Intelligent shelf control method, system and medium

Abstract

The application discloses a kind of intelligent shelf control method, system and medium, it is related to shelf field, the existing intelligent shelf control method is solved and the problem of poor control effect, including steps S1: setting commodity specification measurement area, the feasibility of each to-be-placed commodity is judged by commodity specification measurement area, obtain shelf commodity specification judgment data, step S2: in intelligent shelf division several commodity placement areas, and respectively to the placed goods of commodity placement area corresponding is identified, obtain shelf commodity partition data, step S3: shelf commodity in different commodity placement area is monitored in time period goods, and shelf commodity monitoring data is obtained by analyzing monitoring result, step S4: the placed goods of shelf area is evaluated, and according to the evaluation result, sales state feedback is carried out, the application can improve the control accuracy of intelligent shelf, and functionally expand intelligent shelf.

Description

Technical Field

[0001] This invention belongs to the field of shelving and relates to intelligent control technology, specifically an intelligent shelving control method, system, and medium. Background Technology

[0002] Existing intelligent shelf control methods have the following specific drawbacks when controlling the shelves: 1. Existing shelf control methods cannot automatically obtain the volume of the goods to be placed, nor can they determine the feasibility of placing the goods, resulting in poor shelf control over the placement of goods; 2. Existing shelf control methods cannot collect and analyze customers' body movements in front of smart shelves, nor can they combine the analysis results with the sales volume of the goods displayed on the shelves, thus making it impossible to judge the sales status of the goods.

[0003] Therefore, we propose an intelligent shelf control method, system, and medium. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a smart shelf control method, system and medium. This invention aims to improve the control efficiency of smart shelves and expand their functionality.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent shelf control method, comprising the following specific steps: Step S1: Set up a product specification measurement area, perform projection analysis on each product to be placed through the product specification measurement area, obtain the product placement volume value corresponding to each product to be placed based on the analysis results, obtain the real-time placement volume limit, compare the real-time placement volume limit with the product placement volume value, and obtain shelf product specification judgment data based on the comparison results. Step S2: Divide the smart shelf into several product placement areas, and identify the products placed in each product placement area. Based on the identification results, obtain the products placed in multiple areas to obtain shelf product partitioning data. Step S3: Based on the shelf merchandise zoning data, monitor the shelf merchandise in different merchandise placement areas in a time-based manner, and obtain shelf merchandise monitoring data by analyzing the monitoring results; Step S4: Evaluate the sales status of the products placed in each shelf area based on the shelf product monitoring data and shelf product zoning data, and provide feedback on the sales status based on the evaluation results.

[0006] Furthermore, step S1 also includes the following specific steps: Step S11: Set a rectangular area through the bottom ground area of ​​the smart shelf, and any side of the rectangular area coincides with the bottom edge of the smart shelf. Name the side of the rectangular area that coincides with the bottom edge of the smart shelf as the coincident feature side, and name the set rectangular area as the product specification measurement area. Step S12: Perform projection analysis on the product specification measurement area to obtain the first projection acquisition image and the second projection acquisition image; Step S13: Mark the edges of the projection area occupied by the sample product to be placed in the first projection acquisition image to obtain the projection rectangle corresponding to the sample product to be placed, and name it the first projection rectangle. Mark the edges of the projection area occupied by the sample product to be placed in the second projection acquisition image to obtain the projection rectangle corresponding to the sample product to be placed, and name it the second projection rectangle. Step S14: Select any two perpendicular rectangular sides in the first projection rectangle and name them as the first product volume side and the second product volume side, respectively. Obtain the length values ​​of the first product volume side and the second product volume side, respectively. Step S15: Perform side length analysis on the first projected rectangle to obtain the side length values ​​of the third volume; Step S16: Calculate the product placement volume corresponding to the sample product by taking the first volume side length value, the second volume side length value, and the third volume side length value; The volume of the sample items to be placed is calculated using the following formula: ; Where Vbf is the product placement volume corresponding to the sample product to be placed, Tjb1 is the first volume side length value, Tjb2 is the second volume side length value, and Tjb3 is the third volume side length value. Step S17: Determine the feasibility of placing the sample items based on the corresponding item placement volume values. Step S18: Obtain the volume values ​​corresponding to the products to be placed, and make a feasibility judgment on the placement of each product based on the obtained results to obtain shelf product specification judgment data. Step S17 further includes the following specific steps: Step S171: Obtain the empty placement volume of each empty area of ​​the smart shelf, compare the values ​​of the obtained empty placement volumes of multiple areas, mark the empty placement volume of the area with the largest value as the real-time placement volume limit, and compare the value of the product placement volume corresponding to the sample product to be placed with the real-time placement volume limit. Step S172: If the product placement volume value corresponding to the sample product to be placed is greater than or equal to the real-time placement volume limit, then mark the sample product to be placed as an unplaceable product. Step S173: If the product placement volume value corresponding to the sample product to be placed is less than the real-time placement volume limit, then mark the sample product to be placed as a placeable product.

[0007] Furthermore, step S12 also includes the following specific steps: Step S121: Obtain the diagonal intersection of the rectangle corresponding to the product specification measurement area to obtain the first region feature point. Draw a straight line perpendicular to the product specification measurement area through the first region feature point to obtain the first region feature line. Draw a plane parallel to the product specification measurement area through the first region feature line to obtain the first light projection plane. Mark the plane opposite to the first light projection plane as the first projection plane. Step S122: Draw a plane perpendicular to the product specification measurement area through the overlapping feature edge to obtain the second ray projection plane. Set a projection imaging surface in the spatial area opposite to the first ray projection plane and name it the second projection plane. Step S123: Obtain several items to be placed, and select one sample item from the obtained items to be placed; Step S124: Place a single sample of the product to be placed in the product specification measurement area, emit a parallel projection beam from the first light projection plane to the first projection plane, and acquire the projection image of the sample of the product to be placed on the first projection plane using an image acquisition device to obtain a first projection acquisition image. Then, emit a parallel projection beam from the second light projection plane to the second projection plane, and acquire the projection image of the sample of the product to be placed on the second projection plane using an image acquisition device to obtain a second projection acquisition image.

[0008] Furthermore, step S15 also includes the following specific steps: Step S151: Select any two perpendicular rectangular sides in the second projection rectangle and name them as the first undetermined volume side and the second undetermined volume side. Obtain the length values ​​of the first undetermined volume side and the second undetermined volume side respectively. Step S152: If the length of the first undetermined side is equal to the length of the second undetermined side, then mark the length of the first undetermined side as the length of the third volume side. Step S153: If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the first undetermined side is equal to the length of the first volume side or the length of the second volume side, then mark the length of the first undetermined side as the length of the third volume side. Step S154: If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the second undetermined side is equal to the length of the first volume side or the length of the second volume side, then mark the length of the second undetermined side as the length of the third volume side.

[0009] Furthermore, step S2 also includes the following specific steps: Step S21: Obtain shelf product specification judgment data, and obtain each placeable product based on the shelf product specification judgment data; Step S22: Divide the smart shelf into several product placement areas and label the divided product placement areas as B1 product placement area to Ba product placement area respectively; Step S23: Place each placeable item in the B1 item placement area to the Ba item placement area, and name the items placed in the B1 item placement area to the Ba item placement area as B1 shelf items to Ba shelf items respectively. Step S24: Define the product placement area from B1 to Ba as the product placement area and the product on shelf B1 to Ba as shelf product partition data.

[0010] Furthermore, step S3 also includes the following specific steps: Step S31: Obtain shelf product partitioning data, and obtain the product placement areas from B1 to Ba and the shelf products from B1 to Ba based on the shelf product partitioning data; Step S32: In the historical period when the smart shelf sells goods from shelf B1 to shelf Ba, select several historical sales sub-periods of equal duration, and mark the selected historical sales sub-periods in chronological order as L1 historical sales sub-period to Lb historical sales sub-period. Step S33: Obtain the shelf video monitoring of the B1 product placement area during the historical sales sub-period of L1 to obtain the L1B1 monitoring video. Based on the L1B1 monitoring video, perform image monitoring on the B1 shelf products in the historical sales sub-period of L1. Based on the monitoring results, obtain the L1 time period product popularity coefficient corresponding to the B1 shelf products. Step S34: Obtain the product popularity coefficients for the L2 historical sales sub-period to the Lb historical sales sub-period corresponding to the products on shelf B1, and obtain the product popularity coefficients for the L2 period to the Lb period. Step S35: Calculate the average of the product popularity coefficients from time period L1 to time period Lb to obtain the product popularity evaluation coefficient for B1. Step S36: Obtain the product popularity evaluation coefficients for products on shelves B2 to Ba respectively, to obtain the product popularity evaluation coefficients for products B2 to Ba. Step S37: Define the product popularity evaluation coefficients from B1 to Ba and the historical sales sub-periods from L1 to Lb as shelf product monitoring data.

[0011] Furthermore, step S33 also includes the following specific steps: Step S331: Identify the customers appearing in the L1B1 surveillance video using an image recognition algorithm, and mark the identified customers sequentially as G1 surveillance customers to Gc surveillance customers according to the time sequence. Step S332: In the L1B1 surveillance video, extract video images from the time periods when the G1 monitoring customer to the Gc monitoring customer appears in the video to obtain the G1 customer monitoring image to the Gc customer monitoring image. Step S333: Perform image analysis on the monitoring image of customer G1, and obtain the product intention distance duration ratio and hand trajectory movement complexity of the monitored customer G1 based on the analysis results, and name them as G1 intention distance duration ratio and G1 hand trajectory complexity. Step S334: Obtain the G2 intention distance duration ratio to the Gc intention distance duration ratio and the G2 hand trajectory complexity to the Gc hand trajectory complexity respectively; Step S335: Calculate the product popularity coefficient for the L1 time period corresponding to the product on shelf B1 by combining the G1 intention distance duration ratio to the Gc intention distance duration ratio and the G1 hand trajectory complexity to the Gc hand trajectory complexity. The popularity coefficient of the products in time period L1 corresponding to the products on shelf B1 is calculated using the following formula: ; Where Rdl1 is the product popularity coefficient of L1 time period corresponding to the product on shelf B1, Yjli is the Gi intention distance duration ratio, Sfli is the Gi hand trajectory complexity, and c is the number of monitored customers.

[0012] Furthermore, step S4 also includes the following specific steps: Step S41: Obtain shelf merchandise monitoring data, and obtain the B1 merchandise popularity evaluation coefficient to the Ba merchandise popularity evaluation coefficient and the L1 historical sales sub-period to the Lb historical sales sub-period based on the shelf merchandise monitoring data; Step S42: Obtain shelf product partitioning data, and obtain the product placement areas from B1 to Ba and the shelf products from B1 to Ba based on the shelf product partitioning data; Step S43: Obtain monitoring images of the B1 product placement area from the L1 historical sales sub-period to the Lb historical sales sub-period. Based on the monitoring images, obtain the sales quantity of B1 shelf products from the L1 historical sales sub-period to the Lb historical sales sub-period, and calculate the average to obtain the sales quantity of B1 products during the time period. Step S44: Obtain the sales quantity of the products on shelf B2 to shelf Ba for each time period, and obtain the sales quantity of products on shelf B2 to shelf Ba for each time period; Step S45: Obtain shelf merchandise monitoring data, and obtain the popularity evaluation coefficients of merchandise B1 to Ba respectively based on the shelf merchandise monitoring data; Step S46: Calculate the ratio of the sales volume of product B1 during the time period to the popularity evaluation coefficient of product B1 to obtain the sales popularity ratio of product B1. Calculate the ratio of the sales volume of product B2 during the time period to the popularity evaluation coefficient of product B3 to obtain the sales popularity ratio of product B2. And so on, calculate the ratio of the sales volume of product Ba during the time period to the popularity evaluation coefficient of product Ba to obtain the sales popularity ratio of product Ba. Step S47: Create a product sales popularity histogram based on the sales popularity ratio of product B1 to product Ba, and set the sales popularity baseline in the product sales popularity histogram. Step S49: If the bar corresponding to a product on the shelf is lower than the sales popularity baseline, the smart shelf will assess the sales status of the corresponding product as abnormal and issue an abnormality warning. If the bar corresponding to a product on the shelf is higher than or equal to the sales popularity baseline, the smart shelf will assess the sales status of the corresponding product as normal.

[0013] An intelligent shelf control system includes: Product placement module: Set up a product specification measurement area, perform projection analysis on each product to be placed through the product specification measurement area, obtain the product placement volume value corresponding to each product to be placed based on the analysis results, obtain the real-time placement volume limit, compare the real-time placement volume limit with the product placement volume value, and obtain shelf product specification judgment data based on the comparison results. Placement zoning module: Divide the smart shelf into several product placement areas, identify the products placed in each product placement area, and obtain the product placement data of multiple areas based on the identification results. Shelf monitoring module: Based on shelf product zoning data, it performs time-based product monitoring of products in different product placement areas and obtains shelf product monitoring data by analyzing the monitoring results; Intelligent feedback module: Based on the monitoring data of goods on the shelves and the data of goods zoning on the shelves, the module evaluates the sales status of the goods placed in each shelf area and provides feedback on the sales status based on the evaluation results.

[0014] A storage medium storing a computer program thereon, characterized in that, when the computer program is executed by a processor, it performs the steps of the control method.

[0015] In summary, the beneficial effects of this invention are: 1. This invention improves the effectiveness of intelligent shelves in controlling the placement of goods by automatically acquiring the volume of the goods to be placed and combining it with the volume of the empty area of ​​the shelf to determine the feasibility of placing the goods. 2. This invention collects and analyzes customers' body movements in front of the smart shelf, and combines the analysis results with the sales volume of the goods displayed on the shelf to accurately judge the sales status of the goods. Attached Figure Description

[0016] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 This is a diagram illustrating the implementation steps of the present invention; Figure 2 This is an overall system block diagram of the present invention; Figure 3 This is the product specification measurement area of ​​the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 Please see Figure 1 The present invention provides a technical solution: an intelligent shelf control method, which further includes the following specific steps: Step S1: Set up a product specification measurement area, perform projection analysis on each product to be placed through the product specification measurement area, obtain the product placement volume value corresponding to each product to be placed based on the analysis results, obtain the real-time placement volume limit, compare the real-time placement volume limit with the product placement volume value, and obtain shelf product specification judgment data based on the comparison results. Step S1 further includes the following specific steps: Step S11: Set a rectangular area through the bottom ground area of ​​the smart shelf, and any side of the rectangular area coincides with the bottom edge of the smart shelf. Name the side of the rectangular area that coincides with the bottom edge of the smart shelf as the coincident feature side, and name the set rectangular area as the product specification measurement area. Step S12: Perform projection analysis on the product specification measurement area to obtain the first projection acquisition image and the second projection acquisition image; Step S12 further includes the following specific steps: Step S121: Obtain the diagonal intersection of the rectangle corresponding to the product specification measurement area to obtain the first region feature point. Draw a straight line perpendicular to the product specification measurement area through the first region feature point to obtain the first region feature line. Draw a plane parallel to the product specification measurement area through the first region feature line to obtain the first light projection plane. Mark the plane opposite to the first light projection plane as the first projection plane. Step S122: Draw a plane perpendicular to the product specification measurement area through the overlapping feature edge to obtain the second ray projection plane. Set a projection imaging surface in the spatial area opposite to the first ray projection plane and name it the second projection plane. Step S123: Obtain several items to be placed, and select one sample item from the obtained items to be placed; Step S124: Place a single sample of the product to be placed in the product specification measurement area, emit a parallel projection beam from the first light projection plane to the first projection plane, and acquire the projection image of the sample of the product to be placed on the first projection plane through an image acquisition device to obtain a first projection acquisition image. Emit a parallel projection beam from the second light projection plane to the second projection plane, and acquire the projection image of the sample of the product to be placed on the second projection plane through an image acquisition device to obtain a second projection acquisition image. Step S13: Mark the edges of the projection area occupied by the sample product to be placed in the first projection acquisition image to obtain the projection rectangle corresponding to the sample product to be placed, and name it the first projection rectangle. Mark the edges of the projection area occupied by the sample product to be placed in the second projection acquisition image to obtain the projection rectangle corresponding to the sample product to be placed, and name it the second projection rectangle. Step S14: Select any two perpendicular rectangular sides in the first projection rectangle and name them as the first product volume side and the second product volume side, respectively. Obtain the length values ​​of the first product volume side and the second product volume side, respectively. Step S15: Perform side length analysis on the first projected rectangle to obtain the side length values ​​of the third volume; Step S15 further includes the following specific steps: Step S151: Select any two perpendicular rectangular sides in the second projection rectangle and name them as the first undetermined volume side and the second undetermined volume side. Obtain the length values ​​of the first undetermined volume side and the second undetermined volume side respectively. Step S152: If the length of the first undetermined side is equal to the length of the second undetermined side, then mark the length of the first undetermined side as the length of the third volume side. Step S153: If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the first undetermined side is equal to the length of the first volume side or the length of the second volume side, then mark the length of the first undetermined side as the length of the third volume side. Step S154: If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the second undetermined side is equal to the length of the first volume side or the length of the second volume side, then mark the length of the second undetermined side as the length of the third volume side. Step S16: Calculate the product placement volume corresponding to the sample product by taking the first volume side length value, the second volume side length value, and the third volume side length value; The volume of the sample items to be placed is calculated using the following formula: ; Where Vbf is the product placement volume corresponding to the sample product to be placed, Tjb1 is the first volume side length value, Tjb2 is the second volume side length value, and Tjb3 is the third volume side length value. Step S17: Determine the feasibility of placing the sample items based on the corresponding item placement volume values. Step S17 further includes the following specific steps: Step S171: Obtain the empty placement volume of each empty area of ​​the smart shelf, compare the values ​​of the obtained empty placement volumes of multiple areas, mark the empty placement volume of the area with the largest value as the real-time placement volume limit, and compare the value of the product placement volume corresponding to the sample product to be placed with the real-time placement volume limit. Step S172: If the product placement volume value corresponding to the sample product to be placed is greater than or equal to the real-time placement volume limit, then mark the sample product to be placed as an unplaceable product. Step S173: If the product placement volume value corresponding to the sample product to be placed is less than the real-time placement volume limit, then mark the sample product to be placed as a placeable product. Step S18: Obtain the volume values ​​corresponding to the products to be placed, and make a feasibility judgment on the placement of each product based on the obtained results to obtain shelf product specification judgment data.

[0020] Step S2: Divide the smart shelf into several product placement areas, and identify the products placed in each product placement area. Based on the identification results, obtain the products placed in multiple areas to obtain shelf product partitioning data. Step S2 further includes the following specific steps: Step S21: Obtain shelf product specification judgment data, and obtain each placeable product based on the shelf product specification judgment data; Step S22: Divide the smart shelf into several product placement areas and label the divided product placement areas as B1 product placement area to Ba product placement area respectively; Step S23: Place each placeable item in the B1 item placement area to the Ba item placement area, and name the items placed in the B1 item placement area to the Ba item placement area as B1 shelf items to Ba shelf items respectively. Step S24: Define the product placement area from B1 to Ba as the product placement area and the products on shelf B1 to Ba as shelf product partition data; Step S3: Based on the shelf merchandise zoning data, monitor the shelf merchandise in different merchandise placement areas in a time-based manner, and obtain shelf merchandise monitoring data by analyzing the monitoring results; Step S3 further includes the following specific steps: Step S31: Obtain shelf product partitioning data, and obtain the product placement areas from B1 to Ba and the shelf products from B1 to Ba based on the shelf product partitioning data; Step S32: In the historical period when the smart shelf sells goods from shelf B1 to shelf Ba, select several historical sales sub-periods of equal duration, and mark the selected historical sales sub-periods in chronological order as L1 historical sales sub-period to Lb historical sales sub-period. Step S33: Obtain the shelf video monitoring of the B1 product placement area during the historical sales sub-period of L1 to obtain the L1B1 monitoring video. Based on the L1B1 monitoring video, perform image monitoring on the B1 shelf products in the historical sales sub-period of L1. Based on the monitoring results, obtain the L1 time period product popularity coefficient corresponding to the B1 shelf products. Step S33 further includes the following specific steps: Step S331: Identify the customers appearing in the L1B1 surveillance video using an image recognition algorithm, and mark the identified customers sequentially as G1 surveillance customers to Gc surveillance customers according to the time sequence. Step S332: In the L1B1 surveillance video, extract video images from the time periods when the G1 monitoring customer to the Gc monitoring customer appears in the video to obtain the G1 customer monitoring image to the Gc customer monitoring image. Step S333: Perform image analysis on the monitoring image of customer G1, and obtain the product intention distance duration ratio and hand trajectory movement complexity of the monitored customer G1 based on the analysis results, and name them as G1 intention distance duration ratio and G1 hand trajectory complexity. Step S334: Obtain the G2 intention distance duration ratio to the Gc intention distance duration ratio and the G2 hand trajectory complexity to the Gc hand trajectory complexity respectively; Step S335: Calculate the product popularity coefficient for the L1 time period corresponding to the product on shelf B1 by combining the G1 intention distance duration ratio to the Gc intention distance duration ratio and the G1 hand trajectory complexity to the Gc hand trajectory complexity. The popularity coefficient of the products in time period L1 corresponding to the products on shelf B1 is calculated using the following formula: ; Where Rdl1 is the product popularity coefficient of L1 time period corresponding to the product of B1 shelf, Yjli is the Gi intention distance time ratio, Sfli is the Gi hand trajectory complexity, and c is the number of monitored customers. Step S34: Obtain the product popularity coefficients for the L2 historical sales sub-period to the Lb historical sales sub-period corresponding to the products on shelf B1, and obtain the product popularity coefficients for the L2 period to the Lb period. Step S35: Calculate the average of the product popularity coefficients from time period L1 to time period Lb to obtain the product popularity evaluation coefficient for B1. Step S36: Obtain the product popularity evaluation coefficients for products on shelves B2 to Ba respectively, to obtain the product popularity evaluation coefficients for products B2 to Ba. Step S37: Define the product popularity evaluation coefficients from B1 to Ba and the historical sales sub-periods from L1 to Lb as shelf product monitoring data; Step S4: Evaluate the sales status of the products placed in each shelf area based on the shelf product monitoring data and shelf product zoning data, and provide feedback on the sales status based on the evaluation results; Step S4 also includes the following specific steps: Step S41: Obtain shelf merchandise monitoring data, and obtain the B1 merchandise popularity evaluation coefficient to the Ba merchandise popularity evaluation coefficient and the L1 historical sales sub-period to the Lb historical sales sub-period based on the shelf merchandise monitoring data; Step S42: Obtain shelf product partitioning data, and obtain the product placement areas from B1 to Ba and the shelf products from B1 to Ba based on the shelf product partitioning data; Step S43: Obtain monitoring images of the B1 product placement area from the L1 historical sales sub-period to the Lb historical sales sub-period. Based on the monitoring images, obtain the sales quantity of B1 shelf products from the L1 historical sales sub-period to the Lb historical sales sub-period, and calculate the average to obtain the sales quantity of B1 products during the time period. Step S44: Obtain the sales quantity of the products on shelf B2 to shelf Ba for each time period, and obtain the sales quantity of products on shelf B2 to shelf Ba for each time period; Step S45: Obtain shelf merchandise monitoring data, and obtain the popularity evaluation coefficients of merchandise B1 to Ba respectively based on the shelf merchandise monitoring data; Step S46: Calculate the ratio of the sales volume of product B1 during the time period to the popularity evaluation coefficient of product B1 to obtain the sales popularity ratio of product B1. Calculate the ratio of the sales volume of product B2 during the time period to the popularity evaluation coefficient of product B3 to obtain the sales popularity ratio of product B2. And so on, calculate the ratio of the sales volume of product Ba during the time period to the popularity evaluation coefficient of product Ba to obtain the sales popularity ratio of product Ba. Step S47: In the existing histogram template, mark the shelf products as the horizontal axis and the product sales popularity ratio as the vertical axis. Mark the product sales popularity ratio from B1 to Ba in the form of histogram bars to obtain the product sales popularity histogram. Step S48: Select a sales popularity benchmark point on the y-axis of the product sales popularity histogram, and draw a straight line perpendicular to the y-axis through the sales popularity benchmark point to obtain the sales popularity benchmark line. Step S49: If the bar corresponding to a product on the shelf is lower than the sales popularity baseline, the smart shelf will assess the sales status of the corresponding product as abnormal and issue an abnormality warning. If the bar corresponding to a product on the shelf is higher than or equal to the sales popularity baseline, the smart shelf will assess the sales status of the corresponding product as normal.

[0021] In this application, if a corresponding calculation formula appears, the above calculation formula is a dimensionless calculation. The weighting coefficient, proportional coefficient and other coefficients in the formula are set to quantify each parameter to obtain a result value. The size of the weighting coefficient and proportional coefficient is only required to not affect the proportional relationship between the parameter and the result value.

[0022] Example 2 Please see Figure 2 Based on another concept of the same invention, an intelligent shelf control system is proposed. The shelf control system includes a product placement module, a placement zoning module, a shelf monitoring module, an intelligent feedback module, and a server. The product placement module, placement zoning module, shelf monitoring module, and intelligent feedback module are respectively connected to the server, and the server controls the product placement module, placement zoning module, shelf monitoring module, and intelligent feedback module respectively. The product placement module sets up a product specification measurement area. The module performs projection analysis on each product to be placed through the product specification measurement area. Based on the analysis results, it obtains the product placement volume value corresponding to each product to be placed and obtains the real-time placement volume limit. The real-time placement volume limit is compared with the product placement volume value. Based on the comparison results, the shelf product specification judgment data is obtained. Specifically as follows: Set a rectangular area at the bottom of the smart shelf, and make sure that any side of the rectangular area coincides with the bottom edge of the smart shelf. Name the side of the rectangular area that coincides with the bottom edge of the smart shelf as the coincident feature side, and name the rectangular area as the product specification measurement area. It should be noted here that: In this application, the size of the rectangular area can be set according to the size of the smart shelf. In this application, the size of the rectangular area is 0.5m × 0.8m.

[0023] Please see Figure 3 The first feature point of the first region is obtained by finding the intersection of the diagonals of the rectangle corresponding to the product specification measurement area. A straight line perpendicular to the product specification measurement area is drawn through the first feature point of the first region to obtain the first feature line of the first region. A plane parallel to the product specification measurement area is drawn through the first feature line of the first region to obtain the first ray projection plane. The plane opposite to the first ray projection plane is marked as the first projection plane. A plane perpendicular to the product specification measurement area is drawn through the overlapping feature edge to obtain the second ray projection plane. A projection imaging surface is set in the spatial area opposite to the first ray projection plane and named the second projection plane. Obtain several items to be placed, and select one sample item from the obtained items to be placed; It should be noted here that: In this application, the outer packaging of the various goods to be placed is a cuboid.

[0024] Please see Figure 3A single sample of the product to be placed is placed in the product specification measurement area. A parallel projection beam is emitted from the first light projection plane to the first projection plane. The image acquisition device acquires the projection image of the sample of the product to be placed on the first projection plane to obtain the first projection image. A parallel projection beam is emitted from the second light projection plane to the second projection plane. The image acquisition device acquires the projection image of the sample of the product to be placed on the second projection plane to obtain the second projection image. In the first projection acquisition image, the projection area occupied by the sample product to be placed is marked with edges to obtain the projection rectangle corresponding to the sample product to be placed, and it is named the first projection rectangle. In the second projection acquisition image, the projection area occupied by the sample product to be placed is marked with edges to obtain the projection rectangle corresponding to the sample product to be placed, and it is named the second projection rectangle. In the first projection rectangle, select any two mutually perpendicular rectangular sides and name them the first product volume side and the second product volume side, respectively. Obtain the length values ​​of the first product volume side and the second product volume side, respectively. In the second projection rectangle, select any two mutually perpendicular rectangular sides and name them the first undetermined volume side and the second undetermined volume side. Obtain the length values ​​of the first undetermined volume side and the second undetermined volume side respectively. If the length of the first undetermined side is equal to the length of the second undetermined side, then the length of the first undetermined side is marked as the length of the third volume side. If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the first undetermined side is equal to the length of the first volume side or the length of the second volume side, then the length of the first undetermined side is marked as the length of the third volume side. If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the second undetermined side is equal to the length of the first volume side or the length of the second volume side, then the length of the second undetermined side is marked as the length of the third volume side. It should be noted here that: In this application, by determining any two mutually perpendicular faces of a cuboid that share a common edge, the length, width, and height of the corresponding cuboid can be obtained.

[0025] The product placement volume corresponding to the sample product is obtained by calculating the first volume side length value, the second volume side length value, and the third volume side length value. The volume of the sample items to be placed is calculated using the following formula: ; Where Vbf is the product placement volume corresponding to the sample product to be placed, Tjb1 is the first volume side length value, Tjb2 is the second volume side length value, and Tjb3 is the third volume side length value. The feasibility of placing the sample items is judged based on the corresponding volume values ​​of the sample items to be placed. Specifically as follows: Obtain the empty placement volume corresponding to each empty area of ​​the smart shelf, compare the values ​​of multiple empty placement volumes, mark the empty placement volume of the area with the largest value as the real-time placement volume limit, and compare the value of the product placement volume corresponding to the sample product to be placed with the real-time placement volume limit. If the product placement volume value corresponding to the sample product to be placed is greater than or equal to the real-time placement volume limit, then the sample product to be placed will be marked as an unplaceable product. If the product placement volume of the sample product is less than the real-time placement volume limit, then the sample product will be marked as a placeable product. Repeat the process of obtaining the volume values ​​of the products to be placed in the sample, obtain the volume values ​​of the products to be placed, and make a feasibility judgment on the placement of each product based on the results, so as to obtain the shelf product specification judgment data.

[0026] The product placement zoning module divides the smart shelf into several product placement areas and identifies the products placed in each product placement area. Based on the identification results, it obtains the products placed in multiple areas and obtains the shelf product zoning data. Specifically as follows: Obtain shelf product specification judgment data, and obtain each placeable product based on the shelf product specification judgment data; The smart shelf is divided into several product placement areas, and these areas are labeled as B1 product placement area to Ba product placement area. It should be noted here that: In this application, B is the identifier corresponding to the product placement area, a is the quantity value corresponding to the divided product placement area, and a is an integer greater than 0. In this application, the shelf goods referred to herein are fast-moving consumer goods (FMCG), which include, but are not limited to, toothpaste and shower gel in rectangular packaging.

[0027] It should be noted here that: In this application, the goods placed in the same product display area are the same type of goods, and there is no situation where the same type of goods are placed in different areas.

[0028] Place each item in the B1 item placement area to the Ba item placement area, and name the items placed in the B1 item placement area to the Ba item placement area respectively as B1 shelf items to Ba shelf items; It should be noted here that: In this application, the quantity of goods placement areas and the quantity of goods on shelves are both referred to as 'a'.

[0029] Define the product placement area from B1 to Ba as the product placement area from B1 shelf to Ba shelf as shelf product partitioning data; The product placement zoning module acquires the product zoning data on the shelves and transmits it to the shelf monitoring module, product analysis module, quality monitoring module, and intelligent control module.

[0030] The shelf monitoring module monitors the shelf merchandise in different merchandise placement areas in a time-based manner based on the shelf merchandise zoning data, and obtains shelf merchandise monitoring data by analyzing the monitoring results. Specifically as follows: Obtain shelf product zoning data, and based on the shelf product zoning data, obtain the product placement area from B1 to Ba and the shelf products from B1 to Ba respectively. In the historical period when the smart shelf sells goods from shelf B1 to shelf Ba, select several historical sales sub-periods of equal duration and mark them in chronological order as L1 historical sales sub-period to Lb historical sales sub-period. It should be noted here that: In this application, L is the identifier corresponding to the historical sales sub-period, a is the quantity value corresponding to the historical sales sub-period, and a is an integer greater than 0; In this application, during the historical sales sub-periods L1 to Lb, the goods on shelves B1 to Ba are placed in the same position on the shelves.

[0031] Obtain video surveillance of the shelves in the B1 product placement area during the historical sales sub-period of L1 to obtain L1B1 surveillance video. Based on the L1B1 surveillance video, perform image monitoring on the B1 shelf products in the historical sales sub-period of L1, and obtain the L1 time period product popularity coefficient corresponding to the B1 shelf products based on the monitoring results. Specifically as follows: The image recognition algorithm is used to identify customers appearing in the L1B1 surveillance video, and the identified customers are marked as G1 surveillance customers to Gc surveillance customers in chronological order. It should be noted here that: In this application, G is the identifier corresponding to the monitored customer, c is the quantity value corresponding to the monitored customer, and c is an integer greater than 0; In the L1B1 surveillance video, video images were extracted from the time periods when the G1 monitoring customer to the Gc monitoring customer appeared in the video, resulting in the monitoring images of the G1 customer to the Gc customer. It should be noted here that: In this application, the customer surveillance images referred to herein are 3D surveillance images; Image analysis was performed on the monitoring images of customer G1. Based on the analysis results, the product intention distance-time ratio and hand trajectory movement complexity of the monitored customer G1 were obtained and named G1 intention distance-time ratio and G1 hand trajectory complexity, respectively. Specifically as follows: The feature customer monitoring image is transposed into a three-dimensional spatial model using a three-dimensional reconstruction algorithm to obtain the feature three-dimensional spatial model. Please see Figure 2 In the feature 3D spatial model, any feature point in the B1 product placement area is selected as the origin of the coordinate system. A plane parallel to the shelf plane is drawn from the origin to obtain the first feature plane. A plane perpendicular to the first feature plane is drawn through the origin to obtain the second feature plane. In the first feature plane, any straight line is drawn through the origin to obtain the x-axis. A straight line perpendicular to the x-axis is drawn through the origin to obtain the y-axis. In the second feature plane, a straight line perpendicular to the first feature plane is drawn through the origin to obtain the z-axis. The spatial coordinate system composed of the origin, x-axis, y-axis and z-axis is named the model spatial coordinate system. In the feature 3D spatial model, the space area occupied by the customer's hand is 3D modeled to obtain the hand 3D model, the space area occupied by the goods on the B1 shelf is 3D modeled to obtain the goods 3D model, the geometric center point of the model corresponding to the hand 3D model is obtained to obtain the hand feature monitoring point, and the geometric center point of the model corresponding to the goods 3D model is obtained to obtain the goods feature monitoring point. It should be noted here that: In this application, if both of the customer's hands are in the feature three-dimensional space model, the space area occupied by the customer's hands is specifically the space area occupied by the customer's right hand.

[0032] In the model space coordinate system, the three-dimensional coordinates of the hand feature monitoring point at each time point are obtained to obtain the hand monitoring three-dimensional coordinates. The three-dimensional coordinates of the product feature monitoring point at each time point are obtained to obtain the product monitoring three-dimensional coordinates. The product contact distance is obtained by calculating the hand monitoring three-dimensional coordinates and the product monitoring three-dimensional coordinates. The formula for calculating the contact distance with goods is as follows: ; Where Sjj is the product contact distance, (x s1 ,y s1 ,z s1 (x) represents the three-dimensional coordinates for hand monitoring. p2 ,y p2 ,z p2 ( ) represents the three-dimensional coordinates for commodity monitoring; Obtain the preset range of product contact distance. If the product contact distance is within the preset range, the corresponding time point is marked as the intended distance time point. The intended distance time ratio is obtained by the proportion of time occupied by the intended distance time point in the feature customer monitoring image. It should be noted here that: In this application, the preset range of product contact distance needs to be set according to the specific type of product on shelf B1. In specific implementation, if the product on shelf B1 is "toothpaste", there is test data showing that the product contact distance is 5cm, and the preset range of product contact distance is set to [0,50], then the corresponding time point is the intended distance time point.

[0033] Perform motion trajectory analysis on hand feature monitoring points in characteristic customer monitoring images, and obtain the motion complexity of hand trajectory based on the analysis results; Specifically as follows: In the characteristic customer monitoring image, select a movement trajectory analysis period from the time period when the contact distance of the goods corresponding to the hand feature monitoring point is within the preset range of the contact distance of the goods; During the movement trajectory analysis period, mark several trajectory time points with equal intervals, and name the marked trajectory time points sequentially from J1 trajectory time point to Je trajectory time point in chronological order. It should be noted here that: In this application, J is the identifier corresponding to the trajectory time point, e is the quantity value corresponding to the trajectory time point, and e is an integer greater than 0; In the model space coordinate system, the coordinates of the hand feature monitoring points at the time points from the J1 trajectory to the Je trajectory are obtained respectively, thus obtaining the coordinate points from the J1 trajectory to the Je trajectory. Connect the coordinates of trajectory J1 and trajectory J2 to obtain trajectory line P1; connect the coordinates of trajectory J2 and trajectory J3 to obtain trajectory line P2; and so on. Connect the coordinates of trajectory Je-1 and trajectory Je to obtain trajectory line Pd. It should be noted here that: In this application, d is the quantity value corresponding to the trajectory connection line, d is an integer greater than 0 and d=e-1.

[0034] The length values ​​of the lines connecting P1 trajectory and Pd trajectory are obtained respectively, thus obtaining the length values ​​of P1 trajectory and Pd trajectory. Calculate the difference between the length values ​​of trajectory P1 and trajectory P2, and take the absolute value of the difference to obtain the trajectory length deviation of P1. Similarly, calculate the difference between the length values ​​of trajectory P2 and trajectory P3, and take the absolute value of the difference to obtain the trajectory length deviation of P2. Continue this process, calculating the difference between the length values ​​of trajectory Pd-1 and trajectory Pd, and taking the absolute value of the difference to obtain the trajectory length deviation of Pd-1. In the model space coordinate system, the plane where the line connecting P1 trajectory and P2 trajectory is located is marked as the P1 trajectory plane, the plane where the line connecting P2 trajectory and P3 trajectory is located is marked as the P2 trajectory plane, and so on, the plane where the line connecting Pd-1 trajectory and Pd trajectory is located is marked as the Pd-1 trajectory plane. Obtain the angle between the line connecting P1 trajectory and the line connecting P2 trajectory in the plane of P1 trajectory, and obtain the angle value of P1 trajectory. Obtain the angle between the line connecting P2 trajectory and the line connecting P3 trajectory in the plane of P2 trajectory, and obtain the angle value of P1 trajectory. Similarly, obtain the angle between the line connecting Pd-1 trajectory and the line connecting Pd trajectory in the plane of Pd-1 trajectory, and obtain the angle value of Pd-1 trajectory. The hand trajectory movement complexity is obtained by calculating the trajectory length deviation from P1 to Pd-1 and the trajectory angle value from P1 to Pd-1. The complexity of hand trajectory movement is calculated using the following formula: ; Where Sfd is the complexity of the hand trajectory movement, Cdpi is the deviation of the Pi trajectory length, Jdpi is the Pi trajectory angle value, and d is the number of corresponding trajectory lines.

[0035] It should be noted here that: In practice, the experiment showed that the trajectory length deviation of P1 was 5cm, the trajectory length deviation of P2 was 3cm, the trajectory length deviation of P3 was 7cm, the trajectory angle value of P1 was 31 degrees, the trajectory angle value of P2 was 15 degrees, and the trajectory angle value of P3 was 41 degrees. The complexity of the hand trajectory movement can be calculated to be 162.4.

[0036] Repeat the process of obtaining the product intention distance duration ratio and hand trajectory movement complexity corresponding to the customer monitored by G1, and obtain the intention distance duration ratio from G2 to Gc and the hand trajectory complexity from G2 to Gc respectively. The product popularity coefficient of L1 time period corresponding to the product on shelf B1 is obtained by calculating the ratio of intention distance duration from G1 to Gc and the ratio of intention distance duration from G1 to Gc, as well as the hand trajectory complexity from G1 to Gc. The popularity coefficient of the products in time period L1 corresponding to the products on shelf B1 is calculated using the following formula: ; Where Rdl1 is the product popularity coefficient of L1 time period corresponding to the product of B1 shelf, Yjli is the Gi intention distance time ratio, Sfli is the Gi hand trajectory complexity, and c is the number of monitored customers. The popularity coefficients of the products in the L2 historical sales sub-period to the Lb historical sales sub-period corresponding to the products in the B1 shelf are obtained respectively, and the popularity coefficients of the products in the L2 period to the Lb period are obtained. The average of the product popularity coefficients from time period L1 to time period Lb is calculated to obtain the product popularity evaluation coefficient for time period B1. It should be noted here that: In this application, the B1 product popularity assessment coefficient represents the popularity or demand intensity of the B1 product in the market. The larger the B1 product popularity assessment coefficient, the greater the demand intensity of the B1 product in the market.

[0037] Repeat the process of obtaining the popularity evaluation coefficient of product B1, and obtain the popularity evaluation coefficients of products corresponding to products on shelves B2 to Ba respectively, to obtain the popularity evaluation coefficients of products B2 to Ba. The product popularity evaluation coefficients from B1 to Ba and the historical sales sub-periods from L1 to Lb are defined as shelf product monitoring data. The intelligent feedback module assesses the sales status of the products placed in each shelf area based on the monitoring data of the products on the shelf and the data of the product zoning on the shelf, and provides feedback on the sales status based on the assessment results; Specifically as follows: Obtain shelf merchandise monitoring data, and based on the shelf merchandise monitoring data, obtain the B1 merchandise popularity evaluation coefficient to the Ba merchandise popularity evaluation coefficient and the L1 historical sales sub-period to the Lb historical sales sub-period. Obtain shelf product zoning data, and based on the shelf product zoning data, obtain the product placement area from B1 to Ba and the shelf products from B1 to Ba respectively. Acquire monitoring images of the B1 product display area from the historical sales sub-period of L1 to the historical sales sub-period of Lb. Based on the monitoring images, obtain the sales quantity of B1 shelf products from the historical sales sub-period of L1 to the historical sales sub-period of Lb, and calculate the average to obtain the sales quantity of B1 products in each period. Repeat the process of obtaining the sales quantity of product B1 during the same time period, and then obtain the sales quantity of products from shelf B2 to shelf Ba during the same time period to obtain the sales quantity of product B2 during the same time period to product Ba during the same time period. Obtain shelf merchandise monitoring data, and based on the shelf merchandise monitoring data, obtain the popularity evaluation coefficients for products B1 to Ba respectively; Calculate the ratio of the sales volume of product B1 during the time period to the popularity evaluation coefficient of product B1 to obtain the sales popularity ratio of product B1. Calculate the ratio of the sales volume of product B2 during the time period to the popularity evaluation coefficient of product B3 to obtain the sales popularity ratio of product B2. And so on, calculate the ratio of the sales volume of product Ba during the time period to the popularity evaluation coefficient of product Ba to obtain the sales popularity ratio of product Ba. In the existing histogram template, the shelf products are marked as the horizontal axis and the product sales popularity ratio is marked as the vertical axis. The product sales popularity ratio from B1 to Ba is marked in the histogram template in the form of histogram bars to obtain the product sales popularity histogram. Select a sales popularity benchmark point on the y-axis of the product sales popularity histogram, and draw a straight line perpendicular to the y-axis through the sales popularity benchmark point to obtain the sales popularity benchmark line. If the corresponding bar for a product on the shelf is lower than the sales popularity benchmark, the smart shelf will assess the abnormal sales status of the corresponding product and issue an abnormality warning. If the corresponding bar of a product on the shelf is higher than or equal to the sales popularity baseline, the smart shelf will assess that the sales status of the corresponding product is normal. A storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the training method.

[0038] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for controlling an intelligent shelf, characterized in that, Includes the following steps: Step S1: Set up a product specification measurement area, perform projection analysis on each product to be placed through the product specification measurement area, obtain the product placement volume value corresponding to each product to be placed based on the analysis results, obtain the real-time placement volume limit, compare the real-time placement volume limit with the product placement volume value, and obtain shelf product specification judgment data. Step S2: Divide the smart shelf into product placement areas, and identify the products placed in each product placement area. Based on the identification results, obtain the products placed in multiple areas to obtain shelf product partitioning data. Step S3: Based on the shelf merchandise zoning data, monitor the shelf merchandise in different merchandise placement areas in a time-based manner, and obtain shelf merchandise monitoring data by analyzing the monitoring results; Step S4: Evaluate the sales status of the products placed in each shelf area based on the shelf product monitoring data and shelf product zoning data, and provide feedback on the sales status based on the evaluation results.

2. The intelligent shelf control method according to claim 1, characterized in that, Step S1 further includes the following steps: Step S11: Set up a product specification measurement area on the ground at the bottom of the smart shelf; Step S12: Perform projection analysis on the product specification measurement area to obtain the first projection acquisition image and the second projection acquisition image; Step S13: Mark the edges of the projection area occupied by the sample product to be placed in the first projection acquisition image to obtain the first projection rectangle; mark the edges of the projection area occupied by the sample product to be placed in the second projection acquisition image to obtain the second projection rectangle. Step S14: Select the first product volume side and the second product volume side in the first projection rectangle and obtain the length values ​​of the first volume side and the second volume side. Step S15: Perform side length analysis on the first projected rectangle to obtain the side length values ​​of the third volume; Step S16: Calculate the product placement volume Vbf corresponding to the sample product by using the first volume side length value Tjb1, the second volume side length value Tjb2, and the third volume side length value Tjb3. ; Step S17: Obtain the real-time placement volume limit. If the placement volume value of the sample product to be placed is greater than or equal to the real-time placement volume limit, then mark the sample product to be placed as an unplaceable product. If the placement volume value of the sample product to be placed is less than the real-time placement volume limit, then mark the sample product to be placed as a placeable product. Step S18: Obtain the volume values ​​corresponding to the products to be placed, and make a feasibility judgment on the placement of each product based on the obtained results to obtain shelf product specification judgment data.

3. The intelligent shelf control method according to claim 2, characterized in that, Step S12 includes the following steps: Step S121: Obtain the intersection of the diagonals of the rectangle corresponding to the product specification measurement area to obtain the first region feature point. Draw a straight line perpendicular to the product specification measurement area through the first region feature point to obtain the first region feature line. Draw a plane parallel to the product specification measurement area through the first region feature line to obtain the first light projection plane. Mark the plane opposite to the first light projection plane as the first projection plane. Step S122: Draw a plane perpendicular to the product specification measurement area through the overlapping feature edge to obtain the second ray projection plane. Set a projection imaging surface in the spatial area opposite to the first ray projection plane and name it the second projection plane. Step S123: Obtain several items to be placed, and select one sample item to be placed; Step S124: Place a single sample of the product to be placed in the product specification measurement area, emit a parallel projection beam from the first light projection plane to the first projection plane, and acquire the projection image of the sample of the product to be placed on the first projection plane to obtain the first projection acquisition image. Then emit a parallel projection beam from the second light projection plane to the second projection plane, and acquire the projection image of the sample of the product to be placed on the second projection plane to obtain the second projection acquisition image.

4. The intelligent shelf control method according to claim 2, characterized in that, Step S15 includes the following steps: Step S151: Select any two mutually perpendicular rectangular sides in the second projection rectangle as the first undetermined volume side and the second undetermined volume side, and obtain their length values ​​to get the length values ​​of the first undetermined side and the second undetermined side. Step S152: If the length of the first undetermined side is equal to the length of the second undetermined side, then mark the length of the first undetermined side as the length of the third volume side. Step S153: If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the first undetermined side is equal to the length of the first volume side or the length of the second volume side, then mark the length of the first undetermined side as the length of the third volume side. Step S154: If the length of the first undetermined side is not equal to the length of the second undetermined side, and the length of the second undetermined side is equal to the length of the first volume side or the length of the second volume side, then mark the length of the second undetermined side as the length of the third volume side.

5. The intelligent shelf control method according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Obtain shelf product specification judgment data, and obtain each placeable product based on the shelf product specification judgment data; Step S22: Divide the smart shelf into product placement areas B1 to Ba; Step S23: Place each placeable item in the B1 item placement area to the Ba item placement area, and name the items placed in the B1 item placement area to the Ba item placement area as B1 shelf items to Ba shelf items respectively. Step S24: Define the product placement area from B1 to Ba as the product placement area and the product on shelf B1 to Ba as shelf product partition data.

6. The intelligent shelf control method according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Obtain shelf product partitioning data, and obtain the product placement areas from B1 to Ba and the shelf products from B1 to Ba based on the shelf product partitioning data; Step S32: In the historical period when the smart shelf sold goods from shelf B1 to shelf Ba, select the historical sales sub-periods from L1 to Lb. Step S33: Obtain the shelf video monitoring of the B1 product placement area during the historical sales sub-period of L1, obtain the L1B1 monitoring video, analyze the L1B1 monitoring video, and obtain the product popularity coefficient of the L1 period corresponding to the B1 shelf product. Step S34: Obtain the product popularity coefficients for the L2 historical sales sub-period to the Lb historical sales sub-period corresponding to the products on shelf B1, and obtain the product popularity coefficients for the L2 period to the Lb period. Step S35: Calculate the average of the product popularity coefficients from time period L1 to time period Lb to obtain the product popularity evaluation coefficient for B1. Step S36: Obtain the product popularity evaluation coefficients for products on shelves B2 to Ba respectively, to obtain the product popularity evaluation coefficients for products B2 to Ba. Step S37: Define the product popularity evaluation coefficients from B1 to Ba and the historical sales sub-periods from L1 to Lb as shelf product monitoring data.

7. The intelligent shelf control method according to claim 6, characterized in that, Step S33 includes the following steps: Step S331: Identify the customers appearing in the L1B1 surveillance video and mark the identified customers in chronological order as G1 surveillance customers to Gc surveillance customers; Step S332: In the L1B1 surveillance video, extract video images from the time periods when the G1 monitoring customer to the Gc monitoring customer appears in the video to obtain the G1 customer monitoring image to the Gc customer monitoring image. Step S333: Perform image analysis on the monitoring image of customer G1, and obtain the product intention distance duration ratio and hand trajectory movement complexity of the monitored customer G1 based on the analysis results, and name them as G1 intention distance duration ratio and G1 hand trajectory complexity. Step S334: Obtain the G2 intention distance duration ratio to the Gc intention distance duration ratio and the G2 hand trajectory complexity to the Gc hand trajectory complexity; Step S335: Calculate the product popularity coefficient for the L1 time period corresponding to the product on shelf B1 by combining the G1 intention distance duration ratio to the Gc intention distance duration ratio and the G1 hand trajectory complexity to the Gc hand trajectory complexity. The popularity coefficient of the products in time period L1 corresponding to the products on shelf B1 is calculated using the following formula: ； Where Rdl1 is the product popularity coefficient of L1 time period corresponding to the product on shelf B1, Yjli is the Gi intention distance duration ratio, Sfli is the Gi hand trajectory complexity, and c is the number of monitored customers.

8. The intelligent shelf control method according to claim 1, characterized in that, Step S4 includes the following steps: Step S41: Obtain shelf merchandise monitoring data, and obtain the B1 merchandise popularity evaluation coefficient to the Ba merchandise popularity evaluation coefficient and the L1 historical sales sub-period to the Lb historical sales sub-period based on the shelf merchandise monitoring data; Step S42: Obtain shelf product partitioning data, and obtain the product placement areas from B1 to Ba and the shelf products from B1 to Ba based on the shelf product partitioning data; Step S43: Obtain monitoring images of the B1 product placement area from the L1 historical sales sub-period to the Lb historical sales sub-period. Based on the monitoring images, obtain the sales quantity of B1 shelf products from the L1 historical sales sub-period to the Lb historical sales sub-period, and calculate the average to obtain the sales quantity of B1 products during the time period. Step S44: Obtain the sales quantity of the products on shelf B2 to shelf Ba for each time period, and obtain the sales quantity of products on shelf B2 to shelf Ba for each time period; Step S45: Obtain shelf merchandise monitoring data, and obtain the popularity evaluation coefficients of B1 merchandise to Ba merchandise based on the shelf merchandise monitoring data; Step S46: Calculate the ratio of the sales volume of product B1 during the time period to the popularity evaluation coefficient of product B1 to obtain the sales popularity ratio of product B1. Similarly, obtain the sales popularity ratios of products B2 to Ba. Step S47: Create a product sales popularity histogram based on the sales popularity ratio of product B1 to product Ba, and set the sales popularity baseline in the product sales popularity histogram. Step S49: If the bar corresponding to a product on the shelf is lower than the sales popularity baseline, the sales status of the corresponding product is assessed as abnormal, and an abnormality warning is issued. If the bar corresponding to a product on the shelf is higher than or equal to the sales popularity baseline, the sales status of the corresponding product is assessed as normal.

9. An intelligent shelf control system, applicable to the intelligent shelf control method according to any one of claims 1-8, characterized in that, The intelligent shelf control system includes: Product placement module: Set up a product specification measurement area, perform projection analysis on each product to be placed through the product specification measurement area, obtain the product placement volume value corresponding to each product to be placed based on the analysis results, obtain the real-time placement volume limit, compare the real-time placement volume limit with the product placement volume value, and obtain shelf product specification judgment data based on the comparison results. Placement zoning module: Divide the smart shelf into several product placement areas, identify the products placed in each product placement area, and obtain the product placement data of multiple areas based on the identification results. Shelf monitoring module: Based on shelf product zoning data, it performs time-based product monitoring of products in different product placement areas and obtains shelf product monitoring data by analyzing the monitoring results; Intelligent feedback module: Based on the monitoring data of goods on the shelves and the data of goods zoning on the shelves, the module evaluates the sales status of the goods placed in each shelf area and provides feedback on the sales status based on the evaluation results.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it performs the steps of the control method as described in any one of claims 1-8.

Images