CN-116011822-B - Store cashing management method and device, computer equipment and storage medium

CN116011822BCN 116011822 BCN116011822 BCN 116011822BCN-116011822-B

Abstract

The invention discloses a store cashing management method, a store cashing management device, computer equipment and a storage medium, wherein the method comprises the steps of obtaining private cashing characteristics of each store, and constructing a potential private cashing shop atmosphere risk identification model according to the private cashing characteristics; predicting potential private receipt probability of each store by utilizing a potential private receipt gate shop atmosphere risk identification model, selecting n stores with the highest potential private receipt probability as first target stores, acquiring a private receipt tool image, constructing a private receipt tool identification model according to the private receipt tool image, carrying out private receipt tool identification on the first target stores by utilizing the private receipt tool identification model, extracting a receipt area, identifying employee and customer information in the receipt area, and judging whether private receipt conditions exist in the corresponding stores by combining the private receipt tool information. According to the method and the device, whether the private collection condition exists or not is judged by identifying the private collection tool and identifying whether the condition conforming to the private collection exists or not for the collection area, so that the store collection management efficiency can be improved, and the private collection condition of the store is avoided or reduced.

Inventors

ZHANG GUITIAN
YUAN LISHA
ZHANG ZHIBIN

Assignees

广州市钱大妈信息科技有限公司

Dates

Publication Date: 20260508
Application Date: 20230118

Claims (8)

1. A store cashing management method, comprising: acquiring private cashing characteristics of each store, and constructing a potential private cashing gate shop atmosphere risk identification model according to the private cashing characteristics, wherein the private cashing characteristics comprise store operation characteristics, commodity classification characteristics, time interval sales characteristics and private cashing tool characteristics; Predicting potential privacy probability for each store by using the potential privacy gate shop atmosphere risk identification model, so as to select n stores with the highest potential privacy probability as first target stores; acquiring a private tool image, and constructing a private tool identification model according to the private tool image; carrying out private tool identification on the first target store by using the private tool identification model, outputting the first target store containing the private tool as a second target store, and outputting private tool information corresponding to the private tool, wherein the private tool information refers to the front and rear time points of the appearance of the private tool; Extracting a cashing area from the second target store, identifying employee and customer information in the cashing area, and judging whether a private cashing situation exists in the corresponding store according to the private cashing tool information; the step of extracting the cashing area from the second target store, identifying staff and customer information in the cashing area, and judging whether the corresponding store has a private cashing condition according to the private cashing tool information, comprising the following steps: performing image recognition on the second target store by using yolov l6 detection models so as to extract a cashing area in the second target store; the employee characteristics are obtained and input into yolov l6 detection models, so that an employee identification model is constructed; Performing person identification on the cashing area by using the yolov l6 detection model, and performing employee identification on the cashing area by using the employee identification model; judging whether staff and/or customers exist in the cashing area or not based on the person identification result and the staff identification result; When staff and customers exist in the cashing area at the same time, identifying customer information corresponding to the customers by utilizing the yolov l6 detection model, wherein the customer information comprises commodities around the customers and customer payment tools; Acquiring transaction information of the second target store; and carrying out private cashing judgment by combining the transaction information, the private tool information, the customer information and the person identification result of the cashing area.
2. The store cashing management method according to claim 1, wherein the acquiring the private cashing characteristics of each store and constructing the potential private gate shop atmosphere risk identification model according to the private cashing characteristics comprises: Calculating a contribution value of each private cashing feature by adopting a random forest model; Selecting m pieces with highest contribution values as input features; And inputting the input characteristics into xgboost algorithm to perform classification prediction so as to construct the potential privacy gate shop atmosphere risk identification model.
3. The store cashing management method according to claim 2, wherein the calculating the contribution value of each private cashing feature using a random forest model includes: and adopting a base value as the contribution value, and calculating the base value of each private cashing feature according to the following formula: Wherein D represents the private cash feature, k represents a private cash feature number, P k represents a feature weight of k, k=1, 2,3.
4. The store cashing management method according to claim 1, wherein the acquiring a private tool image and constructing a private tool identification model according to the private tool image comprises: Inputting the private tool image into yolov l6 detection models for training and learning so as to construct the private tool identification model; and optimizing and updating the private tool identification model by using a loss function according to the following steps: Loss=γ 1 L cls +γ 2 L obj +γ 3 L loc Where Loss represents a Loss function, γ 1 、γ 2 、γ 3 represents a balance coefficient, L cls represents a classification Loss, L obj represents a confidence Loss, and L loc represents a positioning Loss.
5. The store cashing management method of claim 4, wherein the inputting the private tool image into yolov l6 detection model for training learning, thereby constructing the private tool identification model, comprises: carrying out random data enhancement processing on the private tool image through the input end of the yolov l6 detection model; Inputting the private tool image subjected to random data enhancement processing to a backstage layer to extract image characteristics of the private tool image; performing feature enhancement processing on the image features by utilizing a neg layer of the yolov l6 detection model; And carrying out output prediction on the image characteristics subjected to the characteristic enhancement processing through the head layer of the yolov l6 detection model.
6. A store cashing management device, comprising: The first model construction unit is used for acquiring private cashing characteristics of each store and constructing a potential private cashing door shop atmosphere risk identification model according to the private cashing characteristics, wherein the private cashing characteristics comprise store operation characteristics, commodity classification characteristics, time period sales characteristics and private cashing tool characteristics; The first store selection unit is used for predicting potential private probability of each store by using the potential private gate shop atmosphere risk identification model so as to select n stores with the highest potential private probability as first target stores; the second model building unit is used for obtaining a private tool image and building a private tool identification model according to the private tool image; The second store selection unit is used for identifying the private tools of the first target store by utilizing the private tool identification model, outputting the first target store containing the private tools as a second target store, and outputting private tool information corresponding to the private tools, wherein the private tool information refers to the time points before and after the occurrence of the private tools; the cashing management unit is used for extracting a cashing area for the second target store, identifying staff and customer information in the cashing area, and judging whether a private cashing situation exists in the corresponding store according to the private cashing tool information; the cashier management unit includes: the image recognition unit is used for carrying out image recognition on the second target store by utilizing yolov l6 detection models so as to extract a cashing area in the second target store; the employee model building unit is used for obtaining employee characteristics and inputting the employee characteristics into the yolov l6 detection model so as to build an employee identification model; the person identification unit is used for carrying out person identification on the cashing area by utilizing the yolov l6 detection model and carrying out employee identification on the cashing area by utilizing the employee identification model; The person judging unit is used for judging whether staff and/or customers exist in the cashing area or not based on the person identification result and the staff identification result; an information identification unit, configured to identify customer information corresponding to a customer by using the yolov l6 detection model when staff and the customer exist in the cashing area at the same time, where the customer information includes a customer-surrounding commodity and a customer payment tool; The cashier management unit further includes: a transaction information acquisition unit configured to acquire transaction information of the second target store; and the private cashing judging unit is used for carrying out private cashing judgment by combining the transaction information, the private cashing tool information, the customer information and the person identification result of the cashing area.
7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the store cashing management method of any one of claims 1 to 5 when the computer program is executed.
8. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the store cashing management method according to any one of claims 1 to 5.

Description

Store cashing management method and device, computer equipment and storage medium Technical Field The present invention relates to the field of computer software technologies, and in particular, to a store cashing management method, a store cashing management device, a store cashing management computer device, and a store medium. Background The store private cash register refers to a cash register in which a store clerk presents a cash register tool such as a personal cash register or pos machine to a customer, or a cash register in which the cash register is not specified by an enterprise, and the cash register originally belonging to the store is brought into a pocket of the store clerk. The store cash system is not passed through when the store is specifically operated, so the store cash system does not have direct data representation, and the related amount is not too large at one time, so private stores are difficult to dig. For enterprises with a large number of stores, a plurality of cameras are usually installed in one store, and the monitoring video of each camera has more than ten hours, so that if each second of each frame of image is detected only through image algorithm detection, huge calculation force is involved, the detection efficiency is low, huge GPU operation resources are needed in the detection process, network factors are considered, and a GPU server is needed to be deployed for each store, so that the overall detection cost is too high and the detection is difficult to realize. Therefore, how to improve the store cashing management efficiency and efficiently detect the store private cashing situation so as to reduce or even avoid the store private cashing situation is a problem to be solved by the person skilled in the art. Disclosure of Invention The embodiment of the invention provides a store cashing management method, a store cashing management device, computer equipment and a storage medium, and aims to improve store cashing management efficiency and avoid or reduce private cashing situations of stores. In a first aspect, an embodiment of the present invention provides a store cashing management method, including: acquiring private cashing characteristics of each store, and constructing a potential private cashing gate shop atmosphere risk identification model according to the private cashing characteristics, wherein the private cashing characteristics comprise store operation characteristics, commodity classification characteristics, time interval sales characteristics and private cashing tool characteristics; Predicting potential privacy probability for each store by using the potential privacy gate shop atmosphere risk identification model, so as to select n stores with the highest potential privacy probability as first target stores; acquiring a private tool image, and constructing a private tool identification model according to the private tool image; carrying out private tool identification on the first target store by using the private tool identification model, outputting the first target store containing the private tool as a second target store, and outputting private tool information corresponding to the private tool; And extracting a cashing area from the second target store, identifying staff and customer information in the cashing area, and judging whether a private cashing situation exists in the corresponding store by combining the private cashing tool information. In a second aspect, an embodiment of the present invention provides a store cashing management device, including: The first model construction unit is used for acquiring private cashing characteristics of each store and constructing a potential private cashing door shop atmosphere risk identification model according to the private cashing characteristics, wherein the private cashing characteristics comprise store operation characteristics, commodity classification characteristics, time period sales characteristics and private cashing tool characteristics; The first store selection unit is used for predicting potential private probability of each store by using the potential private gate shop atmosphere risk identification model so as to select n stores with the highest potential private probability as first target stores; the second model building unit is used for obtaining a private tool image and building a private tool identification model according to the private tool image; The second store selection unit is used for identifying the private tool of the first target store by using the private tool identification model, outputting the first target store containing the private tool as a second target store, and outputting private tool information corresponding to the private tool; and the cashing management unit is used for extracting a cashing area from the second target store, identifying staff and customer information in the cashing area, and judging whether the corresponding store has a private