Replenishment demand analysis method, device and equipment and readable storage medium

Abstract

The invention discloses a replenishment demand analysis method, device and equipment and a readable storage medium, and the method comprises the steps: obtaining an initial image of a goods shelf region shot at multiple angles, carrying out the enhancement and denoising of the initial image of the goods shelf region, and generating a detection image of the goods shelf region; inputting the shelf area detection image into a pre-trained goods identification model to obtain current goods data output by the goods identification model; generating current sales data based on the called current cashier system data and the online sales platform data, and taking the current sales data as demand prediction basic data; inputting the demand prediction basic data and the current goods data into a pre-trained sales demand model to obtain a sales demand prediction result; and determining a current replenishment demand according to the current goods data, the sales demand prediction result and preset safety inventory setting. According to the invention, the real-time performance, the accuracy, the intelligence and the automation level of the goods replenishment management of the convenience store are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent replenishment, and more specifically, to a replenishment demand analysis method, device, equipment and readable storage medium. Background Art

[0002] Currently, there are two main ways to manage replenishment in shops such as convenience stores: one is to manually inspect shelves regularly, where store clerks visually observe the remaining quantity of goods to determine whether replenishment is needed; the other is to use a simple inventory management system that records the quantity of goods in and out of the warehouse and sets an inventory threshold. When the inventory is lower than the threshold, the system automatically issues a replenishment reminder. These two methods together constitute the mainstream technical framework for replenishment management in convenience stores.

[0003] Although existing technologies have met the basic needs of convenience store replenishment to a certain extent, there are still many shortcomings. First, the manual inspection method is inefficient, not only consuming a lot of time and manpower costs, but also prone to omissions and misjudgments due to human factors (such as fatigue and negligence), thus affecting sales continuity and customer satisfaction. Secondly, relying on the threshold setting of the inventory management system for replenishment, although it can reduce manual intervention to a certain extent, this method cannot reflect the sales of goods on the shelves in real time, and it is easy to cause untimely or excessive replenishment due to information asymmetry, increasing inventory costs and waste. In addition, the existing system relies on the sales that have already occurred for replenishment, that is, it replenishes what is selling fast or is about to sell out in the store. It cannot make intelligent predictions and adjustments based on dynamic factors such as sales trends and seasonal changes of goods, and lacks flexibility and foresight.

[0004] Based on this, the present application provides a replenishment demand analysis solution to solve the problems of low efficiency, insufficient accuracy, lack of flexibility and foresight in replenishment management in the prior art. Summary of the invention

[0005] In view of this, the present application provides a replenishment demand analysis method, device, equipment and readable storage medium, which introduces a product recognition model to quickly and accurately identify the products in the processed images, and at the same time introduces a sales demand model to realize intelligent prediction of product sales demand, and finally generates the current replenishment demand, thereby realizing the improvement of the real-time, accuracy, intelligence and automation level of convenience store product replenishment management.

[0006] A replenishment demand analysis method, comprising:

[0007] Acquire initial images of the shelf area taken from multiple angles, and perform enhancement and denoising processing on the initial images of the shelf area to generate a shelf area detection image;

[0008] Inputting the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images labeled with product type, quantity and placement position;

[0009] Generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as basic data for demand forecasting;

[0010] Input the demand forecast basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecast result output by the sales demand model, wherein the sales demand model is trained based on historical sales data by combining a long short-term memory network LSTM with an attention mechanism;

[0011] The current replenishment demand is determined based on the current product data, the sales demand forecast result and the preset safety stock setting.

[0012] Optionally, after obtaining the initial image of the shelf area shot from multiple angles, the method further includes:

[0013] The initial image of the shelf area is subjected to image sharpening processing and light correction processing.

[0014] Optionally, also include:

[0015] Retrieve product sales data of associated stores;

[0016] The current sales data and the product sales data of the associated stores are used as basic data for demand forecasting.

[0017] Optionally, also include:

[0018] Get a list of best-selling products across all channels;

[0019] The current sales data and the omni-channel hot-selling product list are used as basic data for demand forecasting.

[0020] Optionally, also include:

[0021] Determine slow-moving goods based on the current sales data;

[0022] According to the omni-channel hot-selling product list, determining replacement products belonging to the omni-channel hot-selling product list or the acquired new product list, and generating a replacement list recording the slow-moving products and the replacement products, wherein the replacement products are used to replace the slow-moving products;

[0023] Extracting information about slow-moving goods from the current goods data and the current sales data, and designing a promotion strategy for the slow-moving goods based on the information about the slow-moving goods;

[0024] Determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, including:

[0025] The current replenishment demand is determined based on the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replacement list.

[0026] Optionally, also include:

[0027] Determine the replenishment cycle of each product according to the replenishment process corresponding to each product and the supplier response time;

[0028] Determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, including:

[0029] The current replenishment demand is determined according to the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replenishment cycle of each product.

[0030] Optionally, the attention mechanism sets corresponding weights for each product based on the distribution of consumer groups, current temperature, current promotional strategies, and upcoming holidays.

[0031] A replenishment demand analysis device, comprising:

[0032] The area detection unit is used to obtain the initial image of the shelf area taken from multiple angles, and perform enhancement and denoising processing on the initial image of the shelf area to generate a shelf area detection image;

[0033] A product recognition unit, used for inputting the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images marked with product type, quantity and placement position;

[0034] A forecasting data unit, used to generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as demand forecasting basic data;

[0035] A demand forecasting unit, used for inputting the demand forecasting basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecasting result output by the sales demand model, wherein the sales demand model is obtained by combining a long short-term memory network LSTM with an attention mechanism and is trained based on historical sales data;

[0036] The demand determination unit is used to determine the current replenishment demand based on the current product data, the sales demand forecast result and the preset safety stock setting.

[0037] A replenishment demand analysis device, comprising a memory and a processor;

[0038] The memory is used to store programs;

[0039] The processor is used to execute the program to implement each step of the replenishment demand analysis method as described in any one of the above items.

[0040] A readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, each step of the replenishment demand analysis method as described in any one of the above items is implemented.

[0041] It can be seen from the above technical solutions that the replenishment demand analysis method, device, equipment and readable storage medium provided by the embodiment of the present application first obtain the initial image of the shelf area taken from multiple angles, and enhance and denoise the initial image of the shelf area to generate a shelf area detection image. The shelf area detection image is input into a pre-trained product recognition model to obtain the current product data output by the product recognition model. At this time, based on the retrieved current cash register system data and online sales platform data, the current sales data is generated, and the current sales data is used as the demand forecasting basic data. After that, the demand forecasting basic data and the current product data are input into a pre-trained sales demand model to obtain the sales demand forecasting result output by the sales demand model. The sales demand model adopts a combination of a long short-term memory network LSTM and an attention mechanism, and is obtained based on historical sales data training, which can realize intelligent prediction of product sales demand. Finally, the current replenishment demand is determined based on the current product data, the sales demand forecasting result and the preset safety stock setting.

[0042] On the one hand, this application uses a pre-trained product recognition model to quickly and accurately identify the processed images, obtain current product data, realize real-time and accurate monitoring of shelf status, and ensure the accuracy of product recognition. On the other hand, an intelligent prediction mechanism is introduced. By combining the current cash register system data, online sales platform data, and historical sales data, a sales demand model combining a long short-term memory network (LSTM) and an attention mechanism is used to accurately predict the sales demand of goods. This intelligent prediction not only takes into account historical sales trends, but also incorporates dynamic factors such as temperature seasonal changes, holidays, and promotional activities, making the replenishment plan more flexible and adaptable. In addition, from image acquisition, product recognition to sales forecasting and replenishment demand determination, the entire process of this application does not require a lot of manual intervention, which greatly reduces labor costs and reduces the possibility of human errors. This highly automated replenishment management method not only improves the efficiency of replenishment, but also ensures the accuracy and timeliness of the replenishment plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0044] Figure 1 A flowchart of a replenishment demand analysis method disclosed in an embodiment of the present application;

[0045] Figure 2 A schematic diagram of a replenishment demand analysis device disclosed in an embodiment of the present application;

[0046] Figure 3 This is a hardware structure block diagram of a replenishment demand analysis device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] The present application can be used in many general or special computing device environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multi-processor devices, distributed computing environments including any of the above devices or equipment, etc.

[0049] The embodiment of the present application provides a replenishment demand analysis method, which can be applied to various systems or platforms involving product management or replenishment management, and can also be applied to various computer terminals or smart terminals. The execution subject can be a processor or server of the computer terminal or smart terminal.

[0050] Next, the present application scheme is introduced. The present application proposes the following technical scheme, please see below for details.

[0051] Figure 1 This is a flow chart of a replenishment demand analysis method disclosed in an embodiment of the present application.

[0052] like Figure 1 As shown, the method may include:

[0053] Step S1, obtaining an initial image of the shelf area taken from multiple angles, and performing enhancement and denoising processing on the initial image of the shelf area to generate a shelf area detection image.

[0054] Specifically, in order to efficiently monitor shelf status, adjust inventory in a timely manner, and ensure the attractiveness of product displays, high-definition cameras are usually installed above the shelves to collect real-time image information of the goods on the shelves. In order to obtain a more comprehensive and detailed initial image of the shelf area, these cameras are carefully arranged at multiple angles to ensure that the entire shelf area is fully covered from the front, side, and even top, thereby capturing key information such as the placement of goods, quantity changes, and customer interactions.

[0055] When installing cameras, not only the number and location of cameras should be considered, but also the autofocus and light adjustment functions should be paid special attention to. The autofocus function ensures that the focus can be adjusted quickly and accurately to maintain image clarity regardless of the distance between the camera and the shelf. After obtaining the initial image of the shelf area taken from multiple angles, it is necessary to enhance and denoise it to generate a high-quality shelf area detection image that is conducive to model analysis and processing.

[0056] The process of acquiring shelf area detection images not only relies on camera technology and intelligent image processing algorithms, but also requires a deep understanding and precise control of the internal environment of the convenience store to ensure that the final detection image can accurately reflect the actual situation of the shelf area.

[0057] In addition, after obtaining the initial image of the shelf area shot from multiple angles, the following may also be included:

[0058] The initial image of the shelf area is subjected to image sharpening processing and light correction processing.

[0059] After obtaining the initial image of the shelf area taken from multiple angles, the next step is image enhancement and denoising to further improve the image quality. In addition, this application can also perform image sharpening and light correction on the initial image of the shelf area. Image sharpening mainly enhances the image edges and detail information to make the outline of the goods on the shelf clearer and reduce blur. This processing helps the algorithm to more accurately identify the type, quantity and location information of the goods, and improve the accuracy of shelf management.

[0060] At the same time, due to the complex and changeable lighting conditions in the store, there may be uneven brightness and shadows in the initial image, which will affect the effect of image analysis. Through light correction, the overall brightness and contrast of the image can be adjusted, the lighting differences in different areas can be balanced, the shadow effect can be eliminated or reduced, and the shelf area detection image can be closer to the actual scene, improving the accuracy and reliability of image analysis.

[0061] Step S2: input the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model.

[0062] Specifically, after completing the pre-processing steps such as enhancement and denoising of the shelf area detection images, the next key step is to input these high-quality images into the pre-trained product recognition model to obtain the product data on the current shelf. This process relies on the powerful capabilities of deep learning technology, especially the convolutional neural network (CNN) model for image recognition and classification tasks.

[0063] The product recognition model is trained based on a large number of sample shelf area images labeled with product types, quantities, and placements. These sample images cover a variety of different product types, arrangements, and possible interference factors (such as lighting changes, obstructions, etc.), ensuring the comprehensiveness and robustness of model training. Through repeated learning of these labeled data by deep learning algorithms, the model gradually acquires the ability to identify the characteristics of different goods.

[0064] In actual applications, when the preprocessed shelf area detection image is input into the model, the model will use the feature representation and classification rules learned during its training process to quickly and accurately analyze the goods in the image. In this process, the CNN model automatically extracts useful information from the image, such as edges and textures, through its unique convolutional layer, pooling layer and other structures, and then constructs a high-dimensional feature vector that can represent different goods. Subsequently, these feature vectors will be sent to the fully connected layer for classification, and finally output key information such as the type, quantity and placement of the goods in the current image.

[0065] The product identification model can realize fast and accurate product identification in practical applications, which not only greatly improves the efficiency of shelf management, but also realizes real-time and accurate monitoring of shelf status, ensuring the timely update and accuracy of product data. It is convenient for managers to quickly understand the real-time status of goods on the shelves, adjust inventory in time, and optimize display layout, thereby improving customer satisfaction and sales.

[0066] Step S3: Generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as basic data for demand forecasting.

[0067] Specifically, collect sales data from convenience stores, including the type, quantity and time of goods in each transaction, and obtain sales data in real time by integrating with the cash register system. At the same time, data from online sales platforms can also be collected to integrate sales data from multiple sources. This process may include processing steps such as data cleaning, deduplication, and format unification to ensure that data from different sources can be seamlessly connected and used for subsequent analysis. By integrating sales data from multiple sources, a more complete and accurate current sales data can be obtained, providing a solid basic data for demand forecasting. Ultimately, these current product sales data will be used as the basis for demand forecasting.

[0068] Step S4: input the demand forecast basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecast result output by the sales demand model.

[0069] Specifically, the sales demand model adopts a combination of a long short-term memory network LSTM and an attention mechanism, and is trained based on historical sales data. The attention mechanism sets the corresponding weights of each product based on the distribution of consumer groups, current temperature, current promotional strategies, and holiday expiration situations. This application uses a sales forecasting model that combines a long short-term memory network (LSTM) with an attention mechanism to forecast sales demand. The sales forecasting model can process time series data, capture long-term trends and cyclical changes in sales data, and dynamically assign weights according to the degree of impact of different factors on sales. This makes the forecast results more accurate, and can accurately estimate the demand for goods in advance to avoid out-of-stock or over-replenishment. Compared with the prior art that relies on fixed thresholds and simple inventory management systems, the timeliness and accuracy of replenishment are greatly improved.

[0070] Specifically, the Long Short-Term Memory (LSTM) network is mainly used to process time series information in sales data. Since sales data has obvious time characteristics, such as sales trends vary in different seasons, holidays, and promotions, LSTM can capture these long-term trends and cyclical changes, thereby more accurately predicting future sales demand.

[0071] LSTM controls the flow and storage of information through its special gating mechanism, namely the input gate, forget gate, and output gate. The input gate determines how new information is added to the cell state, the forget gate determines which old information should be discarded, and the output gate determines which parts of the cell state should be output. Through such a mechanism, LSTM can remember long-term dependencies. For sales data with seasonality and periodicity, it can effectively capture the long-term trends, seasonal fluctuations, and periodic change patterns and trends in sales data, providing support for accurate sales forecasts.

[0072] The introduction of the attention mechanism further enhances the intelligence and adaptability of the model. It dynamically adjusts the weight of each product in the prediction process according to the current sales environment, market conditions and external influencing factors. This flexible weight allocation mechanism enables the model to more accurately predict sales demand in specific situations. The attention mechanism assigns weights according to the importance of different influencing factors to sales forecasts. For example, in summer, rising temperatures may lead to increased sales of refreshing beverages. At this time, the attention mechanism will assign a higher weight to the temperature factor; during holidays, such as National Day, people's consumption habits and needs will change, and the attention mechanism will assign a higher weight to holiday factors; and during promotional activities, the attention mechanism will increase the weight of promotional factors accordingly. Through the attention mechanism, the model can more flexibly and intelligently consider the dynamic changes of various influencing factors, making the replenishment prompts more in line with the actual sales situation and market demand, and reducing the risk of out-of-stock or over-replenishment.

[0073] Step S5: Determine the current replenishment demand based on the current product data, the sales demand forecast result and the preset safety stock setting.

[0074] Specifically, in the process of determining the current replenishment demand, the current product data, sales demand forecast results and preset safety stock settings will be comprehensively considered. First, the actual inventory of each product in the current inventory will be understood based on the current product data collected in real time. Then, the trained sales demand forecasting model will be used to accurately forecast the sales demand in the future. This forecast result not only takes into account the trend and cyclical changes of historical sales data, but also integrates the comprehensive consideration of current market conditions, consumer behavior and external influencing factors, so it has high accuracy and reliability.

[0075] After understanding the current inventory and future sales demand of the goods, the preset safety stock settings will be further referenced. Safety stock is the amount of inventory reserved to cope with uncertainties such as market fluctuations and supply chain delays. Its setting is usually based on historical experience, supplier delivery cycle, and corporate risk tolerance. By comparing the differences between the current inventory, future sales forecasts, and safety stock settings, it is possible to accurately determine which goods’ inventory levels are about to or have fallen below the safety threshold, thereby determining replenishment needs. It is understandable that replenishment needs are not sent just because the inventory level is lower than the preset safety stock setting. For example, if the high temperature continues in the next week, the demand for cooling and cooling products will increase significantly. At this time, although the cooling and cooling products are not lower than the preset safety stock settings, considering that the sales of cooling and cooling products will increase explosively in the future, it is still necessary to replenish the products.

[0076] In addition, this application can also calculate the best replenishment time in advance based on the replenishment process and the supplier's response time to ensure timely replenishment of goods, including:

[0077] Determine the replenishment cycle of each product according to the replenishment process corresponding to each product and the supplier response time;

[0078] Determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, including:

[0079] The current replenishment demand is determined according to the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replenishment cycle of each product.

[0080] Specifically, this application is not limited to simple replenishment demand calculations, but also further incorporates the consideration of replenishment processes and supplier response times to optimize replenishment strategies and ensure that goods can be replenished in a timely and efficient manner. First, the replenishment cycle of each product will be accurately determined based on the unique replenishment process of each product and the response time of the supplier. This step fully considers the circulation speed of different products in the supply chain, the delivery capacity of suppliers, and the replenishment process within the convenience store.

[0081] Subsequently, the system comprehensively evaluates and determines the current replenishment needs based on the current product data, sales demand forecast results, and preset safety stock settings. In this process, not only the matching degree between the current inventory status and future sales trends is analyzed, but also the replenishment cycle of each product is combined to ensure that the replenishment plan can meet the sales demand in the short term without increasing the operating cost due to frequent replenishment.

[0082] Ultimately, the current replenishment demand generated by the system may include information such as the type of goods to be replenished, the specific replenishment quantity (calculated based on sales forecast results and safety stock settings), and the recommended replenishment time (taking into account factors such as supplier delivery cycle and goods shelf preparation time).

[0083] It can be seen from the above technical solutions that the replenishment demand analysis method, device, equipment and readable storage medium provided by the embodiment of the present application first obtain the initial image of the shelf area taken from multiple angles, and enhance and denoise the initial image of the shelf area to generate a shelf area detection image. The shelf area detection image is input into a pre-trained product recognition model to obtain the current product data output by the product recognition model. At this time, based on the retrieved current cash register system data and online sales platform data, the current sales data is generated, and the current sales data is used as the demand forecasting basic data. After that, the demand forecasting basic data and the current product data are input into a pre-trained sales demand model to obtain the sales demand forecasting result output by the sales demand model. The sales demand model adopts a combination of a long short-term memory network LSTM and an attention mechanism, and is obtained based on historical sales data training, which can realize intelligent prediction of product sales demand. Finally, the current replenishment demand is determined based on the current product data, the sales demand forecasting result and the preset safety stock setting.

[0084] On the one hand, this application uses a pre-trained product recognition model to quickly and accurately identify the processed images, obtain current product data, realize real-time and accurate monitoring of shelf status, and ensure the accuracy of product recognition. On the other hand, an intelligent prediction mechanism is introduced. By combining the current cash register system data, online sales platform data, and historical sales data, a sales demand model combining a long short-term memory network (LSTM) and an attention mechanism is used to accurately predict the sales demand of goods. This intelligent prediction not only takes into account historical sales trends, but also incorporates dynamic factors such as temperature seasonal changes, holidays, and promotional activities, making the replenishment plan more flexible and adaptable. In addition, from image acquisition, product recognition to sales forecasting and replenishment demand determination, the entire process of this application does not require a lot of manual intervention, which greatly reduces labor costs and reduces the possibility of human errors. This highly automated replenishment management method not only improves the efficiency of replenishment, but also ensures the accuracy and timeliness of the replenishment plan.

[0085] In some embodiments of the present application, in order to make the forecast results more accurate, the demand forecast basic data can also be linked with other system data. Two optional methods are provided below, but it can be understood that the present application should include but is not limited to the following methods. In practical applications, various data that may affect the sales of goods and are related to the replenishment demand forecast can be included in the calculation.

[0086] The first method is to incorporate the product sales data of related stores into demand forecasts.

[0087] Retrieve product sales data of associated stores;

[0088] The current sales data and the product sales data of the associated stores are used as basic data for demand forecasting.

[0089] Specifically, in order to improve the accuracy of demand forecasting, an innovative method was adopted to incorporate the sales data of related stores into the forecasting model. First, through efficient data integration technology, the sales data of all related stores were fully retrieved and integrated. Related stores include similar stores in the same type of business district, other chain stores, etc. These data not only reflect the sales trends of each store itself, but also contain the commonalities and differences in market demand. Subsequently, the sales data of these related stores are combined with the sales data of the current store and used together as the basic data for demand forecasting. This cross-store and cross-regional data fusion greatly enriches the input dimensions of the forecasting model, enabling it to capture market dynamics more comprehensively, thereby generating more accurate and forward-looking sales demand forecasting results.

[0090] The second method is to include the list of best-selling products across all channels into demand forecasts.

[0091] Get a list of best-selling products across all channels;

[0092] The current sales data and the omni-channel hot-selling product list are used as basic data for demand forecasting.

[0093] Specifically, in order to keep up with market trends and quickly respond to consumer demand, the omni-channel hot-selling product list is included in the demand forecast. First, through big data analysis technology, the current hot-selling product list is screened from all channels (including online e-commerce platforms, offline physical stores, social media, etc.). This list not only represents the hot trends in the market, but also reflects consumer preferences. Subsequently, this omni-channel hot-selling product list is combined with the current store sales data to form the basic data for demand forecasting. This method ensures that the forecast results are close to market reality and can capture emerging consumer hotspots, providing strong data support for store inventory management and product procurement.

[0094] On this basis, this application can also further identify slow-moving goods and make promotion and replenishment adjustments, including:

[0095] ① Based on the current sales data, determine the slow-moving goods;

[0096] ② According to the omni-channel hot-selling product list, determine the replacement products belonging to the omni-channel hot-selling product list or the acquired new product list, and generate a replacement list recording the slow-selling products and the replacement products, wherein the replacement products are used to replace the slow-selling products;

[0097] ③ extracting the slow-moving product information from the current product data and the current sales data, and designing a promotion strategy for the slow-moving product based on the slow-moving product information;

[0098] Determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, including:

[0099] The current replenishment demand is determined based on the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replacement list.

[0100] Specifically, this application further introduces a mechanism for identifying slow-moving goods and adjusting promotional replenishment to comprehensively improve commodity turnover efficiency and sales. First, the system uses intelligent analysis algorithms based on current sales data to accurately identify slow-moving goods. These goods face inventory backlogs due to insufficient market demand or improper sales strategies. Subsequently, in order to effectively alleviate the slow-moving situation, the system selects replaceable popular goods or new products based on the omni-channel hot-selling goods list and new product list. These replacement goods not only meet current market trends, but also have the potential to attract consumers. Then generate a detailed replacement list to clearly record the correspondence between slow-moving goods and replacement goods. For slow-moving goods, promotional strategies can be designed and implemented according to their characteristics and market positioning, such as price discounts, bundling sales, limited-time discounts, etc., to stimulate consumers' desire to buy and accelerate the digestion of slow-moving goods.

[0101] While implementing the promotion strategy, the current product data, sales demand forecast results, and preset safety stock settings are comprehensively considered, combined with the adjustment information in the replacement list, to re-evaluate and determine the current replenishment needs. This step ensures the flexibility and accuracy of the replenishment plan, which not only meets the continuous supply demand of hot-selling products, but also effectively avoids the inventory backlog problem caused by too many slow-moving products.

[0102] A replenishment demand analysis device provided in an embodiment of the present application is described below. The replenishment demand analysis device described below and the replenishment demand analysis method described above can refer to each other.

[0103] See also Figure 2 , Figure 2 This is a schematic diagram of a replenishment demand analysis device disclosed in an embodiment of the present application.

[0104] like Figure 2As shown, the replenishment demand analysis device may include:

[0105] The area detection unit 110 is used to obtain an initial image of the shelf area taken from multiple angles, and to perform enhancement and denoising processing on the initial image of the shelf area to generate a shelf area detection image;

[0106] A product recognition unit 120 is used to input the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images labeled with product type, quantity and placement position;

[0107] The forecasting data unit 130 is used to generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as demand forecasting basic data;

[0108] The demand forecasting unit 140 is used to input the demand forecasting basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecasting result output by the sales demand model, wherein the sales demand model is obtained by combining a long short-term memory network LSTM with an attention mechanism and is trained based on historical sales data;

[0109] The demand determination unit 150 is used to determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting.

[0110] It can be seen from the above technical solutions that the replenishment demand analysis method, device, equipment and readable storage medium provided by the embodiment of the present application first obtain the initial image of the shelf area taken from multiple angles, and enhance and denoise the initial image of the shelf area to generate a shelf area detection image. The shelf area detection image is input into a pre-trained product recognition model to obtain the current product data output by the product recognition model. At this time, based on the retrieved current cash register system data and online sales platform data, the current sales data is generated, and the current sales data is used as the demand forecasting basic data. After that, the demand forecasting basic data and the current product data are input into a pre-trained sales demand model to obtain the sales demand forecasting result output by the sales demand model. The sales demand model adopts a combination of a long short-term memory network LSTM and an attention mechanism, and is obtained based on historical sales data training, which can realize intelligent prediction of product sales demand. Finally, the current replenishment demand is determined based on the current product data, the sales demand forecasting result and the preset safety stock setting.

[0111] On the one hand, this application uses a pre-trained product recognition model to quickly and accurately identify the processed images, obtain current product data, realize real-time and accurate monitoring of shelf status, and ensure the accuracy of product recognition. On the other hand, an intelligent prediction mechanism is introduced. By combining the current cash register system data, online sales platform data, and historical sales data, a sales demand model combining a long short-term memory network (LSTM) and an attention mechanism is used to accurately predict the sales demand of goods. This intelligent prediction not only takes into account historical sales trends, but also incorporates dynamic factors such as temperature seasonal changes, holidays, and promotional activities, making the replenishment plan more flexible and adaptable. In addition, from image acquisition, product recognition to sales forecasting and replenishment demand determination, the entire process of this application does not require a lot of manual intervention, which greatly reduces labor costs and reduces the possibility of human errors. This highly automated replenishment management method not only improves the efficiency of replenishment, but also ensures the accuracy and timeliness of the replenishment plan.

[0112] Optionally, the area detection unit may further perform image sharpening processing and light correction processing on the initial image of the shelf area after acquiring the initial image of the shelf area shot at multiple angles.

[0113] Optionally, the forecasting data unit may also retrieve product sales data of associated stores, and use the current sales data and the product sales data of the associated stores as basic data for demand forecasting.

[0114] Optionally, the forecasting data unit may also obtain a list of best-selling products across all channels, and use the current sales data and the list of best-selling products across all channels as basic data for demand forecasting.

[0115] Optionally, a replenishment demand analysis device may further include a slow-moving processing unit;

[0116] The slow-selling processing unit is used to determine the slow-selling goods based on the current sales data, and determine the replacement goods belonging to the omni-channel hot-selling goods list or the acquired new product list according to the omni-channel hot-selling goods list, and generate a replacement list recording the slow-selling goods and the replacement goods, the replacement goods are used to replace the slow-selling goods, and extract the slow-selling goods information from the current goods data and the current sales data, and design a promotion strategy for the slow-selling goods based on the slow-selling goods information;

[0117] The demand determination unit determines the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, which may include:

[0118] The current replenishment demand is determined based on the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replacement list.

[0119] Optionally, the demand determination unit may also determine the replenishment cycle of each product according to the replenishment process corresponding to each product and the supplier response time;

[0120] The demand determination unit determines the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, which may include:

[0121] The current replenishment demand is determined according to the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replenishment cycle of each product.

[0122] Optionally, the attention mechanism sets corresponding weights for each product based on the distribution of consumer groups, current temperature, current promotional strategies, and upcoming holidays.

[0123] The replenishment demand analysis device provided in the embodiment of the present application can be applied to replenishment demand analysis equipment. Figure 3 The hardware structure diagram of the replenishment demand analysis device is shown in FIG. Figure 3 ,The hardware structure of the replenishment demand analysis device may include: at least one processor 1, at least one communication interface 2, at least one memory 3, and at least one communication bus 4;

[0124] In the embodiment of the present application, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 communicate with each other through the communication bus 4;

[0125] The processor 1 may be a central processing unit CPU, or an application-specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention, etc.;

[0126] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), etc., such as at least one disk memory;

[0127] The memory stores a program, and the processor can call the program stored in the memory, wherein the program is used to:

[0128] Acquire initial images of the shelf area taken from multiple angles, and perform enhancement and denoising processing on the initial images of the shelf area to generate a shelf area detection image;

[0129] Inputting the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images labeled with product type, quantity and placement position;

[0130] Generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as basic data for demand forecasting;

[0131] Input the demand forecast basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecast result output by the sales demand model, wherein the sales demand model is trained based on historical sales data by combining a long short-term memory network LSTM with an attention mechanism;

[0132] The current replenishment demand is determined based on the current product data, the sales demand forecast result and the preset safety stock setting.

[0133] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0134] The embodiment of the present application further provides a readable storage medium, which may store a program suitable for execution by a processor, wherein the program is used to:

[0135] Acquire initial images of the shelf area taken from multiple angles, and perform enhancement and denoising processing on the initial images of the shelf area to generate a shelf area detection image;

[0136] Inputting the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images labeled with product type, quantity and placement position;

[0137] Generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as basic data for demand forecasting;

[0138] Input the demand forecast basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecast result output by the sales demand model, wherein the sales demand model is trained based on historical sales data by combining a long short-term memory network LSTM with an attention mechanism;

[0139] The current replenishment demand is determined based on the current product data, the sales demand forecast result and the preset safety stock setting.

[0140] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0141] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0142] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0143] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A replenishment demand analysis method, It is characterized in that include: Acquire initial images of the shelf area taken from multiple angles, and perform enhancement and denoising processing on the initial images of the shelf area to generate a shelf area detection image; Inputting the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images labeled with product type, quantity and placement position; Generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as basic data for demand forecasting; Input the demand forecast basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecast result output by the sales demand model, wherein the sales demand model is trained based on historical sales data by combining a long short-term memory network LSTM with an attention mechanism; The current replenishment demand is determined based on the current product data, the sales demand forecast result and the preset safety stock setting.

2. The method according to claim 1, It is characterized in that After obtaining the initial images of the shelf area taken from multiple angles, the following steps are also included: The initial image of the shelf area is subjected to image sharpening processing and light correction processing.

3. The method according to claim 1, It is characterized in that Also includes: Retrieve product sales data from associated stores; The current sales data and the product sales data of the associated stores are used as basic data for demand forecasting.

4. The method according to claim 1, It is characterized in that Also includes: Get a list of best-selling products across all channels; The current sales data and the omni-channel hot-selling product list are used as basic data for demand forecasting.

5. The method according to claim 4, It is characterized in that Also includes: Determine slow-moving goods based on the current sales data; According to the omni-channel hot-selling product list, determining replacement products belonging to the omni-channel hot-selling product list or the acquired new product list, and generating a replacement list recording the slow-moving products and the replacement products, wherein the replacement products are used to replace the slow-moving products; Extracting information about slow-moving goods from the current goods data and the current sales data, and designing a promotion strategy for the slow-moving goods based on the information about the slow-moving goods; Determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, including: The current replenishment demand is determined based on the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replacement list.

6. The method according to claim 1, It is characterized in that Also includes: Determine the replenishment cycle of each product according to the replenishment process corresponding to each product and the supplier response time; Determine the current replenishment demand according to the current product data, the sales demand forecast result and the preset safety stock setting, including: The current replenishment demand is determined according to the current product data and the sales demand forecast result, combined with the preset safety stock setting and the replenishment cycle of each product.

7. The method according to claim 1, It is characterized in that The attention mechanism sets the corresponding weights of each product based on the distribution of consumer groups, current temperature, current promotion strategy, and upcoming holidays.

8. A replenishment demand analysis device, It is characterized in that include: The area detection unit is used to obtain the initial image of the shelf area taken from multiple angles, and perform enhancement and denoising processing on the initial image of the shelf area to generate a shelf area detection image; A product recognition unit, used for inputting the shelf area detection image into a pre-trained product recognition model to obtain current product data output by the product recognition model, wherein the product recognition model is trained based on sample shelf area images marked with product type, quantity and placement position; A forecasting data unit, used to generate current sales data based on the retrieved current cash register system data and online sales platform data, and use the current sales data as demand forecasting basic data; A demand forecasting unit, used for inputting the demand forecasting basic data and the current product data into a pre-trained sales demand model to obtain a sales demand forecasting result output by the sales demand model, wherein the sales demand model is obtained by combining a long short-term memory network LSTM with an attention mechanism and is trained based on historical sales data; The demand determination unit is used to determine the current replenishment demand based on the current product data, the sales demand forecast result and the preset safety stock setting.

9. A replenishment demand analysis device, It is characterized in that including memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the replenishment demand analysis method according to any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, each step of the replenishment demand analysis method according to any one of claims 1 to 7 is implemented.

Images