trajectory curve and the second trajectory curve, obtaining the enclosed figure area information, denoted as the first store-shopping trajectory deviation coefficient, and adding it to the set of store-shopping trajectory deviation coefficients. Find the starting positions of the first trajectory curve and the second trajectory curve and connect them, then connect the ending positions of these two curves. The two trajectory curves and the two connecting lines together form an enclosed figure. Calculate the area of this enclosed figure, which reflects the spatial deviation between the two trajectory curves, that is, their relative positions and shape differences in the three-dimensional virtual space. Set it as the first store-shopping trajectory deviation coefficient, which quantifies the deviation degree between the first trajectory curve and the second trajectory curve. The larger the coefficient value, the greater the deviation between the two trajectories; the smaller the coefficient value, the smaller the deviation. Finally, add the first store-shopping trajectory deviation coefficient to the set of store-shopping trajectory deviation coefficients, which contains all the calculated trajectory deviation coefficients for subsequent clustering analysis or other data analysis tasks.

[0026] In a possible implementation, step S230 further includes step S231 of obtaining the first cluster of portrait information among the multiple clusters of portrait information, where the first cluster of portrait information has a portrait number M. Select one cluster from among the numerous clustering results, that is, the first cluster of portraits, and determine the number of portraits M included in this cluster. It also includes step S232 of setting an integer k value, with the initial value set to 2, and performing k-item enumeration combinations on the first cluster of portraits to generate a first set of k-item portrait combination schemes. The k value represents the number of portraits to be selected from the first cluster of portraits. Under this initial setting, what is concerned is the combination of all pairs of portraits. Perform enumeration combinations on the first cluster of portraits, that is, list all possible combination situations, and find all possible combinations of k portraits in the first cluster of portraits. For example, if the first cluster of portraits contains 5 portraits (A, B, C, D, E) and k = 2, it is necessary to find all pairs of portraits, that is, (A,B), (A,C), (A,D), (A,E), (B,C), (B,D), (B,E), (C,D), (C,E), and (D,E), and summarize all the found k-item combinations into a set, that is, the first set of k-item portrait combination schemes. It also includes step S233 of traversing the first set of k-item portrait combination schemes to perform two-person portrait timestamp deviation verification based on the first checkout timestamp up to the Nth checkout timestamp. Traverse each combination of two portraits in the first set of k-item portrait combination schemes, compare their checkout timestamps, and quantify the difference in the time when these two people leave the store by calculating the difference between the two timestamps and comparing whether they are within the same time period.

[0027] In a possible implementation, step S233 further includes step S2331 of eliminating the corresponding first set of k-item portrait combination schemes when any one timestamp deviation is greater than or equal to the timestamp deviation threshold. The timestamp deviation threshold is a preset value used to determine whether the difference in the time when two portraits leave the store has reached an unacceptable level, and this threshold can be set according to actual situations. For example, based on the business hours of the store, the average shopping duration of customers, etc. It also includes step S2332 of adding the corresponding first set of k-item portrait combination schemes to the portrait combination scheme when each timestamp deviation is less than the timestamp deviation threshold. It also includes step S2333 of, when the traversal of the first set of k-item portrait combination schemes is completed, determining whether k is greater than or equal to M. If k is less than M, increment k by one and loop for k-item timestamp verification; if k is greater than or equal to M, output the portrait combination scheme.

[0028] Step S300: Extract the goods category combination plan that meets the portrait combination plan from the first pick-up category information to the Nth pick-up category information. The first pick-up category information to the Nth pick-up category information of customers reflects different customers' shopping choices in the store. Specifically, classify and count the pick-up category information of customers to understand which goods categories are hot-selling, which are unpopular, as well as the relevance and purchase combinations between categories. Match this category information with the previously obtained portrait combination plan. Through comparison, identify which goods categories are more favored by a certain customer group, or which category combinations have a higher purchase rate among specific customer groups. Based on the above analysis results, extract the goods category combination plan that meets the portrait combination plan, which may include recommendations for popular category combinations for high-frequency buyers, recommendations for new products or promotional categories for occasional buyers, personalized category combination suggestions for specific customer groups, etc.

[0029] Step S400: Traverse the goods category combination plan, based on the discount constraint conditions, count the first list of receivables, and match the first list of theoretical collection times to the first receivables information based on the first pick-up timestamp to the Nth pick-up timestamp, and the first store exit timestamp to the Nth store exit timestamp. The discount constraint conditions refer to the discount policies or constraints that need to be considered when counting receivables, which may include quantity discounts, time discounts, member discounts, etc. For example, if a certain goods category combination plan can enjoy a discount when purchased within a specific time period, then the price needs to be adjusted according to this discount condition when counting receivables. Specifically, traverse the extracted goods category combination plan that meets the portrait combination plan, and calculate the receivables corresponding to each combination plan according to the traversed goods category combination plan and the discount constraint conditions, and form a list of receivables, including information such as the product details, quantity, unit price, and discounted amount of each combination plan. Then, based on the pick-up timestamp and store exit timestamp of the customer, match a theoretical collection time for each receivables information, that is, the first list of theoretical collection times. For example, if a certain customer picks up and exits the store at a specific time, theoretically this payment should be collected shortly after the customer exits the store. By matching the timestamps, a list of theoretical collection times can be established to help the store better arrange the collection plan and track the collection progress. Through this process, the store can not only more accurately grasp the receivables of each goods category combination plan, but also reasonably arrange the collection time according to the customer's shopping time, improve the collection efficiency, and reduce the bad debt risk.

[0030] Step S500: Based on the first pick-up category information up to the Nth pick-up category information, and based on the discount constraint conditions, count the second list of receivables. Based on the first pick-up timestamp up to the Nth pick-up timestamp, and the first store exit timestamp up to the Nth store exit timestamp, match a second list of theoretical collection times for the second receivables information. The actual pick-up category information of customers (from the pick-up categories of the first customer to the Nth customer) and the discount constraint conditions of the store are used to count the receivables. Consider what goods the customers have purchased and calculate the receivable amount according to whether these goods meet the discount conditions (such as quantity discounts, membership discounts, etc.). The second list of receivables refers to the detailed list of payments for goods sold individually and not according to the goods combination plan. For each entry in the second list of receivables, a theoretical collection time, that is, the second list of theoretical collection times, is matched according to the pick-up timestamp and store exit timestamp of the customer.

[0031] Step S600: Conduct order analysis based on the first list of receivables and the first list of theoretical collection times, as well as the second list of receivables and the second list of theoretical collection times, and generate abnormal order identification information to send to the management terminal for warning. Order analysis refers to conducting in-depth reviews and evaluations of all orders in the store in order to identify potential abnormalities or risks. Specifically, the first list of receivables and the second list of receivables will serve as the main sources of financial data, and the first list of theoretical collection times and the second list of theoretical collection times provide the theoretical time references for order receivables. By comparing the differences between the actual receivables and the theoretical receivables, and analyzing the deviation of the collection time, orders that may have problems can be initially screened out. Once an abnormal order is identified, the system generates abnormal order identification information, including the order number, customer information, abnormal type (such as overdue payment, amount mismatch, etc.), abnormal details, etc. After the abnormal order identification information is generated, the system will send this abnormal information to the management terminal of the store to trigger the warning mechanism, such as the store's back-end management system, the mobile devices of the person in charge, etc. Through the warning, the situation of the abnormal order can be quickly understood, and corresponding countermeasures can be taken, such as investigating the cause of the abnormality, contacting the customer for payment collection, adjusting the sales strategy, etc., to reduce the bad debt risk and improve the financial health of the store.

[0032] In a possible implementation, step S600 further includes step S610 of obtaining, through a cash register device, the uploaded order information for the preset time zone, where the uploaded order information includes the total amount of recorded payments and the number of actual payment accounts. Obtaining the uploaded order information for the preset time zone through a cash register device means collecting and uploading order-related information, including the total amount of recorded payments and the number of actual payment accounts, through a cash register system or related devices during a specific time period (i.e., the preset time zone). Specifically, the total amount of recorded payments refers to the total recorded amount of all orders within the preset time zone; the number of actual payment accounts refers to the number of accounts that use different payment methods for payment during this time period, which may be personal bank accounts, credit card accounts, third-party payment accounts (such as Alipay, WeChat Pay, etc.). It further includes step S620 of obtaining the total amount of the first receivables and the number of the first payment accounts according to the first list of receivables, and obtaining the total amount of the second receivables and the number of the second payment accounts according to the second list of receivables. The total amount of the first receivables and the total amount of the second receivables represent the total amount of receivables under different conditions or during different time periods. For example, the total amount of the first receivables may be calculated based on the sales data of the previous week, while the total amount of the second receivables may be calculated based on the sales data of the previous month. It further includes step S630 of constructing a theoretical collection range according to the total amount of the first receivables and the total amount of the second receivables. Using the two total amounts of receivables (the total amount of the first receivables and the total amount of the second receivables) to set an expected or possible collection range, that is, the theoretical collection range. It further includes step S640 of constructing a theoretical account number range according to the number of the first payment accounts and the number of the second payment accounts. Using two different data points of the number of payment accounts (i.e., the number of the first payment accounts and the number of the second payment accounts) to set an expected or possible account number range, that is, the theoretical account number range. It further includes step S650 of generating the abnormal order identification information when the total amount of the recorded payments does not belong to the theoretical collection range, and / or the number of actual payment accounts does not belong to the theoretical account number range, and if the uploaded order information does not meet the first list of receivables and the first theoretical collection time list, and does not meet the second list of receivables and the second theoretical collection time list. If there is a deviation between the actual collection amount, the number of actual payment accounts used and the expected range, and the uploaded order information does not meet the first collection time list and the second collection time list, that is, there is no identical order information, then an abnormal identification is made, which may be caused by abnormal sales activities, system errors, human operation mistakes, etc. The abnormal order identification information is generated. By identifying abnormal orders, merchants can quickly identify and process orders that may have problems.

[0033] In the foregoing, with reference to Figure 1A monitoring and early warning method for an enterprise cashier system according to an embodiment of the present invention is described in detail. Next, a monitoring and early warning system for an enterprise cashier system according to an embodiment of the present invention will be described with reference to Figure 2 Describe a monitoring and early warning system for an enterprise cashier system according to an embodiment of the present invention.

[0034] A monitoring and early warning system for an enterprise cashier system according to an embodiment of the present invention is used to solve the technical problem that the current cashier system monitoring scheme of retail enterprises has insufficient involved elements, resulting in regulatory loopholes in the system, and achieves the technical effects of improving the supervision quality of the cashier system, ensuring the healthy development of enterprises, and protecting the rights and interests of consumers. A monitoring and early warning system for an enterprise cashier system includes: a video stream information acquisition module 10, a portrait combination evaluation module 20, a goods category combination scheme extraction module 30, a first receivables list statistics module 40, a second receivables list statistics module 50, and an abnormal order identification information generation module 60.

[0035] The video stream information acquisition module 10, the video stream information acquisition module 10 is used to obtain the first pick-up timestamp, the first pick-up category information, the first store exit timestamp, and the first shopping trajectory information of the first portrait in a preset time zone through the in-store video stream uploaded by the image acquisition device, until the Nth pick-up timestamp, the Nth pick-up category information, the Nth store exit timestamp, and the Nth shopping trajectory information of the Nth portrait; The portrait combination evaluation module 20, the portrait combination evaluation module 20 is used to perform portrait combination evaluation based on the first store exit timestamp until the Nth store exit timestamp, and the first shopping trajectory information until the Nth shopping trajectory information, and obtain a portrait combination scheme; The goods category combination scheme extraction module 30, the goods category combination scheme extraction module 30 is used to extract the goods category combination scheme that meets the portrait combination scheme from the first pick-up category information until the Nth pick-up category information; The first receivables list statistics module 40, the first receivables list statistics module 40 is used to traverse the goods category combination scheme, based on the discount constraint conditions, to count the first receivables list, and based on the first pick-up timestamp until the Nth pick-up timestamp, and the first store exit timestamp until the Nth store exit timestamp, to match the first collection theory time list for the first receivables information; The second receivables list statistics module 50, the second receivables list statistics module 50 is used to based on the first pick-up category information until the Nth pick-up category information, based on the discount constraint conditions, to count the second receivables list, and based on the first pick-up timestamp until the Nth pick-up timestamp, and the first store exit timestamp until the Nth store exit timestamp, to match the second collection theory time list for the second receivables information; An abnormal order identification information generation module 60, which is configured to perform order analysis based on the first receivables list and the first theoretical collection time list, as well as the second receivables list and the second theoretical collection time list, and generate abnormal order identification information to be sent to the management terminal for warning.

[0036] Next, the specific configuration of the video stream information acquisition module 10 will be described in detail. The video stream information acquisition module 10 further includes: performing semantic segmentation on the in-store video stream to obtain a first sequence of portrait video identifiers; performing pick-up analysis based on the first sequence of portrait video identifiers to obtain the first pick-up timestamp and the first pick-up category information; performing extraction of the temporal sequence of movement positions based on the first sequence of portrait video identifiers to obtain the first store-shopping trajectory information; and performing extraction of the end time moment based on the first sequence of portrait video identifiers and setting it as the first store-exit timestamp.

[0037] Next, the specific configuration of the video stream information acquisition module 10 will be further described in detail. The video stream information acquisition module 10 may further include: when the first item is taken out by the first portrait and not put back into the storage location, recording the first item pick-up timestamp, the first item category information, and the first item pick-up quantity information; adding the first item pick-up timestamp to the first pick-up timestamp; and adding the first item category information and the first item pick-up quantity information to the first pick-up category information.

[0038] Next, the specific configuration of the portrait combination evaluation module 20 will be described in detail. The portrait combination evaluation module 20 may further include: performing pairwise trajectory deviation analysis based on the first store-shopping trajectory information to the Nth store-shopping trajectory information to generate a set of store-shopping trajectory deviation coefficients; performing clustering analysis on the first portrait to the Nth portrait based on the set of store-shopping trajectory deviation coefficients to generate multi-cluster portrait information; and traversing the multi-cluster portrait information and performing k-item timestamp verification based on the first store-exit timestamp to the Nth store-exit timestamp to generate the portrait combination scheme.

[0039] Next, the specific configuration of the portrait combination evaluation module 20 will be further described in detail. The portrait combination evaluation module 20 further includes: positioning the ith store-shopping trajectory information based on the three-dimensional virtual space to generate a first trajectory curve, and positioning the jth store-shopping trajectory information to generate a second trajectory curve, where i and j are any two different trajectories from the first store-shopping trajectory information to the Nth store-shopping trajectory information, N≥i≥1, N≥j≥1, and i≠j; connecting the starting positions of the first trajectory curve and the second trajectory curve, and connecting the ending positions of the first trajectory curve and the second trajectory curve to obtain the enclosed graphic area information, which is set as the first store-shopping trajectory deviation coefficient and added to the set of store-shopping trajectory deviation coefficients.

[0040] Next, the specific configuration of the portrait combination evaluation module 20 will be further described in detail. The portrait combination evaluation module 20 further includes: obtaining the first cluster of portrait information of the multiple clusters of portrait information, where the first cluster of portrait information has a portrait quantity M; setting an integer k value, with the initial value set to 2, performing k-item enumeration combinations on the first cluster of portrait information to generate a first set of k-item portrait combination schemes; based on the first store exit timestamp until the Nth store exit timestamp, traversing the first set of k-item portrait combination schemes to perform pairwise portrait timestamp deviation verification: when any timestamp deviation is greater than or equal to the timestamp deviation threshold, eliminating the corresponding first set of k-item portrait combination schemes; when each timestamp deviation is less than the timestamp deviation threshold, adding the corresponding first set of k-item portrait combination schemes to the portrait combination scheme; when the traversal of the first set of k-item portrait combination schemes is completed, determining whether k is greater than or equal to M; if k is less than M, incrementing k by one and looping through the k-item timestamp verification; if k is greater than or equal to M, outputting the portrait combination scheme.

[0041] Next, the specific configuration of the abnormal order identification information generation module 60 will be described in detail. The abnormal order identification information generation module 60 may further include: obtaining the uploaded order information of the preset time zone through the cashier device, where the uploaded order information includes the total amount of incoming funds and the number of actual payment accounts; obtaining the total amount of the first receivables and the number of the first payment accounts according to the first list of receivables, and obtaining the total amount of the second receivables and the number of the second payment accounts according to the second list of receivables; constructing a theoretical collection interval according to the total amount of the first receivables and the total amount of the second receivables; constructing a theoretical account number interval according to the number of the first payment accounts and the number of the second payment accounts; when the total amount of the incoming funds does not belong to the theoretical collection interval, and / or the number of the actual payment accounts does not belong to the theoretical account number interval, if the uploaded order information does not meet the first list of receivables and the first list of theoretical collection times, and does not meet the second list of receivables and the second list of theoretical collection times, generating the abnormal order identification information.

[0042] The monitoring and warning system of an enterprise cashier system provided by an embodiment of the present invention can execute the monitoring and warning method of an enterprise cashier system provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0043] Although this application makes various references to certain modules in the system according to the embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or the server. The various units and modules included are only divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of mutual distinction and are not used to limit the protection scope of the present invention.

[0044] The above specific embodiments do not constitute a limitation on the protection scope of this application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of this application shall be included within the protection scope of this application.

Claims

1. A monitoring and early warning method for an enterprise cash register system, characterized in that: include: Obtaining, through the in-store video stream uploaded by the image acquisition device, the first pickup timestamp, the first pickup category information, the first store exit timestamp, and the first store browsing trajectory information of the first person in the preset time zone, until the Nth pickup timestamp, the Nth pickup category information, the Nth store exit timestamp, and the Nth store browsing trajectory information of the Nth person; Performing a portrait combination evaluation based on the first store exit timestamp to the Nth store exit timestamp, and the first store visiting trajectory information to the Nth store visiting trajectory information, to obtain a portrait combination solution; Extracting a goods category combination scheme that satisfies the portrait combination scheme from the first pickup category information to the Nth pickup category information; Traversing the product category combination scheme, based on the discount constraint condition, counting the first account receivable list, matching the first account receivable information with the first theoretical collection time list based on the first pickup timestamp to the Nth pickup timestamp, and the first store exit timestamp to the Nth store exit timestamp; Based on the first pickup category information until the Nth pickup category information and the discount constraint condition, a second account receivable list is counted, and based on the first pickup timestamp until the Nth pickup timestamp and the first store exit timestamp until the Nth store exit timestamp, a second theoretical collection time list is matched for the second account receivable information; Order analysis is performed based on the first accounts receivable list and the first theoretical collection time list, as well as the second accounts receivable list and the second theoretical collection time list, and abnormal order identification information is generated and sent to the management end for early warning.

2. The monitoring and early warning method of the enterprise cash register system according to claim 1, characterized in that: Obtain the first pickup timestamp, first pickup category information, first store exit timestamp, and first store browsing trajectory information of the first person portrait in the preset time zone, including: Performing semantic segmentation on the in-store video stream to obtain a first portrait video identification sequence; Performing a pickup analysis according to the first portrait video identification sequence to obtain the first pickup timestamp and the first pickup category information; Extracting the motion position time sequence according to the first portrait video identification sequence to obtain the first store-shopping trajectory information; The end time is extracted according to the first portrait video identification sequence and is set as the first store exit timestamp.

3. A monitoring and early warning method for an enterprise cash register system as claimed in claim 2, characterized in that: Performing a pickup analysis according to the first portrait video identification sequence to obtain the first pickup timestamp and the first pickup category information includes: When the first cargo is taken out by the first person and not put back into the cargo position, a timestamp of picking up the first cargo, information about the category of the first cargo, and information about the quantity of the first cargo picked up are recorded; Adding the first cargo pickup timestamp to the first cargo pickup timestamp; The first cargo category information and the first cargo pickup quantity information are added to the first cargo pickup category information.

4. The monitoring and early warning method of a cash register system of an enterprise as claimed in claim 1, characterized in that: Based on the first store exit timestamp to the Nth store exit timestamp, and the first store visiting trajectory information to the Nth store visiting trajectory information, a portrait combination evaluation is performed to obtain a portrait combination solution, including: Performing pairwise trajectory deviation analysis based on the first store-shopping trajectory information to the Nth store-shopping trajectory information to generate a store-shopping trajectory deviation coefficient set; performing cluster analysis on the first portrait to the Nth portrait according to the store-shopping trajectory deviation coefficient set to generate multiple clusters of portrait information; The multiple clusters of portrait information are traversed, and k timestamp verifications are performed based on the first store exit timestamp to the Nth store exit timestamp to generate the portrait combination solution.

5. The monitoring and early warning method of the enterprise cash register system as claimed in claim 4, characterized in that: Performing pairwise trajectory deviation analysis based on the first store-shopping trajectory information to the Nth store-shopping trajectory information to generate a store-shopping trajectory deviation coefficient set includes: According to the three-dimensional virtual space, the i-th store-shopping trajectory information is positioned to generate a first trajectory curve, and the j-th store-shopping trajectory information is positioned to generate a second trajectory curve, wherein i and j are any two different trajectories from the first store-shopping trajectory information to the N-th store-shopping trajectory information, N≥i≥1, N≥j≥1, i≠j; The starting points of the first trajectory curve and the second trajectory curve are connected, and the end points of the first trajectory curve and the second trajectory curve are connected to obtain closed figure area information, which is set as the first store shopping trajectory deviation coefficient and added to the store shopping trajectory deviation coefficient set.

6. A monitoring and early warning method for an enterprise cash register system as claimed in claim 4, characterized in that: Traversing the multiple clusters of portrait information, performing k timestamp verifications based on the first store exit timestamp to the Nth store exit timestamp, and generating the portrait combination scheme, including: Obtaining a first cluster of portraits of the plurality of clusters of portrait information, wherein the first cluster of portraits has a number M of portraits; Set an integer k value, the initial value of which is 2, and perform k-item enumeration combinations on the first cluster of portraits to generate the first k-item portrait combination solution set; Based on the first store exit timestamp to the Nth store exit timestamp, the first k portrait combination solution sets are traversed to perform pairwise portrait timestamp deviation verification: When any timestamp deviation is greater than or equal to the timestamp deviation threshold, the corresponding first k portrait combination solutions are eliminated; When each timestamp deviation is less than the timestamp deviation threshold, adding the corresponding first k portrait combination solutions into the portrait combination solution; When the traversal of the first k portrait combination solution sets is completed, determining whether k is greater than or equal to M; If k is less than M, k is increased by one, and the k-item timestamp verification is repeated; If k is greater than or equal to M, the portrait combination solution is output.

7. The monitoring and early warning method of the enterprise cash register system according to claim 1, characterized in that: Performing order analysis based on the first accounts receivable list and the first theoretical collection time list, and the second accounts receivable list and the second theoretical collection time list to generate abnormal order identification information includes: Obtaining the uploaded order information of the preset time zone through the cash register device, wherein the uploaded order information includes the total amount of the deposited funds and the number of actual payment accounts; According to the first accounts receivable list, obtain the total amount of first accounts receivable and the number of first payment accounts, and according to the second accounts receivable list, obtain the total amount of second accounts receivable and the number of second payment accounts; Constructing a theoretical collection interval according to the first total amount of accounts receivable and the second total amount of accounts receivable; Constructing a theoretical account quantity interval according to the first payment account quantity and the second payment account quantity; When the total amount of received funds does not fall within the theoretical collection range, and / or the actual number of payment accounts does not fall within the theoretical account number range, if the uploaded order information does not satisfy the first receivables list and the first theoretical collection time list, and does not satisfy the second receivables list and the second theoretical collection time list, the abnormal order identification information is generated.

8. A monitoring and early warning system for an enterprise cash register system, characterized in that: The system is used to implement the monitoring and early warning method of an enterprise cash register system according to any one of claims 1 to 7, and the system includes: A video stream information acquisition module, the video stream information acquisition module is used to obtain the first pickup timestamp, the first pickup category information, the first store exit timestamp and the first store browsing trajectory information of the first person in a preset time zone through the in-store video stream uploaded by the image acquisition device, until the Nth pickup timestamp, the Nth pickup category information, the Nth store exit timestamp and the Nth store browsing trajectory information of the Nth person; A portrait combination evaluation module, the portrait combination evaluation module is used to perform portrait combination evaluation based on the first store exit timestamp to the N-th store exit timestamp, and the first store visiting trajectory information to the N-th store visiting trajectory information to obtain a portrait combination solution; A goods category combination solution extraction module, the goods category combination solution extraction module is used to extract goods category combination solutions that meet the portrait combination solution from the first pickup category information to the Nth pickup category information; A first account receivable list statistics module, the first account receivable list statistics module is used to traverse the commodity category combination scheme, based on the discount constraint condition, count the first account receivable list, based on the first pickup timestamp to the Nth pickup timestamp, and the first store departure timestamp to the Nth store departure timestamp for matching the first account receivable information with the first collection theoretical time list; A second account receivable list statistics module, the second account receivable list statistics module is used to count the second account receivable list based on the first pickup category information until the Nth pickup category information, based on the discount constraint condition, and match the second account receivable information with a second collection theoretical time list based on the first pickup timestamp until the Nth pickup timestamp, and the first store exit timestamp until the Nth store exit timestamp; An abnormal order identification information generation module is used to perform order analysis based on the first receivables list and the first theoretical collection time list, as well as the second receivables list and the second theoretical collection time list, to generate abnormal order identification information and send it to the management end for early warning.

Images