Promotion method based on visual interaction of equipment side

Abstract

The invention discloses a promotion method based on equipment side visual interaction, and relates to the technical field of marketing and computer vision. According to the method, commodity advantages are quantitatively evaluated through three dimensions, price competitiveness is evaluated through competitiveness indexes, the low-price-only theory is avoided, the re-purchase contribution degree and the sales volume increase degree are fused, public praise type commodities are identified, interaction behaviors such as browsing, clicking, purchase adding and evaluation are calculated through weighting, the user participation depth is quantified, the commodity potential value is mined from different angles, and the user experience is improved. Data support is provided for optimization of subsequent visual interaction, and the problem that in the prior art, a popularization means of the price only theory is adopted, and full mining of potential values of commodities is lacked is solved.

Description

Technical Field

[0001] The present invention relates to the fields of marketing and computer vision technologies, and particularly relates to a promotion method based on device-side visual interaction. Background Art

[0002] In today's digital age, advertisements and promotion activities are everywhere. Traditional promotion methods, such as TV advertisements, print advertisements, etc., although they can cover a certain audience group, lack direct interaction with users. Users are often in a passive state of receiving information, resulting in poor promotion effects. With the development of mobile devices and computer vision technologies, new opportunities have been brought to promotion activities.

[0003] However, the existing vision-based promotion methods in the prior art still have the following deficiencies in the actual application process: When the prior art uses the method of visual interaction for product promotion, most of them only convert the traditional "text list" information into a virtual-real integrated scene display and adopt the promotion means of "price-only theory", lacking full exploration of the potential value of products, and unable to adopt different promotion strategies according to the exploration results of the potential value of products, resulting in poor promotion efficiency and effects.

[0004] Therefore, a promotion method based on device-side visual interaction is introduced. Summary of the Invention

[0005] In view of this, the present invention provides a promotion method based on device-side visual interaction to solve the problems raised in the above background art.

[0006] The object of the present invention can be achieved by the following technical solutions: A promotion method based on device-side visual interaction, including: Interactive content generation: After associating with the ID of the product currently photographed by the user; the ID includes the real-time price, discount information, and price trend in the past X days of the product; in the real-time screen of the device-side screen, the appearance of the product is superimposed through AR technology, and the product price, discount information, and price trend in the past X days are marked in the form of floating labels, where X>10; Intelligent promotion optimization: Based on the ID of the product currently photographed by the user, retrieve the price data, sales data, and user data of the currently photographed product within a set time window before the current time point in the cloud product database, and perform analysis and evaluation after retrieval to determine the price advantage score, sales advantage score, and conversion advantage score of the product within a set time window before the current time point; Promotion decision execution: Based on the price advantage score, sales advantage score, and conversion advantage score of the product currently photographed by the user, execute corresponding steps to optimize the scene of the visual interaction on the device side.

[0007] In some embodiments, the price advantage score, sales advantage score, and conversion advantage score of a product within a set time window before the current time point are determined as follows: S1: Extract the price changes of the currently photographed product within the set time window before the current time point from the price data. First, calculate the average value of the price changes in each group within the set time window as the price level value, and then obtain the product price of the currently photographed product at the current time point, denoted as the current price; Use the formula to calculate the competitiveness index of the currently photographed product at the current time point ; S2: Extract the total sales amount of the currently photographed product within the set time window before the current time point from the sales data. For the total sales amount within the set time window, divide the sales amount of repeat customers, and use the sales amount of repeat customers divided by the total sales amount to calculate the repeat purchase contribution degree of the currently photographed product within the set time window; Extract the total sales amount in the previous set time window as the denominator and the total sales amount in the current set time window as the numerator for ratio calculation, and calculate the sales volume growth rate of the currently photographed product within the set time window; Multiply the repeat purchase contribution degree and the sales volume growth rate of the currently photographed product by the set weight coefficients respectively to obtain the sales advantage index of the currently photographed product; S3: Extract the number of views, clicks, add-to-cart quantities, and evaluation quantities of the currently photographed product within the set time window at the current time point from the user data, and mark them as a, b, c, and d respectively. Use the formula for weighted calculation to obtain the interaction participation index of the currently photographed product ; where g1, g2, g3, and g4 are the weight coefficients of the number of views, clicks, add-to-cart quantities, and evaluation quantities respectively; S4: Based on the pre-constructed conversion rules, convert the competitiveness index , sales advantage index, and interaction participation index into price advantage score, sales advantage score, and conversion advantage score respectively.

[0008] In some embodiments, the specific conversion process of the conversion rules is as follows: Pre-construct the range data sets corresponding to the competitiveness index , sales advantage index, and interaction participation index respectively. The range data sets contain groups of index intervals respectively associated with the competitiveness index , sales advantage index, and interaction participation index ; For the competitiveness index Each group of associated index intervals corresponds to a price advantage score respectively; each group of index intervals associated with the sales advantage index corresponds to a sales advantage score respectively; each group of index intervals associated with the interaction participation index corresponds to a conversion advantage score respectively; Match the competitiveness index of the currently photographed product at the current time point , the sales advantage index, and the interaction participation index in the input range data set, and output the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product.

[0009] In some embodiments, the corresponding steps are performed to optimize the visual interaction of the device side, specifically: M1: Identify the user type to which the current user belongs, and perform corresponding steps to optimize the visual interaction of the device side based on the user type to which the user belongs; the user types include newly registered users and old users.

[0010] In some embodiments, the corresponding steps are performed based on the user type to which the user belongs, specifically: M2: If the current user is a newly registered user, calculate the comprehensive promotion index of the currently photographed product; set the passing reference index range for the comprehensive promotion index corresponding to each type of product. If the comprehensive promotion index of the currently photographed product is higher than the passing reference index range; then use the pre-constructed virtual AR character to voice broadcast that the currently photographed product is a "full-dimensional advantage product", and then mark the specific value by which the comprehensive promotion index exceeds the passing reference index range and attach a "rush to buy now" hot zone link; If the comprehensive promotion index of the currently photographed product is within the passing reference index range, compare the comprehensive promotion index of the product photographed by the current user with the comprehensive promotion index of similar products in the cloud product database. If the comparison result shows that the comprehensive promotion index of the product photographed by the current user is higher than a set percentage of similar products; then use the pre-constructed virtual AR character to voice broadcast that the currently photographed product is a "similar product advantage", and then mark "the current product exceeds Y of similar products", and attach a "rush to buy now" hot zone link; where Y represents the percentage value, calculated using the formula ; and respectively represent the number of similar products lower than the comprehensive promotion index corresponding to the current product and the total number of similar products.

[0011] In some embodiments, the specific calculation process of the comprehensive promotion index is: Extract the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product, and set the weight coefficients for the price advantage score, sales advantage score, and conversion advantage score corresponding to the new user; multiply the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product by the corresponding set weights respectively, and then sum them to obtain the comprehensive promotion index of the currently photographed product.

[0012] In some embodiments, in step M2, if the comprehensive promotion index of the currently photographed product is lower than the passing reference index range or the comprehensive promotion index of the product photographed by the current user is lower than that of similar products within a set percentage, then execute: Extract the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product within the previous set time window, denoted as the previous price advantage score, previous sales advantage score, and previous conversion advantage score; calculate the ratios of the price advantage score, sales advantage score, and conversion advantage score within the current set time window to the corresponding previous price advantage score, previous sales advantage score, and previous conversion advantage score respectively to obtain the price ratio, sales ratio, and conversion ratio. Set the weight coefficients for the price ratio, sales ratio, and conversion ratio corresponding to the new user, multiply the price ratio, sales ratio, and conversion ratio by the corresponding set weight coefficients respectively, and then sum them to obtain the potential index of the currently photographed product. After calculating the potential index, compare it with the corresponding set potential reference index. If the potential index of the currently photographed product is higher than the potential reference index, use the pre-constructed virtual AR character to voice broadcast that the currently photographed product is a "potential advantage product", then mark "The current product has increased significantly compared to the full data of the previous sales cycle", and attach a "rush to buy now" hot zone link.

[0013] In some embodiments, the corresponding steps are executed according to the user's user type, and it further includes: M3: If the current user is an old user, pop up a "preference option". After the old user selects the preference option for the currently photographed product, the preference options include price preference, sales preference, and conversion score; set a group of weight coefficient combinations corresponding to different preference options, and each of the three groups of weight coefficient combinations includes the weight coefficients corresponding to the price advantage score, sales advantage score, and conversion advantage score. After determining the weight coefficient combination, multiply the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product by the corresponding weight coefficients within the weight coefficient combination, and then sum them to obtain the preference promotion index. Extract the preference promotion index of the current product in the previous set time window and compare it with the preference promotion index of the current set time window. Use the pre-constructed virtual AR character to voice broadcast the comparison result. If the preference promotion index of the current set time window is higher than that of the previous set time window, attach the annotation "higher than the previous time zone" and the hot zone link "rush to buy now". Otherwise, directly attach the hot zone link "rush to buy now".

[0014] In some embodiments, the method further includes: Collection of product visual information: The user takes pictures of products in the offline scene through the device-side camera; Product content matching: Locate the product bounding box and identify the core visual elements, including the brand logo, product name, packaging color, and pattern; Extract the text information on the packaging; Combine the core visual elements and text information of the product into a matching data set; Platform data matching: Perform multi-dimensional matching of the combined matching data set with the cloud product database. The cloud product database stores the core visual elements and text information of various products and constitutes a storage data set; After multi-dimensional matching, associate the product category and ID photographed by the current user.

[0015] Compared with the prior art, the beneficial effects of the present invention are: The present invention quantitatively evaluates the advantages of products through three dimensions, evaluates price competitiveness through the competitiveness index, avoids the "only low price theory", integrates the repurchase contribution degree and sales growth degree, identifies word-of-mouth products, and quantitatively evaluates the depth of user participation by weighted calculation of interactive behaviors such as browsing, clicking, adding to cart, and evaluating. It excavates the potential value of products from different angles, provides data support for the optimization of subsequent visual interactions, and solves the problem that the prior art lacks full excavation of the potential value of products by using the promotion method of "only price theory"; The present invention divides users into newly registered users and old users, formulates different promotion strategies for different types of users and the comprehensive promotion index, finds differentiated selling points for top and waist products, makes the promotion content more targeted, and improves the resource utilization efficiency; For products with a low comprehensive promotion index but a high potential index, the present invention gives display opportunities by analyzing the growth trend in historical data, avoids being ignored due to short-term data disadvantages, helps merchants discover and promote potential products, and expands the market space; Through the technological innovation of multi-dimensional data evaluation, user stratification strategy, and scenario-based interaction, the present invention breaks through the single-dimensional limitation of the existing promotion methods, realizes the upgrade from passive promotion to precise adaptation, improves business efficiency through data-driven differentiated strategies, and finally achieves a win-win situation between user experience and business value. Brief Description of the Drawings

[0016] In the following description of exemplary embodiments in conjunction with the accompanying drawings, more details, features, and advantages of the present application are disclosed, in the drawings: Figure 1 This is a flowchart of the present invention. Detailed implementation manners

[0017] Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and for many different purposes and should not be limited to the embodiments set forth herein. These embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the application to those skilled in the art. The embodiments do not limit the present application.

[0018] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning that is consistent with their meaning in the relevant art and / or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0019] Please refer to Figure 1 As shown, a promotion method based on device-side visual interaction includes: Collection of commodity visual information: The user captures commodities in the offline scene through the device-side camera; for example, skin care products on the shelf, clothing tags, food packaging; and preprocesses the image, uses Gaussian filtering to remove image noise, extracts the commodity main area through threshold segmentation, reduces background interference (such as shelf labels, adjacent commodities), and performs perspective correction (based on homography matrix transformation) on the deformed or tilted commodity image to restore the standard perspective of the commodity and improve the subsequent recognition accuracy; Commodity content matching: Use object detection algorithms (such as YOLOv8, FasterR-CNN) to locate the commodity bounding box and identify the core visual elements, including brand LOGO, product name, packaging color, and pattern; combine OCR technology to extract the text information on the packaging, and the text information includes model number, specification, barcode / QR code, and supports multi-language mixed recognition (such as simultaneous parsing of Chinese and English packaging); combine the core visual elements and text information of the commodity into a matching data set; Platform data matching: Perform multi-dimensional matching of the combined matching data set with the cloud commodity database. The cloud commodity database stores the core visual elements and text information of various commodities and constitutes a storage data set; after multi-dimensional matching, associate the category and ID of the commodity captured by the current user; Interactive content generation: After associating with the ID of the product currently being photographed by the user; the ID includes the real-time price, discount information, and price trend in the past X days of the product; in the real-time screen of the device, the appearance of the product is superimposed through AR technology and the product price, discount information, and price trend in the past X days are marked in the form of floating labels, where X > 10, and the specific display duration is set by the technical staff; Intelligent promotion optimization: Based on the ID of the product currently being photographed by the user, retrieve the price data, sales data, and user data of the currently photographed product within a set time window before the current time point in the cloud product database, and perform analysis and evaluation after retrieval to determine the price advantage score, sales advantage score, and conversion advantage score of the product within the set time window before the current time point; Specifically: S1: Extract the price change of the currently photographed product within a set time window before the current time point from the price data. First, calculate the average value of each group of changed prices within the set time window as the price level value, and then obtain the product price of the currently photographed product at the current time point, denoted as the current price; Use the formula Calculate the competitiveness index of the currently photographed product at the current time point ; Competitiveness index > 0, indicating that the currently photographed product has an advantage in price, and the specific degree of the advantage is evaluated according to the size of the index; S2: Extract the total sales amount of the currently photographed product within a set time window before the current time point from the sales data. For the total sales amount within the set time window, divide the sales amount of repeat customers, and calculate the repeat purchase contribution degree of the currently photographed product within the set time window by dividing the sales amount of repeat customers by the total sales amount; Extract the total sales amount in the previous set time window as the denominator and the total sales amount in the current set time window as the numerator for ratio calculation, and calculate the sales volume growth rate of the currently photographed product within the set time window; For the repeat purchase contribution degree and sales volume growth rate of the currently photographed product, multiply them by the set weight coefficients respectively to obtain the sales advantage index of the currently photographed product; S3: Extract the number of views, clicks, add-to-cart amounts, and evaluation amounts of the currently photographed product within the set time window at the current time point from the user data, and mark them as a, b, c, and d respectively. Use the formula Perform weighted calculation to obtain the interactive participation index of the currently photographed product ; where g1, g2, g3, and g4 are the weight coefficients of the number of views, clicks, add-to-cart amounts, and evaluation amounts respectively; the specific values of the weight coefficients are set according to the importance of different interaction behaviors for product promotion; The view count is used as the denominator, representing "how many users have been exposed to the current product", while the numerator (click, add to cart, and evaluation behaviors) represents the number of users who have had effective interactions among the view counts; S4: Pre-build the competitiveness index , the sales advantage index, and the interaction participation index The corresponding range data sets respectively, and each range data set contains groups of index intervals associated with the competitiveness index , the sales advantage index, and the interaction participation index The respective associated groups of index intervals; Each group of index intervals associated with the competitiveness index Corresponds to a price advantage score respectively; The price advantage score range is set from 1 to 10 and is an integer. The larger the competitiveness index > 0 and the larger the value, the higher the possibility of corresponding to a match of 10. The smaller the competitiveness index < 0 and the larger the negative value, the higher the possibility of corresponding to a match of 1; Each group of index intervals associated with the sales advantage index corresponds to a sales advantage score respectively; The sales advantage score range is set from 1 to 10 and is an integer. The larger the value of the sales advantage index, the higher the possibility of corresponding to a match of 10, and the smaller the sales advantage index, the higher the possibility of corresponding to a match of 1; Each group of index intervals associated with the interaction participation index Corresponds to a conversion advantage score respectively; The conversion advantage score range is set from 1 to 10 and is an integer. The larger the value of the interaction participation index The higher the conversion rate during the product browsing process, the higher the possibility of corresponding to a match of 10, and vice versa, the higher the possibility of corresponding to a match of 1; Input the competitiveness index , the sales advantage index, and the interaction participation index of the currently photographed product at the current time point into the range data set for matching, and output the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product; Taking sunscreen as an example, assume that the current user photographs a certain brand of sunscreen (product ID: SPF50+PA++++), and sets the time window as the past 30 days. The specific calculation process is as follows: Calculation of the price advantage score 1. Data extraction Price level value: The price data for the past 30 days is [100 yuan, 100 yuan, 95 yuan, 95 yuan, 100 yuan,...] (a total of 30 groups of data), and the average is 98 yuan; Current price: The real-time price is 90 yuan when scanned by the user; 2. Calculation of competitiveness index 1 - (90 ÷ 98) ≈ 0.0816; 3. Matching of price advantage scores Range dataset definition: Competitiveness index ≥ 0.1 → score 9 0.05 ≤ index < 0.1 → score 8; 0 ≤ index < 0.05 → score 7; In this example, the competitiveness index is 0.0816, belonging to the range of 0.05 - 0.1, and the matching price advantage score is 8; Calculation of sales advantage scores 1. Data extraction Repeat purchase contribution: Total sales in the past 30 days: 100,000 yuan Sales of repeat purchase users: 30,000 yuan (repeat purchase users refer to users who have purchased this product repeatedly within 30 days); Repeat purchase contribution = 30,000 / 100,000 = 30% Sales growth rate: Sales in the previous period (the previous 30 days): 80,000 yuan Sales growth rate = 100,000 / 80,000 = 1.25 (a 25% increase) 2. Calculation of sales advantage index Weight setting: The weight of repeat purchase contribution is 40%, and the weight of sales growth rate is 60%; 0.3×0.4 + 0.25×0.6 = 0.27; 3. Matching of sales advantage scores Range dataset definition: Index ≥ 0.3 → score 10; 0.2 ≤ index < 0.3 → score 8; 0.1 ≤ index < 0.2 → score 6; In this example, the index is 0.27, belonging to the range of 0.2 - 0.3, and the matching sales advantage score is 8; Calculation of conversion advantage scores 1. Data extraction Page views (a): 1000 times; Click-through rate (b): 150 times (click on the product details page); Add-to-cart quantity (c): 50 times; Evaluation quantity (d): 30 times; 2. Calculation of interaction participation index Weight setting: g1 = 0.1 (page views, denominator), g2 = 1 (click), g3 = 2 (add to cart), g4 = 3 (evaluation) Calculated to be 3.4; 3. Transformation Advantage Score Matching Range dataset definition: Index ≥ 3.5 → Score 10; 2.5 ≤ Index < 3.5 → Score 8; 1.5 ≤ Index < 2.5 → Score 6; In this example, the index is 3.4, which belongs to the range of 2.5 - 3.5, and the matching transformation advantage score is 8; Promotion decision execution: Based on the price advantage score, sales advantage score, and transformation advantage score of the product photographed by the current user, execute corresponding steps to optimize the visual interaction scenario on the device side; Specifically: M1: Identify the user type to which the current user belongs, and execute step M2 or M3 accordingly based on the user type to which the user belongs; The user types include newly registered users and old users; M2: If the current user is a newly registered user, extract the price advantage score, sales advantage score, and transformation advantage score of the currently photographed product, and set the weight coefficients for the price advantage score, sales advantage score, and transformation advantage score corresponding to the new user; Multiply the price advantage score, sales advantage score, and transformation advantage score of the currently photographed product by the corresponding set weights respectively, and then sum to obtain the comprehensive promotion index of the currently photographed product; Set the passing reference index range for the comprehensive promotion index corresponding to each type of product. If the comprehensive promotion index of the currently photographed product is higher than the passing reference index range; Then use the pre - constructed virtual AR character to voice - broadcast that the currently photographed product is a "full - dimension advantage product", and then mark the specific value by which the comprehensive promotion index exceeds the passing reference index range and attach a "rush to buy now" hot - zone link; If the comprehensive promotion index of the currently photographed product is within the passing reference index range, compare the comprehensive promotion index of the product photographed by the current user with the comprehensive promotion index of similar products in the cloud product database. If the comparison result shows that the comprehensive promotion index of the product photographed by the current user is higher than a set percentage of similar products; The set percentage can be set to 80%; Then use the pre - constructed virtual AR character to voice - broadcast that the currently photographed product is a "similar - product advantage product", and then mark "The current product exceeds Y of similar products", and attach a "rush to buy now" hot - zone link; Where Y represents the percentage value, calculated using the formula ; and respectively represent the number of similar products lower than the comprehensive promotion index corresponding to the current product and the total number of similar products; If the comprehensive promotion index of the currently photographed product is lower than the qualified reference index range or the comprehensive promotion index of the product photographed by the current user is lower than the set percentage of similar products, the price advantage score, sales advantage score and conversion advantage score of the currently photographed product in the previous set time window will be advanced and recorded as the previous price advantage score, previous sales advantage score and previous conversion advantage score; Calculate the ratios of the price advantage score, sales advantage score and conversion advantage score in the current set time window to the corresponding previous price advantage score, previous sales advantage score and previous conversion advantage score to obtain the price ratio, sales ratio and conversion ratio; Set the weight coefficients of the price ratio, sales ratio and conversion ratio corresponding to the new user, multiply the price ratio, sales ratio and conversion ratio by the corresponding weight coefficients respectively, and then sum them up to obtain the potential index of the currently photographed product; The formula can be expressed as ,in Represents the potential index, Respectively represent the price advantage score, sales advantage score and conversion advantage score within the current set time window; They represent the former price advantage score, the former sales advantage score, and the former conversion advantage score respectively; Respectively represent the weight coefficients of price ratio, sales ratio and conversion ratio; After calculating the potential index, compare it with the corresponding potential reference index. If the potential index of the currently photographed product is higher than the potential reference index, use the pre-built virtual AR character voice to announce that the currently photographed product is a "potential advantage product", and then mark "the current product has a significant increase compared to the full data of the previous sales cycle", and attach a "buy now" hot zone link; Analyze the comprehensive promotion index of the product and adapt the promotion strategy, including: All-dimensional superior products (promotion index > passing reference index range): directly display high promotion index, strengthen the label of "all-dimensional superior products", and attract users who pursue quality; Relatively superior products (promotion index > 80% of similar products): Use the comparison phrase "more than Y% of similar products" to highlight the cost-effectiveness or advantages in a niche area, and adapt to price-sensitive users; Potential advantage products: dig out the growth trend (potential index) in historical data, and display the future value with the label of "potential advantage products" to avoid missing out on potential conversion opportunities due to short-term data disadvantages; This strategy solves the "one-size-fits-all" promotion drawback of existing technologies, allowing both top and mid-tier products to find differentiated selling points, and increasing the conversion rate of promotional content by more than 40%; M3: If the current user is an old user, a pop-up window "bias option" will appear. After the old user selects the bias option for the currently photographed product, the bias options include price bias, sales bias and conversion score; set different bias options to correspond to a set of weight coefficient combinations, and the three sets of weight coefficient combinations all include weight coefficients corresponding to price advantage score, sales advantage score and conversion advantage score; For example, within the price bias, the weight coefficient of the price advantage score in the corresponding weight coefficient combination should be higher than that of the other two groups; the same is true for sales bias and conversion bias, and the specific values ​​are set by technical personnel; After the weight coefficient combination is determined, the price advantage score, sales advantage score and conversion advantage score of the currently photographed product are multiplied by the corresponding weight coefficients in the weight coefficient combination, and then the sum is calculated to obtain the biased promotion index; At the same time, the biased promotion index of the current product in the previous set time window is extracted and compared with the biased promotion index of the current set time window. The comparison result is broadcasted using the pre-built virtual AR character voice. If the biased promotion index of the current set time window is higher than that of the previous set time window, it will be marked as "higher than the previous time zone" and accompanied by a "buy now" hot zone link. Otherwise, it will be directly accompanied by a "buy now" hot zone link. Through the three core innovations of user stratification operation, personalized weight configuration, and multi-dimensional dynamic evaluation, it breaks the limitation of unified standard promotion in existing technologies, not only meets the needs of new users for comprehensive understanding of products, but also gives old users the right to choose their preferences independently. At the same time, it activates the potential of mid-range products through data-driven differentiation strategies, and ultimately achieves a dual improvement in user experience and business efficiency. The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A promotion method based on device-side visual interaction, characterized in that: include: Interactive content generation: after associating the ID of the product photographed by the current user; the ID includes the product's real-time price, discount information, and price trends in the past X days; The AR technology is used to overlay the appearance of the product on the real-time screen of the device, and the product price, discount information and price trend in the past X days are marked in the form of floating labels, where X>10; Smart promotion optimization: Based on the ID of the product photographed by the current user, the price data, sales data and user data of the product photographed within the set time window before the current time point are retrieved from the cloud product database. After the retrieval, analysis and evaluation are performed to determine the price advantage score, sales advantage score and conversion advantage score of the product within the set time window before the current time point; Promotion decision execution: Based on the price advantage score, sales advantage score, and conversion advantage score of the product photographed by the current user, execute corresponding steps to optimize the scene of visual interaction on the device side.

2. A promotion method based on device-side visual interaction according to claim 1, characterized in that: Determine the price advantage score, sales advantage score, and conversion advantage score of the product within the set time window before the current time point, specifically: S1: Extract the price change of the currently photographed product within a set time window before the current time point from the price data. First, calculate the average of the price changes of each group within the set time window as the price level value. Then, obtain the product price of the currently photographed product at the current time point and record it as the current price. Using the formula Calculate the competitiveness index of the product being photographed at the current time point ; S2: extracting the total sales of the currently photographed product within a set time window before the current time point from the sales data, dividing the total sales within the set time window into the sales of repeat purchase users, dividing the sales of repeat purchase users by the total sales, and calculating the repeat purchase contribution of the currently photographed product within the set time window; The total sales amount in the previous set time window is taken as the denominator, and the total sales amount in the current set time window is taken as the numerator to calculate the ratio, and calculate the sales growth rate of the currently photographed product in the set time window; The repurchase contribution and sales growth of the currently photographed product are multiplied by the set weight coefficient to obtain the sales advantage index of the currently photographed product; S3: Extract the number of views, clicks, purchases and reviews of the currently photographed product within the set time window at the current time point from the user data, mark them as a, b, c and d respectively, and use the formula Perform weighted calculation to obtain the interactive participation index of the currently photographed product ; g1, g2, g3 and g4 are the weight coefficients of page views, clicks, added to cart and reviews respectively; S4: Based on pre-built conversion rules, the competitiveness index , sales advantage index and interactive participation index They are converted into price advantage score, sales advantage score and conversion advantage score respectively.

3. A promotion method based on device-side visual interaction according to claim 2, characterized in that: The specific transformation process of the transformation rule is as follows: Pre-built competitiveness index , sales advantage index and interactive participation index The corresponding scope data sets include the competitiveness index , sales advantage index and interactive participation index The index intervals of each group respectively associated; Competitiveness Index Each associated group of index intervals corresponds to a price advantage score; Each index interval associated with the sales advantage index corresponds to a sales advantage score; Each associated group of index intervals corresponds to a conversion advantage score; The competitiveness index of the currently photographed product at the current time point , sales advantage index and interactive participation index Matching is performed within the input range data set, and the price advantage score, sales advantage score, and conversion advantage score of the currently photographed product are output.

4. A promotion method based on device-side visual interaction according to claim 3, characterized in that: The corresponding steps are executed to optimize the scene of the visual interaction on the device side, specifically: M1: Identify the user type of the current user, and perform corresponding steps based on the user type to optimize the scene of visual interaction on the device; user types include newly registered users and old users.

5. A promotion method based on device-side visual interaction according to claim 4, characterized in that: The corresponding steps are executed based on the user type to which the user belongs, specifically: M2: If the current user is a newly registered user, calculate the comprehensive promotion index of the currently photographed product; set the passing reference index range for the comprehensive promotion index of each type of product; if the comprehensive promotion index of the currently photographed product is higher than the passing reference index range; use the pre-built virtual AR character voice to announce that the currently photographed product is a "full-dimensional advantage product", then mark the specific value of the comprehensive promotion index that is higher than the passing reference index range and attach a "buy now" hot zone link; If the comprehensive promotion index of the currently photographed product is within the qualified reference index range, the comprehensive promotion index of the product photographed by the current user is compared with the comprehensive promotion index of similar products in the cloud product database. If the comparison result shows that the comprehensive promotion index of the product photographed by the current user is higher than the set percentage of similar products, the pre-built virtual AR character voice broadcasts the current photographed product as a "similar superior product", and then marks "the current product exceeds Y similar products", and attaches a "buy now" hot zone link; where Y represents the percentage value, which is calculated using the formula ; and They respectively represent the number of similar products with a lower comprehensive promotion index than the current product and the total number of similar products.

6. A promotion method based on device-side visual interaction according to claim 5, characterized in that: The specific calculation process of the comprehensive promotion index is as follows: Extract the price advantage score, sales advantage score and conversion advantage score of the currently photographed product, set the weight coefficients of the price advantage score, sales advantage score and conversion advantage score corresponding to the new user; multiply the price advantage score, sales advantage score and conversion advantage score of the currently photographed product by the corresponding set weights respectively, and then sum them up to obtain the comprehensive promotion index of the currently photographed product.

7. A promotion method based on device-side visual interaction according to claim 6, characterized in that: In step M2, if the comprehensive promotion index of the currently photographed product is lower than the qualified reference index range or the comprehensive promotion index of the product photographed by the current user is lower than the set percentage of similar products, then execute: Extract the price advantage score, sales advantage score and conversion advantage score of the currently photographed product in the previous set time window, and record them as the previous price advantage score, previous sales advantage score and previous conversion advantage score; calculate the ratios of the price advantage score, sales advantage score and conversion advantage score in the current set time window with the corresponding previous price advantage score, previous sales advantage score and previous conversion advantage score, respectively, to obtain the price ratio, sales ratio and conversion ratio; Set the weight coefficients of the price ratio, sales ratio and conversion ratio corresponding to the new user, multiply the price ratio, sales ratio and conversion ratio by the corresponding weight coefficients respectively, and then sum them up to obtain the potential index of the currently photographed product; After the potential index is calculated, it is compared with the corresponding potential reference index. If the potential index of the currently photographed product is higher than the potential reference index, the pre-built virtual AR character voice broadcasts the currently photographed product as a "potential advantage product", and then marks "the current product has a significant increase compared to the full data of the previous sales cycle", and attaches a "buy now" hot zone link.

8. A promotion method based on device-side visual interaction according to claim 7, characterized in that: The step of executing corresponding steps based on the user type to which the user belongs also includes: M3: If the current user is an old user, a "bias option" window will pop up. After the old user selects the bias option for the currently photographed product, the bias options include price bias, sales bias and conversion score; set different bias options to correspond to a set of weight coefficient combinations, and the three sets of weight coefficient combinations all include weight coefficients corresponding to price advantage score, sales advantage score and conversion advantage score; After the weight coefficient combination is determined, the price advantage score, sales advantage score and conversion advantage score of the currently photographed product are multiplied by the corresponding weight coefficients in the weight coefficient combination, and then the sum is calculated to obtain the biased promotion index; At the same time, the biased promotion index of the current product in the previous set time window is extracted and compared with the biased promotion index of the current set time window. The comparison result is broadcasted using the pre-built virtual AR character voice. If the biased promotion index of the current set time window is higher than that of the previous set time window, it will be marked as "higher than the previous time zone" and accompanied by a "buy now" hot zone link; otherwise, it will be directly accompanied by a "buy now" hot zone link.

9. A promotion method based on device-side visual interaction according to claim 8, characterized in that: The promotion method also includes: Product visual information collection: Users use the device camera to take pictures of products in offline scenes; Product content matching: locate the product boundary box and identify the core visual elements, including brand logo, product name, packaging color and pattern; extract the text information on the packaging; combine the core visual elements and text information of the product into a matching data set; Platform data matching: The combined matching data set is matched with the cloud product database in multiple dimensions. The cloud product database stores the core visual elements and text information of various products and constitutes a storage data set. After multi-dimensional matching, the product category and ID photographed by the current user are associated.

Images