A statistical method and system for agricultural trade retail industry based on big data

Abstract

The present invention relates to the field of big data statistics, and provides a statistical method and system for the agricultural and trade retail industry based on big data. By using an agricultural product identification model to identify the product types and sales volumes of different products, the limitation of using product barcodes is avoided, so that it can be applied to the retail scenarios of most farmers' markets or convenience markets, improving the comprehensiveness of statistics. At the same time, the traceability is ensured by generating sales records, and then the sales records are analyzed to obtain the commodity demands in different regions and different time periods, further improving the timeliness of statistics. It realizes feedback adjustment in two directions of space and time, and through market risk judgment, the influence of invalid data and misleading data in the analysis result data is avoided, considering the influence of comprehensive external market factors, improving the authenticity of statistics, so as to form accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the statistical method for the agricultural and trade retail industry.

Description

Technical Field

[0001] The present invention relates to the field of big data statistics, and particularly to a statistical method and system for the agricultural and trade retail industry based on big data. Background Art

[0002] With the rapid development of the agricultural and trade retail industry, higher requirements are put forward for the payment flow and product inflow and outflow in the retail process. In order to conduct market monitoring management and material logistics regulation reasonably and effectively, accurately counting the payment flow and product inflow and outflow is a crucial link.

[0003] In the prior art, it is usually based on scanning barcodes for warehousing and sales. Forming retail data from barcode scanning records is the most common method. However, in the agricultural and trade retail industry, due to many technical conditions and labor cost limitations, in most retail scenarios of farmers' markets or convenience markets, retail statistics cannot be carried out through barcodes, and only the overall market logistics inflow and outflow can be used for statistics. As a result, real-time retail data can only correspond to retail scenarios with barcode recording capabilities such as supermarkets, with a lot of one-sided and incomplete data, greatly increasing the statistical error. At the same time, due to the lag of overall statistics, statistics can only be carried out after a fixed sales cycle, and it is impossible to provide real-time feedback and timely regulation, resulting in the formation of invalid data and misleading data.

[0004] Therefore, how to design a statistical method for the agricultural and trade retail industry to avoid one-sided and lagged statistics and improve the accuracy and comprehensiveness of statistical data. Summary of the Invention

[0005] Based on this, a statistical method and system for the agricultural and trade retail industry based on big data provided by the present invention identify the product types and product sales volumes of different products through an agricultural and trade product identification model, avoiding the limitations of using product barcodes, so that it can be applied to most retail scenarios of farmers' markets or convenience markets, improving the comprehensiveness of statistics. At the same time, the traceability is ensured by generating sales records, and then the sales records are analyzed to obtain the commodity demands in different regions and different time periods, further improving the timeliness of statistics, realizing feedback regulation in two directions of space and time, and avoiding the influence of invalid data and misleading data in the analysis result data through market risk judgment, considering the influence of external comprehensive factors of the market, improving the authenticity of statistics, so as to form accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the statistical method for the agricultural and trade retail industry.

[0006] A statistical method for the agricultural and trade retail industry based on big data proposed by the present invention includes:

[0007] Obtain the product type and product sales volume of the target sales product according to the agricultural trade commodity recognition model, and generate a real-time sales record based on the product type and product sales volume. The agricultural trade commodity recognition model includes a cross-modal feature fusion module based on visible light features and near-infrared features and an overlapping occlusion separation module;

[0008] Conduct demand analysis based on the real-time sales record and historical sales record to respectively obtain the regional demand analysis result and the time period demand analysis result;

[0009] Conduct market risk judgment based on the regional demand analysis result and the time period demand analysis result to obtain a market risk judgment report. The market risk judgment includes product shortage risk judgment and product overstock risk judgment;

[0010] Conduct sales guidance based on the regional demand analysis result, the time period demand analysis result and the market risk judgment report. The sales guidance includes generating relevance suggestions based on the sales region and product type.

[0011] In summary, according to the above-mentioned agricultural trade retail industry statistical method based on big data, the product type and product sales volume of different products are identified through the agricultural trade commodity identification model, which avoids the limitation of using commodity barcodes, so that it can be applied in most farmer's markets or convenience markets. Retail scenarios, improve the comprehensiveness of statistics, and at the same time, by generating sales records to ensure traceability, and then analyzing the sales records to obtain commodity demand in different regions and different time periods, further improve the timeliness of statistics, and realize feedback adjustment from both spatial and temporal directions. Through market risk judgment, the influence of invalid data and misleading data in the analysis result data is avoided, and the influence of comprehensive external market factors is considered to improve the authenticity of statistics, thereby forming accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the agricultural trade retail industry statistical method. Specifically, the product type and product sales volume of the target sales product are obtained according to the agricultural trade commodity recognition model, and the product type and product sales volume are generated into a real-time sales record. The agricultural trade commodity recognition model includes a cross-modal feature fusion module based on visible light features and near-infrared features and an overlapping occlusion separation module, which avoids the limitations of generating and tracking sales records through barcodes in the prior art, and is widely used in most retail scenarios of agricultural markets or convenience markets, thereby improving the comprehensiveness of statistics. Demand analysis is performed according to the real-time sales records and historical sales records to respectively obtain regional demand analysis results and time period demand analysis results, and demand analysis is performed from different spatial and temporal orientations, thereby improving the accuracy of statistical analysis and ensuring data timeliness. Market risk judgment is performed based on the regional demand analysis results and the time period demand analysis results to obtain a market risk judgment report, the market risk judgment includes product shortage risk judgment and product backlog risk judgment, and the product purchase risk judgment is performed based on the influence of comprehensive external market factors to avoid product shortage and product backlog problems caused by comprehensive external market factors, thereby improving the authenticity of statistics and avoiding the influence of data illusions of invalid data and misleading data. Sales guidance is performed based on the regional demand analysis results, the time period demand analysis results and the market risk judgment report, and the sales guidance includes generating correlation suggestions based on sales areas and product types, thereby forming accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the statistical method of the agricultural trade retail industry.

[0012] Furthermore, the step of obtaining the product type and product sales volume of the target sales product according to the agricultural product identification model and generating a real-time sales record based on the product type and product sales volume specifically includes:

[0013] Collect visible light image data and near infrared image data of target sales products and perform pre-processing;

[0014] Input the preprocessed visible light image data and near-infrared image data into the agricultural product recognition model to extract visible light image features and near-infrared image features;

[0015] According to the overlapping occlusion separation algorithm, decouple and separate the visible light image features and near-infrared image features, respectively obtain the prediction masks of the visible light image features and near-infrared image features, and perform overlapping occlusion recognition according to the prediction masks to obtain the product sales volume;

[0016] Perform feature enhancement processing on the visible light image features and near-infrared image features according to the cross-modal feature fusion algorithm to identify and obtain the product type;

[0017] The specific cross-modal feature fusion algorithm is as follows:

[0018]

[0019] Among them, represents the cross-modal fusion feature, represents the activation function, and respectively represent the attention weights of the visible light image features and the attention weights of the near-infrared image features, and respectively represent the visible light image features and the near-infrared image features;

[0020] Generate real-time sales records according to the product type and product sales volume.

[0021] Further, the step of performing demand analysis according to the real-time sales record and the historical sales record specifically includes:

[0022] Retrieve the historical sales records in the preset sales record database according to the real-time sales record, and match the real-time sales record and the historical sales record according to the product type;

[0023] Obtain the location information of the real-time sales record and the historical sales record to generate a sales heat zone map of the target sales product. The sales heat zone map is divided into multiple different market areas according to the urban area, and the market area includes the heat value of the target sales product;

[0024] Judge whether the heat value of the target sales product is greater than or equal to the preset demand threshold within the preset sales cycle;

[0025] If it is determined that the heat value of the target sales product is greater than or equal to the preset demand threshold within the preset sales cycle, mark the target sales product as a region high-demand product in the sales heat zone map;

[0026] If it is determined that the popularity value of the target sales product is less than the preset demand threshold within the preset sales cycle, then mark the target sales product as a non-region high-demand product in the sales heat zone map;

[0027] Generate a regional demand analysis result based on the marked sales heat zone map.

[0028] Further, after the step of obtaining the positioning information of the real-time sales record and the historical sales record to generate the sales heat zone map of the target sales product, the following steps are also included:

[0029] Obtain the product sales amount of the same type of products in each market region in the sales heat zone map;

[0030] Sort the product sales amounts according to the order of the sales time periods in a single sales cycle;

[0031] Judge whether the product sales amount in each sales time period is greater than or equal to the preset product sales amount threshold;

[0032] If it is determined that the product sales amount in each sales time period is greater than or equal to the preset product sales amount threshold, then mark the sales time period as the high-demand sales time period of the current product type in the sales heat zone map;

[0033] If it is determined that the product sales amount in each sales time period is less than the preset product sales amount threshold, then mark the sales time period as the non-high-demand sales time period of the current product type in the sales heat zone map;

[0034] Generate a time period demand analysis result based on the marked sales heat zone map.

[0035] Further, the step of performing market risk judgment based on the regional demand analysis result and the time period demand analysis result to obtain a market risk judgment report specifically includes:

[0036] Obtain the regional demand analysis results and time period demand analysis results of the current sales cycle and multiple historical sales cycles;

[0037] Use the regional demand analysis result and the time period demand analysis result of the current sales cycle as the basic fluctuation parameters to calculate the basic coefficient of variation of the target sales product;

[0038] Then obtain the comprehensive time-varying factors of the current sales cycle, and perform time series decomposition according to the comprehensive time-varying factors to obtain the time series coefficient of variation. The comprehensive time-varying factors include seasonal factors, climate factors, and holiday factors;

[0039] Perform regression analysis based on the basic coefficient of variation and the time series coefficient of variation to obtain the sales volatility risk value;

[0040] Take the regional demand analysis results and time period demand analysis results of all historical sales cycles as the basic sales trend parameters, and calculate the market external correlation parameters of the target sales product;

[0041] Calculate the market dynamic correlation parameters according to the regional demand analysis results and time period demand analysis results of the current sales cycle;

[0042] Conduct multiple linear regression analysis based on the market external correlation parameters and market dynamic correlation parameters to obtain the market correlation risk value;

[0043] Generate a market risk judgment report according to the sales volatility risk value and market correlation risk value.

[0044] Further, the step of conducting sales guidance according to the regional demand analysis results, the time period demand analysis results and the market risk judgment report specifically includes:

[0045] Conduct product screening according to the market risk judgment report to obtain product types of different risk levels, and mark the product types with low risk levels as recommended sales product types;

[0046] Conduct regional screening in the sales hot zone map according to the regional demand analysis results to obtain multiple high-demand area combinations. The high-demand area combinations include multiple high-demand areas. The high-demand areas are market areas with area high-demand products in the sales hot zone map. Each high-demand area combination has only a single corresponding product type as the high-demand product type, and generate logistics allocation suggestions according to the high-demand area combinations;

[0047] Conduct replenishment sales suggestions according to the time period demand analysis results and the market risk judgment report;

[0048] Generate a sales guidance report according to the recommended sales product types, logistics allocation suggestions and replenishment sales suggestions.

[0049] Further, after the step of generating a sales guidance report according to the recommended sales product types, logistics allocation suggestions and replenishment sales suggestions, it further includes:

[0050] When an abnormal sales record is currently detected, conduct traceability management according to the merchant code and the incoming and outgoing goods vouchers;

[0051] The funds statistics module and the funds reconciliation module then conduct anomaly location according to the order list data to obtain the anomaly cause;

[0052] Conduct funds protection freezing and order suspension processing according to the anomaly cause, and send an anomaly warning text to the merchant terminal and the central control terminal.

[0053] The present invention proposes a statistical system for the agricultural trade retail industry based on big data, comprising:

[0054] An identification module, used to obtain the product type and product sales volume of the target sales product according to the agricultural product identification model, and generate real-time sales records for the product type and product sales volume, wherein the agricultural product identification model includes a cross-modal feature fusion module based on visible light features and near infrared features and an overlapping occlusion separation module;

[0055] A demand analysis module, used to perform demand analysis based on the real-time sales records and historical sales records to obtain regional demand analysis results and time period demand analysis results respectively;

[0056] A risk judgment module, used to make market risk judgments based on the regional demand analysis results and the period demand analysis results to obtain a market risk judgment report, wherein the market risk judgments include product shortage risk judgments and product backlog risk judgments;

[0057] A guidance and suggestion module is used to provide sales guidance based on the regional demand analysis results, the period demand analysis results and the market risk judgment report, wherein the sales guidance includes generating correlation suggestions based on sales areas and product types.

[0058] The present invention also provides a storage medium, which stores one or more programs, and when the programs are executed by a processor, they implement the above-mentioned agricultural trade retail industry statistical method based on big data.

[0059] The present invention also provides a computer device, the computer device comprising a memory and a processor, wherein:

[0060] The memory is used to store computer programs;

[0061] When the processor is used to execute the computer program stored in the memory, it implements the above-mentioned agricultural trade and retail industry statistical method based on big data. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of a statistical method for the agricultural trade retail industry based on big data proposed in the first embodiment of the present invention;

[0063] Figure 2 This is a flow chart of a statistical method for the agricultural trade retail industry based on big data proposed in the second embodiment of the present invention;

[0064] Figure 3 This is a schematic diagram of the structure of the agricultural trade retail industry statistics system based on big data proposed in the third embodiment of the present invention.

[0065] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. Detailed implementation manners

[0066] For the convenience of understanding the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0067] It should be noted that when an element is referred to as being "fixedly provided on" another element, it can be directly on the other element or there can also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration.

[0068] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0069] Please refer to Figure 1 , which shows a flowchart of a statistical method for the agricultural and trade retail industry based on big data proposed in the first embodiment of the present invention. This statistical method for the agricultural and trade retail industry based on big data includes steps S01 to S04, where:

[0070] Step S01: Obtain the product type and product sales volume of the target sales product according to the agricultural product identification model, and generate a real-time sales record from the product type and product sales volume;

[0071] It should be noted that in this embodiment, visible light image data and near-infrared image data of the target sales product are collected and preprocessed;

[0072] Input the preprocessed visible light image data and near-infrared image data into the agricultural product identification model to extract visible light image features and near-infrared image features;

[0073] Decouple and separate the visible light image features and near-infrared image features according to the overlapping occlusion separation algorithm, respectively obtain the prediction masks of the visible light image features and near-infrared image features, and perform overlapping occlusion recognition according to the prediction masks to obtain the product sales volume;

[0074] Perform feature enhancement processing on the visible light image features and near-infrared image features according to the cross-modal feature fusion algorithm to identify and obtain the product type;

[0075] The cross-modal feature fusion algorithm is specifically as follows:

[0076]

[0077] Among them, represents the cross-modal fusion feature, represents the activation function, and respectively represent the attention weights of visible light image features and near-infrared image features, and respectively represent the visible light image features and near-infrared image features;

[0078] Generate a real-time sales record according to the product type and product sales volume.

[0079] Step S02: Perform demand analysis based on the real-time sales record and historical sales record to respectively obtain the regional demand analysis result and the time period demand analysis result;

[0080] It should be noted that in this embodiment, the historical sales record in the preset sales record database is retrieved according to the real-time sales record, and the real-time sales record and historical sales record are matched according to the product type;

[0081] Obtain the positioning information of the real-time sales record and historical sales record to generate a sales hot zone map of the target sales product. The sales hot zone map is divided into multiple different market areas according to the urban area, and the market area includes the heat value of the target sales product;

[0082] Judge whether the heat value of the target sales product is greater than or equal to the preset demand threshold within the preset sales cycle;

[0083] If it is determined that the heat value of the target sales product is greater than or equal to the preset demand threshold within the preset sales cycle, mark the target sales product as a regional high-demand product in the sales hot zone map;

[0084] If it is determined that the heat value of the target sales product is less than the preset demand threshold within the preset sales cycle, mark the target sales product as a non-regional high-demand product in the sales hot zone map;

[0085] Generate a regional demand analysis result according to the marked sales hot zone map;

[0086] Obtain the product sales amount of the same type of product in each market area of the sales hot zone map;

[0087] Sort the product sales according to the order of the sales time periods in a single sales cycle;

[0088] Determine whether the product sales in each sales time period are greater than or equal to a preset product sales threshold;

[0089] If it is determined that the product sales in each sales time period are greater than or equal to the preset product sales threshold, mark the sales time period in the sales heat map as the high-demand sales time period for the current product type;

[0090] If it is determined that the product sales in each sales time period are less than the preset product sales threshold, mark the sales time period in the sales heat map as the non-high-demand sales time period for the current product type;

[0091] Generate a time period demand analysis result according to the marked sales heat map.

[0092] Step S03: Perform a market risk judgment based on the regional demand analysis result and the time period demand analysis result to obtain a market risk judgment report;

[0093] It should be noted that in this embodiment, the regional demand analysis results and the time period demand analysis results of the current sales cycle and multiple historical sales cycles are obtained;

[0094] Take the regional demand analysis result and the time period demand analysis result of the current sales cycle as basic fluctuation parameters, and calculate the basic coefficient of variation of the target sales product;

[0095] Then obtain the comprehensive time-varying factors of the current sales cycle, and perform time series decomposition according to the comprehensive time-varying factors to obtain the time series coefficient of variation. The comprehensive time-varying factors include seasonal factors, climate factors, and holiday factors;

[0096] Perform regression analysis according to the basic coefficient of variation and the time series coefficient of variation to obtain the sales volatility risk value;

[0097] Take the regional demand analysis results and the time period demand analysis results of all historical sales cycles as basic sales trend parameters, and calculate the market external correlation parameter of the target sales product;

[0098] Calculate the market dynamic correlation parameter according to the regional demand analysis result and the time period demand analysis result of the current sales cycle;

[0099] Perform multiple linear regression analysis according to the market external correlation parameter and the market dynamic correlation parameter to obtain the market correlation risk value;

[0100] Generate a market risk judgment report according to the sales volatility risk value and the market correlation risk value.

[0101] Step S04: Provide sales guidance based on the results of regional demand analysis, time period demand analysis, and market risk judgment report;

[0102] It should be noted that in this embodiment, product screening is performed according to the market risk judgment report to obtain product types with different risk levels, and the product types with low risk levels are marked as recommended sales product types;

[0103] Perform regional screening in the sales hot zone map according to the results of regional demand analysis to obtain multiple high-demand area combinations. Each high-demand area combination includes multiple high-demand areas. A high-demand area is a market area where there are area high-demand products in the sales hot zone map. Each high-demand area combination has only a single corresponding product type as the high-demand product type, and generate logistics allocation suggestions according to the high-demand area combination;

[0104] Provide replenishment sales suggestions according to the results of time period demand analysis and market risk judgment report;

[0105] Generate a sales guidance report according to the recommended sales product types, logistics allocation suggestions, and replenishment sales suggestions;

[0106] When an abnormal sales record is currently detected, perform traceability management according to the merchant code and incoming and outgoing goods vouchers;

[0107] The fund statistics module and the fund reconciliation module then perform anomaly location based on the order list data to obtain the cause of the anomaly;

[0108] Perform fund protection freezing and order suspension processing according to the cause of the anomaly, and send an anomaly warning text to the merchant terminal and the central control terminal.

[0109] In summary, according to the above-mentioned agricultural trade retail industry statistical method based on big data, the product type and product sales volume of different products are identified through the agricultural trade commodity identification model, which avoids the limitation of using commodity barcodes, so that it can be applied in most farmer's markets or convenience markets. Retail scenarios, improve the comprehensiveness of statistics, and at the same time, by generating sales records to ensure traceability, and then analyzing the sales records to obtain commodity demand in different regions and different time periods, further improve the timeliness of statistics, and realize feedback adjustment from both spatial and temporal directions. Through market risk judgment, the influence of invalid data and misleading data in the analysis result data is avoided, and the influence of comprehensive external market factors is considered to improve the authenticity of statistics, thereby forming accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the agricultural trade retail industry statistical method. Specifically, the product type and product sales volume of the target sales product are obtained according to the agricultural trade commodity recognition model, and the product type and product sales volume are generated into a real-time sales record. The agricultural trade commodity recognition model includes a cross-modal feature fusion module based on visible light features and near-infrared features and an overlapping occlusion separation module, which avoids the limitations of generating and tracking sales records through barcodes in the prior art, and is widely used in most retail scenarios of agricultural markets or convenience markets, thereby improving the comprehensiveness of statistics. Demand analysis is performed according to the real-time sales records and historical sales records to respectively obtain regional demand analysis results and time period demand analysis results, and demand analysis is performed from different spatial and temporal orientations, thereby improving the accuracy of statistical analysis and ensuring data timeliness. Market risk judgment is performed based on the regional demand analysis results and the time period demand analysis results to obtain a market risk judgment report, the market risk judgment includes product shortage risk judgment and product backlog risk judgment, and the product purchase risk judgment is performed based on the influence of comprehensive external market factors to avoid product shortage and product backlog problems caused by comprehensive external market factors, thereby improving the authenticity of statistics and avoiding the influence of data illusions of invalid data and misleading data. Sales guidance is performed based on the regional demand analysis results, the time period demand analysis results and the market risk judgment report, and the sales guidance includes generating correlation suggestions based on sales areas and product types, thereby forming accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the statistical method of the agricultural trade retail industry.

[0110] See also Figure 2 , which is a flow chart of a statistical method for agricultural trade retail industry based on big data proposed in the second embodiment of the present invention, and the statistical method for agricultural trade retail industry based on big data includes steps S11 to S16, wherein:

[0111] Step S11: Collect the visible light image data and near-infrared image data of the target sales product and perform preprocessing. Input the preprocessed visible light image data and near-infrared image data into the agricultural product recognition model to extract visible light image features and near-infrared image features. Decouple and separate the visible light image features and near-infrared image features according to the overlapping occlusion separation algorithm, respectively obtain the prediction masks of the visible light image features and near-infrared image features, perform overlapping occlusion recognition based on the prediction masks to obtain the product sales volume, perform feature enhancement processing on the visible light image features and near-infrared image features according to the cross-modal feature fusion algorithm to identify and obtain the product type, and generate real-time sales records according to the product type and product sales volume;

[0112] It should be noted that in this embodiment, the agricultural product recognition model is used to identify the product type and product sales volume, and includes an overlapping occlusion separation module and a cross-modal feature fusion module. The overlapping occlusion separation module separates the overlapping target sales products by respectively obtaining the prediction masks of the visible light image features and near-infrared image features, and realizes the identification of the product sales volume. The cross-modal feature fusion algorithm adjusts the attention weights of the visible light image features and near-infrared image features to perform feature enhancement fusion and improve the accuracy of product type identification.

[0113] Step S12: Retrieve the historical sales records in the preset sales record database according to the real-time sales records, match the real-time sales records and historical sales records according to the product type to obtain the positioning information of the real-time sales records and historical sales records, generate a sales heat map of the target sales product, and determine whether the heat value of the target sales product is greater than or equal to the preset demand threshold within the preset sales cycle. If it is determined that the heat value of the target sales product is greater than or equal to the preset demand threshold within the preset sales cycle, mark the target sales product as a high-demand product in the sales heat map. If it is determined that the heat value of the target sales product is less than the preset demand threshold within the preset sales cycle, mark the target sales product as a non-high-demand product in the sales heat map, and generate a regional demand analysis result according to the marked sales heat map;

[0114] It should be noted that in this embodiment, the sales heat map is divided into multiple different market areas according to the urban area. The market area includes the heat value of the target sales product. The preset sales record database includes all sales records of different sales areas and different sales cycles. The sales records include sales time points, product types, product sales volumes, product sales amounts, and merchant information. In this embodiment, the preset sales cycle is the total business hours of the market, and the preset demand threshold is the average weekly sales volume of the target sales product.

[0115] Step S13: Obtain the product sales amounts of the same type of products in each market area in the sales hot zone map, sort the product sales amounts according to the order of the sales time periods in a single sales cycle, and determine whether the product sales amount in each sales time period is greater than or equal to the preset product sales amount threshold. If it is determined that the product sales amount in each sales time period is greater than or equal to the preset product sales amount threshold, then mark the sales time period in the sales hot zone map as the high-demand sales time period of the current product type. If it is determined that the product sales amount in each sales time period is less than the preset product sales amount threshold, then mark the sales time period in the sales hot zone map as the non-high-demand sales time period of the current product type, and generate a time period demand analysis result based on the marked sales hot zone map;

[0116] It should be noted that in this embodiment, the preset product sales amount threshold is the average weekly sales amount of the target sales product.

[0117] Step S14: Obtain the regional demand analysis results and time period demand analysis results of the current sales cycle and multiple historical sales cycles. Use the regional demand analysis results and time period demand analysis results of the current sales cycle as the basic fluctuation parameters to calculate the basic coefficient of dispersion of the target sales product. Then obtain the comprehensive time-varying factors of the current sales cycle, perform time series decomposition according to the comprehensive time-varying factors to obtain the time series coefficient of dispersion, perform regression analysis based on the basic coefficient of dispersion and the time series coefficient of dispersion to obtain the sales volatility risk value. Use the regional demand analysis results and time period demand analysis results of all historical sales cycles as the basic sales trend parameters to calculate the market external correlation parameter of the target sales product. Calculate the market dynamic correlation parameter according to the regional demand analysis results and time period demand analysis results of the current sales cycle. Perform multiple linear regression analysis based on the market external correlation parameter and the market dynamic correlation parameter to obtain the market correlation risk value, and generate a market risk judgment report based on the sales volatility risk value and the market correlation risk value;

[0118] It should be noted that in this embodiment, the comprehensive time-varying factors include seasonal factors, climate factors, and holiday factors.

[0119] Step S15: Perform product screening according to the market risk judgment report to obtain product types with different risk levels, mark the low-risk level product types as the recommended sales product types, perform regional screening in the sales hot zone map according to the regional demand analysis results to obtain multiple high-demand area combinations, generate logistics allocation suggestions based on the high-demand area combinations, generate replenishment sales suggestions according to the time period demand analysis results and the market risk judgment report, and generate a sales guidance report based on the recommended sales product types, logistics allocation suggestions, and replenishment sales suggestions;

[0120] It should be noted that in this embodiment, the high-demand area combination includes multiple high-demand areas. The high-demand area is the market area where high-demand products exist in the sales heat map. Each high-demand area combination has only a single corresponding product type as the high-demand product type.

[0121] Step S16: When an abnormal sales record is currently detected, traceability management is performed according to the merchant code and the incoming and outgoing goods vouchers. The funds statistics module and the funds reconciliation module then perform abnormal positioning based on the order list data to obtain the cause of the abnormality, and perform funds protection freezing and order suspension processing according to the cause of the abnormality, and send an abnormal alarm text to the merchant terminal and the central control terminal.

[0122] In summary, according to the above-mentioned agricultural trade retail industry statistical method based on big data, the product type and product sales volume of different products are identified through the agricultural trade commodity identification model, which avoids the limitation of using commodity barcodes, so that it can be applied in most farmer's markets or convenience markets. Retail scenarios, improve the comprehensiveness of statistics, and at the same time, by generating sales records to ensure traceability, and then analyzing the sales records to obtain commodity demand in different regions and different time periods, further improve the timeliness of statistics, and realize feedback adjustment from both spatial and temporal directions. Through market risk judgment, the influence of invalid data and misleading data in the analysis result data is avoided, and the influence of comprehensive external market factors is considered to improve the authenticity of statistics, thereby forming accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the agricultural trade retail industry statistical method. Specifically, the product type and product sales volume of the target sales product are obtained according to the agricultural trade commodity recognition model, and the product type and product sales volume are generated into a real-time sales record. The agricultural trade commodity recognition model includes a cross-modal feature fusion module based on visible light features and near-infrared features and an overlapping occlusion separation module, which avoids the limitations of generating and tracking sales records through barcodes in the prior art, and is widely used in most retail scenarios of agricultural markets or convenience markets, thereby improving the comprehensiveness of statistics. Demand analysis is performed according to the real-time sales records and historical sales records to respectively obtain regional demand analysis results and time period demand analysis results, and demand analysis is performed from different spatial and temporal orientations, thereby improving the accuracy of statistical analysis and ensuring data timeliness. Market risk judgment is performed based on the regional demand analysis results and the time period demand analysis results to obtain a market risk judgment report, the market risk judgment includes product shortage risk judgment and product backlog risk judgment, and the product purchase risk judgment is performed based on the influence of comprehensive external market factors to avoid product shortage and product backlog problems caused by comprehensive external market factors, thereby improving the authenticity of statistics and avoiding the influence of data illusions of invalid data and misleading data. Sales guidance is performed based on the regional demand analysis results, the time period demand analysis results and the market risk judgment report, and the sales guidance includes generating correlation suggestions based on sales areas and product types, thereby forming accurate and comprehensive sales guidance. The present invention improves the comprehensiveness and accuracy of the statistical method of the agricultural trade retail industry.

[0123] See also Figure 3 , which is a schematic diagram of the structure of a statistical system for agricultural trade retail industry based on big data proposed in the third embodiment of the present invention, and the system includes:

[0124] An identification module 10, configured to obtain the product type and product sales volume of a target sales product according to an agricultural product identification model, and generate a real-time sales record based on the product type and product sales volume. The agricultural product identification model includes a cross-modal feature fusion module and an overlapping occlusion separation module based on visible light features and near-infrared features;

[0125] A demand analysis module 20, configured to perform demand analysis according to the real-time sales record and historical sales record to respectively obtain a regional demand analysis result and a time period demand analysis result;

[0126] A risk judgment module 30, configured to perform market risk judgment according to the regional demand analysis result and the time period demand analysis result to obtain a market risk judgment report. The market risk judgment includes product shortage risk judgment and product overstock risk judgment;

[0127] A guidance and suggestion module 40, configured to perform sales guidance according to the regional demand analysis result, the time period demand analysis result, and the market risk judgment report. The sales guidance includes generating a relevance suggestion based on the sales region and product type.

[0128] The present invention also provides a computer storage medium, on which one or more programs are stored. When the program is executed by a processor, the above-mentioned statistical method for the agricultural retail industry based on big data is implemented.

[0129] The present invention also provides a computer device, including a memory and a processor. The memory is used to store a computer program, and the processor is used to execute the computer program stored on the memory to implement the above-mentioned statistical method for the agricultural retail industry based on big data.

[0130] Those skilled in the art can understand that the logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0131] More specific examples (a non-exhaustive list) of computer-readable media include the following: electrical connections (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0132] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0133] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0134] The above-described embodiments merely represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. A statistical method for agricultural trade retail industry based on big data, characterized in that: include: Acquire the product type and product sales volume of the target sales product according to the agricultural product recognition model, and generate real-time sales records for the product type and product sales volume, wherein the agricultural product recognition model includes a cross-modal feature fusion module based on visible light features and near infrared features and an overlapping occlusion separation module; The step of obtaining the product type and product sales volume of the target sales product according to the agricultural product identification model and generating a real-time sales record based on the product type and product sales volume specifically includes: Collect visible light image data and near infrared image data of target sales products and perform pre-processing; Inputting the preprocessed visible light image data and near infrared image data into an agricultural commodity recognition model to extract visible light image features and near infrared image features; Decoupling and separating the visible light image features and the near infrared image features according to an overlapping occlusion separation algorithm, respectively obtaining prediction masks of the visible light image features and the near infrared image features, and performing overlapping occlusion recognition according to the prediction masks to obtain product sales volume; Performing feature enhancement processing on the visible light image features and the near infrared image features according to a cross-modal feature fusion algorithm to identify and acquire the product type; The cross-modal feature fusion algorithm is specifically as follows: in, represents the cross-modal fusion feature, represents the activation function, and They represent the feature attention weights of visible light images and near infrared images, respectively. and Represent visible light image features and near infrared image features respectively; Generate real-time sales records based on the product type and product sales volume; Performing demand analysis based on the real-time sales records and historical sales records to obtain regional demand analysis results and time period demand analysis results respectively; Conduct market risk assessment based on the regional demand analysis results and the period demand analysis results to obtain a market risk assessment report, wherein the market risk assessment includes product shortage risk assessment and product backlog risk assessment; Sales guidance is provided based on the regional demand analysis results, the period demand analysis results and the market risk assessment report, wherein the sales guidance includes generating correlation suggestions based on sales regions and product types.

2. The statistical method for agricultural trade retail industry based on big data according to claim 1 is characterized in that: The step of performing demand analysis based on the real-time sales records and historical sales records specifically includes: Retrieve historical sales records from a preset sales record database based on real-time sales records, and match real-time sales records with historical sales records based on product types; Acquire location information of real-time sales records and historical sales records to generate a sales hotspot map of a target sales product, wherein the sales hotspot map is divided into a plurality of different market areas according to urban areas, and the market areas include heat values ​​of the target sales product; Determine whether the popularity value of the target sales product is greater than or equal to a preset demand threshold within a preset sales cycle; If it is determined that the popularity value of the target sales product is greater than or equal to a preset demand threshold within a preset sales cycle, the target sales product is marked as a regional high-demand product in the sales hot zone map; If it is determined that the heat value of the target sales product is less than a preset demand threshold within a preset sales cycle, the target sales product is marked as a non-regional high-demand product in the sales heat map; Generate regional demand analysis results based on the marked sales hotspot map.

3. The statistical method for agricultural trade retail industry based on big data according to claim 2 is characterized in that: The step of obtaining the location information of the real-time sales records and the historical sales records to generate a sales hotspot map of the target sales product further includes: Get the sales volume of the same type of products in each market area in the sales hotspot map; sorting the sales of the products according to the order of the sales time periods in a single sales cycle; Determine whether the product sales in each sales period is greater than or equal to the preset product sales threshold; If it is determined that the product sales volume in each sales time period is greater than or equal to a preset product sales volume threshold, the sales time period is marked as a high-demand sales time period for the current product type in the sales hotspot map; If it is determined that the product sales in each sales time period is less than a preset product sales threshold, the sales time period is marked as a sales non-high-demand time period for the current product type in the sales heat map; The time period demand analysis result is generated according to the marked sales hot zone map.

4. The statistical method for agricultural trade retail industry based on big data according to claim 1 is characterized in that: The step of performing market risk judgment according to the regional demand analysis results and the period demand analysis results to obtain a market risk judgment report specifically includes: Obtain regional demand analysis results and time period demand analysis results for the current sales cycle and multiple historical sales cycles; The regional demand analysis results and the period demand analysis results of the current sales cycle are used as basic fluctuation parameters to calculate the basic dispersion coefficient of the target sales product; Then, the comprehensive time-varying factors of the current sales cycle are obtained, and the time series is decomposed according to the comprehensive time-varying factors to obtain the time series dispersion coefficient, wherein the comprehensive time-varying factors include seasonal factors, climate factors and holiday factors; Performing regression analysis based on the basic dispersion coefficient and the time series dispersion coefficient to obtain the sales volatility risk value; The regional demand analysis results and time period demand analysis results of all historical sales cycles are used as basic sales trend parameters to calculate the market external correlation parameters of the target sales products; Calculate the market dynamic correlation parameters based on the regional demand analysis results and the period demand analysis results of the current sales cycle; Performing a multiple linear regression analysis based on the market external correlation parameters and the market dynamic correlation parameters to obtain a market correlation risk value; A market risk assessment report is generated based on the sales volatility risk value and the market correlation risk value.

5. The statistical method for agricultural trade retail industry based on big data according to claim 1 is characterized in that: The step of providing sales guidance based on the regional demand analysis results, the time period demand analysis results and the market risk judgment report specifically includes: Screen products based on the market risk assessment report to obtain product types at different risk levels, and mark low-risk product types as recommended sales product types; Performing regional screening in the sales hot zone map according to the regional demand analysis results to obtain multiple high-demand regional combinations, wherein the high-demand regional combinations include multiple high-demand regions, and the high-demand regions are market regions where there are regional high-demand products in the sales hot zone map, and each of the high-demand regional combinations has only a single corresponding product type as a high-demand product type, and generating logistics allocation suggestions according to the high-demand regional combinations; Provide replenishment sales recommendations based on time period demand analysis results and market risk assessment reports; A sales guidance report is generated based on the recommended sales product types, logistics allocation suggestions, and replenishment sales suggestions.

6. The statistical method for agricultural trade retail industry based on big data according to claim 5 is characterized in that: The step of generating a sales guidance report according to the recommended sales product type, logistics allocation suggestion and replenishment sales suggestion further includes: When abnormal sales records are currently detected, traceability management is performed based on the merchant code and inbound and outbound shipment vouchers; The fund statistics module and fund reconciliation module then locate the anomaly based on the order list data to obtain the cause of the anomaly; Funds protection freezing and order suspension processing are performed according to the abnormal reasons, and abnormal alarm texts are sent to the merchant terminal and the central control terminal.

7. A statistical system for agricultural trade retail industry based on big data, characterized in that: include: An identification module, used to obtain the product type and product sales volume of the target sales product according to the agricultural product identification model, and generate real-time sales records for the product type and product sales volume, wherein the agricultural product identification model includes a cross-modal feature fusion module based on visible light features and near infrared features and an overlapping occlusion separation module; The step of obtaining the product type and product sales volume of the target sales product according to the agricultural product identification model and generating a real-time sales record based on the product type and product sales volume specifically includes: Collect visible light image data and near infrared image data of target sales products and perform pre-processing; Inputting the preprocessed visible light image data and near infrared image data into an agricultural commodity recognition model to extract visible light image features and near infrared image features; Decoupling and separating the visible light image features and the near infrared image features according to an overlapping occlusion separation algorithm, respectively obtaining prediction masks of the visible light image features and the near infrared image features, and performing overlapping occlusion recognition according to the prediction masks to obtain product sales volume; Performing feature enhancement processing on the visible light image features and the near infrared image features according to a cross-modal feature fusion algorithm to identify and acquire the product type; The cross-modal feature fusion algorithm is specifically as follows: in, represents the cross-modal fusion feature, represents the activation function, and They represent the feature attention weights of visible light images and near infrared images, respectively. and Represent visible light image features and near infrared image features respectively; Generate real-time sales records based on the product type and product sales volume; A demand analysis module, used to perform demand analysis based on the real-time sales records and historical sales records to obtain regional demand analysis results and time period demand analysis results respectively; A risk judgment module, used to make market risk judgments based on the regional demand analysis results and the period demand analysis results to obtain a market risk judgment report, wherein the market risk judgments include product shortage risk judgments and product backlog risk judgments; A guidance and suggestion module is used to provide sales guidance based on the regional demand analysis results, the period demand analysis results and the market risk judgment report, wherein the sales guidance includes generating correlation suggestions based on sales areas and product types.

8. A storage medium, characterized in that: The storage medium stores one or more programs, which, when executed by the processor, implement the agricultural trade and retail industry statistical method based on big data as described in any one of claims 1 to 6.

9. A computer device, characterized in that: The computer device comprises a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, it implements the agricultural trade retail industry statistical method based on big data as described in any one of claims 1-6.

Images