An advertisement pushing method for internet sale
Patent Information
- Application Number
- CN202611117987.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-27
- Publication Date
- 2026-10-09
AI Technical Summary
这种割裂的推送方式，使得高创意价值的广告易被置于低关注时段，而平庸素材却占据优质流量窗口，既降低了广告库存的整体利用效率，也使广告主预算难以获得稳定可预期的回报
[0012]本发明的有益效果：（1）本发明通过在地面划分等腰梯形检测区域并结合行人移动速度的实时动态变化，构建了一套无需人脸识别、不侵犯个人隐私的广告吸引力评估机制。具体而言，利用行人进入检测区域后的速度降幅比例，将行人自然划分为“正常移动”“减速关注”与“停留观看”三种细粒度状态，并进一步通过第一、第二权重系数对减速与停留行为进行差异化加权，从而计算出每个广告独立的吸引度系数X。优势在于：将传统的曝光量统计转化为有效注意力留存统计——只有真正被广告内容吸引并产生行为反馈的行人才被纳入评价体系，有效剔除了单纯路过或无效流量带来的数据噪声。同时，检测区域采用LED屏垂直投影结合屏幕物理尺寸进行几何约束，确保采集范围始终处于行人能够清晰观看屏幕内容的有效视距内，进一步提升了吸引度系数的客观性与信噪比。与现有固定轮播或单一点击率排序方案相比，本发明能够实时捕获每一则广告在线下场景中的吸引力强弱，为后续的时段匹配提供了精准、可信且符合隐私合规要求的数据基础，解决了背景技术中缺乏广告本身吸引力实时量化手段的长期技术痛点。
Smart Images
Figure CN122887553A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of advertising push technology, and specifically to an advertising push method for internet sales. Background Technology
[0002] Internet advertising revenue heavily relies on the synergistic matching of ad content appeal and the traffic value of the time slot. However, existing push mechanisms mostly employ static ranking strategies based on fixed rotation or single exposure, failing to establish a dynamic relationship between content quality and time slot conversion potential. Under the current model, ad slots are allocated broadly based on chronological order or historical performance, lacking real-time quantification of ad appeal and fine-grained evaluation of audience engagement and deep interaction within a given time slot. This fragmented push approach results in highly creative ads being placed in low-attention slots, while mediocre content occupies prime traffic windows, reducing the overall utilization efficiency of ad inventory and making it difficult for advertisers to achieve stable and predictable returns. These technical deficiencies prevent the simultaneous quantification and incorporation of ad content appeal and effective traffic into push decisions, leading to significant fluctuations in revenue and becoming a key bottleneck hindering the refined operation and intelligent delivery of internet advertising. Summary of the Invention
[0003] The purpose of this invention is to provide an advertising push method for internet sales, thereby solving the above-mentioned technical problems.
[0004] The objective of this invention can be achieved through the following technical solutions: A method for pushing advertisements for internet sales includes the following steps: S1: The detection area is divided on the ground area according to the LED screen in the square. When a pedestrian enters the detection area, the pedestrian's moving speed v is obtained. The pedestrian's moving state is divided based on the pedestrian's moving speed v. The moving state includes normal movement, deceleration and attention, and stopping to watch. S2: Obtain the playback time period of advertisement A on the LED screen, filter all pedestrians entering the detection area during the playback time period, and obtain the number N of pedestrians remaining for advertisement A. A The number of pedestrians remaining, N A This refers to the number of pedestrians in the detection area whose movement is slowed down to focus or who stop to watch during the playback period; S3: Number of pedestrians retained based on ad A (N) A Calculate the attractiveness coefficient X of advertisement A, repeat the above steps to calculate the attractiveness coefficient of the remaining advertisements, and sort all advertisements in descending order of attractiveness coefficient; S4: Obtain the preset advertising time periods and calculate the effective traffic coefficient M within each advertising time period. This includes the following steps: Obtain the number of unique visitors (UV) during the advertising period, and calculate the traffic coefficient σ based on the number of unique visitors (UV); Calculate the effective flow coefficient Where λ represents the preset flow coefficient adjustment value, and μ represents the preset correction value; S5: Sort the advertising time slots in descending order of effective traffic coefficient, and push the i-th advertisement in the i-th advertising time slot for playback. Repeat the above steps to push the advertisements in sequence.
[0005] As a further aspect of the present invention: In step S1, the detection area is divided into two zones based on the method of the plaza LED screen on the ground: The detection area is an isosceles trapezoidal region extending outward along the normal direction of the screen from the reference point on the ground, with the center point of the LED screen vertically projected onto it. The near boundary of the isosceles trapezoidal region is parallel to the lower edge of the LED screen and the distance is L, where L is the preset minimum viewing distance. The far boundary is parallel to the upper edge of the LED screen. The distance H between the upper and lower edges of the LED screen is obtained. Let the distance between the far boundary and the near boundary of the isosceles trapezoidal region be η×H, where η represents the preset magnification factor and η>1.
[0006] As a further aspect of the present invention: In step S2, the method for classifying the pedestrian's movement state based on the pedestrian's moving speed v is as follows: Obtain the mean v of the pedestrian's moving speed. avg Given the standard deviation s, calculate the threshold proportion R = 2s / v avg ; When a pedestrian enters the detection area, if the decrease in movement speed is less than or equal to R, it is recorded as normal movement; if the decrease in movement speed is greater than R but less than 1, it is recorded as deceleration and attention; if the pedestrian's movement speed becomes 0, it is recorded as stopping and watching.
[0007] As a further aspect of the present invention: In step S3, the number N of pedestrians remaining based on advertisement A is... A Method for calculating the attractiveness coefficient X of advertisement A: Obtain the total number N of pedestrians passing through the detection area during the playback period of advertisement A. all Calculate the attraction coefficient X = α1 × (N) A1 / N all )+α2×[(N A -N A1 ) / N all ], where N A1 The number of pedestrians to be slowed down is represented by α1 and α2, which represent the preset first and second weighting coefficients, respectively, with 0 < α1 < α2 < 1.
[0008] As a further aspect of the present invention: the duration T of a pedestrian's single stay within the detection area is obtained; if the duration T of a single stay is less than a preset minimum duration T... min The pedestrian is marked, and the marked pedestrian is not included in the total number N. all middle.
[0009] As a further aspect of the present invention: in step S3, if there are advertisements with equal attraction coefficients, the advertisement with shorter playback time is placed in a higher ranking position.
[0010] As a further aspect of the present invention: In step S4, the method for calculating the traffic coefficient σ based on the number of unique visitors (UV) is as follows: Get the order rate Y=I1 / UV, payment rate Z=I2 / I1, and payment completion rate W=I3 / I2 of visitors during the advertising period, where I1 represents the number of users who placed orders, I2 represents the number of users who made payments, I3 represents the number of users who successfully paid, and payment without refund is counted as payment success; Calculate the flow coefficient ,in, .
[0011] As a further aspect of the present invention: in step S5, when there are advertising periods with equal effective traffic coefficients, the total page views within each advertising period are obtained, and the advertising period with higher total page views is placed in a higher sorting position.
[0012] The beneficial effects of the present invention are as follows: (1) The present invention constructs an advertising attractiveness evaluation mechanism that does not require face recognition and does not infringe on personal privacy by dividing the ground into isosceles trapezoidal detection areas and combining the real-time dynamic changes of pedestrian movement speed. Specifically, by using the speed reduction ratio of pedestrians after entering the detection area, pedestrians are naturally divided into three fine-grained states: "normal movement", "deceleration and attention" and "stopping to watch". Furthermore, the deceleration and stopping behaviors are differentiated and weighted by the first and second weight coefficients, thereby calculating the independent attractiveness coefficient X of each advertisement. The advantage is that the traditional exposure statistics are transformed into effective attention retention statistics - only pedestrians who are truly attracted by the advertising content and generate behavioral feedback are included in the evaluation system, effectively eliminating the data noise caused by simply passing by or invalid traffic. At the same time, the detection area adopts the vertical projection of the LED screen combined with the physical size of the screen for geometric constraints, ensuring that the collection range is always within the effective viewing distance where pedestrians can clearly see the screen content, further improving the objectivity and signal-to-noise ratio of the attractiveness coefficient. Compared with existing fixed carousel or single click-through rate ranking schemes, this invention can capture the attractiveness of each advertisement in offline scenarios in real time, providing an accurate, reliable and privacy-compliant data foundation for subsequent time-slot matching, and solving the long-standing technical pain point of lacking real-time quantification of the attractiveness of advertisements in the background technology.
[0013] (2) A two-way quantitative and collaborative matching framework for advertising content value and time-period traffic value was constructed, breaking through the limitations of traditional static ranking strategies. In terms of time-period value assessment, the solution not only introduces unique visitor count (UV) as the traffic base, but also innovatively integrates three funnel conversion indicators: order rate (Y), payment rate (Z), and payment completion rate (W). The standard deviation σ of the three indicators is calculated to quantify the fluctuation of traffic quality within a time period, and an effective traffic coefficient M is constructed based on the natural logarithm and exponential decay function. The technical essence of this design is: the smaller σ is (the more stable the conversion funnel), the higher the e -σ The closer the coefficient is to 1, the higher the effective traffic coefficient, thus identifying high-quality prime time slots that are not only populated but also have stable and predictable purchasing behavior. In terms of matching decisions, this invention arranges ads in descending order of attractiveness coefficient X and time slots in descending order of effective traffic coefficient M, and performs dynamic, optimized push notifications. This one-to-one priority mapping mechanism ensures that ads with high creative value occupy high-quality, high-conversion-stability traffic windows first, while mediocre creatives are automatically relegated to inefficient time slots. Through a unified quantitative dimension and sorting alignment strategy, this invention significantly reduces the marginal benefit mismatch loss of ad inventory, enabling advertisers to obtain stable and predictable returns on their budget investments, while simultaneously improving the platform's overall effective display revenue, achieving a technological leap in refined operation and intelligent delivery of internet advertising. Attached Figure Description
[0014] The invention will now be further described with reference to the accompanying drawings.
[0015] Figure 1 This is a schematic diagram of the structure of an advertising push method for internet sales according to the present invention; Figure 2 This is a flowchart illustrating an internet sales advertising push method according to the present invention. Detailed Implementation
[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.
[0017] Please see Figure 1 As shown, this invention is a method for pushing advertisements for internet sales, comprising the following steps: S1: The detection area is divided on the ground area according to the LED screen in the square. When a pedestrian enters the detection area, the pedestrian's moving speed v is obtained. The pedestrian's moving state is divided based on the pedestrian's moving speed v. The moving state includes normal movement, deceleration and attention, and stopping to watch. S2: Obtain the playback time period of advertisement A on the LED screen, filter all pedestrians entering the detection area during the playback time period, and obtain the number N of pedestrians remaining for advertisement A. A The number of pedestrians remaining, N A This refers to the number of pedestrians in the detection area whose movement is slowed down to focus or who stop to watch during the playback period; S3: Number of pedestrians retained based on ad A (N) A Calculate the attractiveness coefficient X of advertisement A, repeat the above steps to calculate the attractiveness coefficient of the remaining advertisements, and sort all advertisements in descending order of attractiveness coefficient; S4: Obtain the preset advertising time periods and calculate the effective traffic coefficient M within each advertising time period. This includes the following steps: Obtain the number of unique visitors (UV) during the advertising period, and calculate the traffic coefficient σ based on the number of unique visitors (UV); Calculate the effective flow coefficient Where λ represents the preset flow coefficient adjustment value, and μ represents the preset correction value; S5: Sort the advertising time slots in descending order of effective traffic coefficient, and push the i-th advertisement in the i-th advertising time slot for playback. Repeat the above steps to push the advertisements in sequence.
[0018] It should be noted that this invention provides an advertising push method for internet sales, aiming to solve the technical problem of low push efficiency caused by the inability to quantify the attractiveness of advertising content and the traffic value of the time period in existing advertising push mechanisms. This method is applied to squares or outdoor venues equipped with LED displays. By dividing specific pedestrian detection zones on the ground, it collects pedestrian movement behavior data in real time and simultaneously obtains traffic conversion indicators for online advertising time periods. Ultimately, it establishes a two-way ranking and matching mechanism between content attractiveness and time period traffic value, enabling precise tilting of high-attractive advertisements towards high-quality time periods.
[0019] First, to assess the attractiveness of the advertising content, an isosceles trapezoidal detection area is defined on the ground directly in front of the LED screen. Centered on a reference point where the center of the LED screen is vertically projected onto the ground, the area extends outwards along the screen's normal direction. The near boundary is parallel to the lower edge of the LED screen and at a preset minimum viewing distance L, while the far boundary is determined by multiplying the screen height H by a magnification factor η, ensuring the entire detection area covers the optimal viewing distance range for pedestrians to clearly see the screen content. When a pedestrian enters this detection area, the system uses visual sensors or radar speed measurement devices deployed within the area to acquire the pedestrian's moving speed v in real time. To eliminate judgment bias caused by differences in individual walking habits, this invention introduces an adaptive speed threshold mechanism, first calculating the average moving speed v of all pedestrians within the detection area. avg And the standard deviation s, and calculate the threshold proportion R = 2s / v avg Based on this, pedestrian movement is categorized into three types: if a pedestrian's speed decreases by less than or equal to R after entering the area, they are considered unaffected by the advertising content and classified as normal movement; if their speed decreases by more than R but has not completely stopped, it indicates they are interested in the advertising content and have slowed down, and are classified as slowing down to pay attention; if their speed drops to zero, it indicates they are deeply attracted to the advertisement and have stopped to watch, and are classified as stopping to watch. It should be noted that in practical applications, when pedestrian traffic is sparse or speed fluctuations are too large, causing the R value to be abnormally high, an upper limit can be set for R to avoid the unreasonable situation where a speed decrease must exceed 100% to be considered slowing down. Simultaneously, for pedestrians whose single stop duration within the detection area is too short, the system will mark them and exclude them from subsequent statistics to eliminate data noise caused by occasional pauses.
[0020] Based on this, for a specific advertisement A, the system obtains its complete playback time on the LED screen and filters all pedestrians entering the detection area during that time. The system then extracts the number of pedestrians whose movement status is slowed down to show interest and those who stopped to watch; the sum of these two is the pedestrian retention count N for advertisement A. A Furthermore, to quantify the attractiveness coefficient X of advertisement A, we simultaneously obtain the total number N of pedestrians passing through the detection area during the playback period. total The attractiveness coefficient X is calculated using a weighted summation method, which involves multiplying the first weighted coefficient by the proportion of those who slowed down to the total number of viewers, and then adding the second weighted coefficient multiplied by the proportion of those who remained on screen. Since remaining on screen represents a deeper level of attention investment, the second weighted coefficient should be greater than the first weighted coefficient for easier subsequent ranking. All ads to be pushed are iterated through using the above method, their respective X values are calculated, and an ad sequence is generated in descending order.
[0021] Secondly, regarding the evaluation of traffic value during advertising periods, this invention obtains several preset advertising periods and independently calculates the effective traffic coefficient M for each period. This calculation process integrates two factors: the scale of visitors and the stability of conversion quality within the period. Specifically, the system first obtains the number of unique visitors (UV) within that period, and simultaneously counts the number of users placing orders, making payments, and successfully paying within that period, thereby calculating three funnel conversion indicators: order rate Y, payment rate Z, and payment completion rate W. To measure the volatility of the conversion process within that period, this invention calculates the standard deviation of Y, Z, and W. The smaller the standard deviation, the more stable and predictable the user conversion behavior is during that period; conversely, a larger standard deviation indicates significant fluctuations in traffic quality and a higher risk of ad placement. When constructing the effective traffic coefficient M, the natural logarithm of UV is used as the base term for traffic scale, multiplied by two correction factors: the first correction factor is (1 + λ × e) / ... -σ The factor is in the form of (1+λ), where λ is a preset adjustment value. When σ is small, the factor approaches (1+λ), thus amplifying the effect. When σ is large, e -σ Approaching 0, the factor approaches 1, and no longer provides additional gain; the second correction factor introduces the gap between the weakest and strongest links in the conversion funnel, μ is a preset correction value, and this term further penalizes or compensates for periods of unbalanced conversion structure. Combining the product of the above three parts, the effective flow coefficient M for each period is obtained, and the periods are sorted from high to low according to the M value.
[0022] Finally, the aforementioned list of ads sorted in descending order of attractiveness coefficient is matched one-to-one with the list of time slots sorted in descending order of effective traffic coefficient. That is, the ad ranked first is pushed to the time slot ranked first, the ad ranked second is pushed to the time slot ranked second, and so on. When ads with equal attractiveness coefficients appear, the ad with shorter playback duration is prioritized to improve ad rotation efficiency per unit time. When time slots with equal effective traffic coefficients appear, the total page views within the time slot are further compared, and the time slot with higher page views is ranked first to fully utilize high exposure windows. Furthermore, the compatibility between ad duration and time slot length must be considered during the actual push process. If an ad duration exceeds the remaining idle time of the matched time slot, it is automatically postponed to the next cycle for re-matching to ensure physical playback feasibility. Through the above-described quantitative perception and dynamic sorting throughout the entire process, this invention achieves precise matching between ad content attractiveness and time slot traffic value, significantly improving the overall utilization efficiency of ad inventory and the stability of advertisers' return on investment.
[0023] In another preferred embodiment of the present invention, the method for dividing the detection area in the ground area using a plaza LED screen is as follows: The detection area is an isosceles trapezoidal region extending outward along the normal direction of the screen from the reference point on the ground, with the center point of the LED screen vertically projected onto it. The near boundary of the isosceles trapezoidal region is parallel to the lower edge of the LED screen and the distance is L, where L is the preset minimum viewing distance. The far boundary is parallel to the upper edge of the LED screen. The distance H between the upper and lower edges of the LED screen is obtained. Let the distance between the far boundary and the near boundary of the isosceles trapezoidal region be η×H, where η represents the preset magnification factor and η>1.
[0024] It is worth noting that by using the vertical projection of the LED screen's center point as a reference and extending along the normal direction to form an isosceles trapezoidal area, the detection range is precisely aligned with the screen's geometric position and the pedestrian's natural line of sight. This ensures that speed change data is only included in the data collection system when the pedestrian is within an effective viewing distance where they can clearly see the screen content. This avoids the dual drawbacks of detection areas that are too close, leading to misjudgment before the pedestrian can clearly see the content, and too far, causing data interference from irrelevant passersby. Furthermore, it can adapt to LED screens of different sizes and specifications. Through this geometric constraint, the data source for subsequent determination of deceleration, attention, and viewing status is fundamentally guaranteed to be authentic and reliable, making the quantification result of the attraction coefficient X more objective and accurate.
[0025] In another preferred embodiment of the present invention, the method for classifying the pedestrian movement state based on the pedestrian's moving speed v is as follows: Obtain the mean v of the pedestrian's moving speed. avg Given the standard deviation s, calculate the threshold proportion R = 2s / v avg ; When a pedestrian enters the detection area, if the decrease in movement speed is less than or equal to R, it is recorded as normal movement; if the decrease in movement speed is greater than R but less than 1, it is recorded as deceleration and attention; if the pedestrian's movement speed becomes 0, it is recorded as stopping and watching.
[0026] Understandably, by introducing the mean and standard deviation of pedestrian speed to construct an adaptive threshold ratio R, the classification of movement states no longer depends on fixed speed values, but is determined based on the deceleration magnitude of an individual relative to the walking characteristics of their own group, which can adapt to different pedestrian flow densities and walking habits.
[0027] In another preferred embodiment of the present invention, the number of pedestrians retained based on advertisement A is N. A Method for calculating the attractiveness coefficient X of advertisement A: Obtain the total number N of pedestrians passing through the detection area during the playback period of advertisement A. all Calculate the attraction coefficient X = α1 × (N) A1 / N all )+α2×[(N A -N A1 ) / N all ], where NA1 The number of pedestrians to be slowed down is represented by α1 and α2, which represent the preset first and second weighting coefficients, respectively, with 0 < α1 < α2 < 1.
[0028] In another preferred embodiment of the present invention, the duration T of a single stop by a pedestrian within the detection area is obtained. If the duration T of a single stop is less than a preset minimum duration T... min The pedestrian is marked, and the marked pedestrian is not included in the total number N. all middle.
[0029] It should be noted that a single stay duration threshold T is introduced. min This filter eliminates brief pauses within the detection area caused by non-advertising-related factors. In real-world scenarios, pedestrians may briefly pause within the detection area for accidental reasons such as tying shoelaces, checking their phones, talking to companions, or waiting for others. These actions are not caused by advertising content and are therefore included in the pedestrian retention count N. all This will directly pollute the data source for calculating the attractiveness coefficient X, causing highly attractive ads to be underestimated or mediocre ads to be overvalued.
[0030] In another preferred embodiment of the present invention, if there are advertisements with equal attractiveness coefficients, the advertisement with shorter playback time is placed in a higher ranking position.
[0031] Understandably, when multiple ads have the same attractiveness coefficient, using playback duration as a secondary ranking factor effectively breaks the ranking deadlock and ensures the uniqueness and determinism of the ranking sequence. Prioritizing the playback of shorter ads can increase the frequency of ad rotation per unit time under the same content attractiveness conditions, increase the absolute number of ads played during prime time slots, thereby improving the time utilization efficiency of ad inventory and the throughput capacity of the overall push system, achieving a dual optimization of the completeness of the ranking rules and commercial benefits.
[0032] In another preferred embodiment of the present invention, a method for calculating the traffic coefficient σ based on the number of unique visitors (UV) is as follows: Get the order rate Y=I1 / UV, payment rate Z=I2 / I1, and payment completion rate W=I3 / I2 of visitors during the advertising period, where I1 represents the number of users who placed orders, I2 represents the number of users who made payments, I3 represents the number of users who successfully paid, and payment without refund is counted as payment success; Calculate the flow coefficient ,in, .
[0033] In another preferred embodiment of the present invention, when there are advertising periods with equal effective traffic coefficients, the total page views within each advertising period are obtained, and the advertising period with higher total page views is placed in a higher sorting position.
[0034] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.
Claims
1. A method for pushing advertisements for internet sales, characterized in that, Includes the following steps: S1: The detection area is divided on the ground area according to the LED screen in the square. When a pedestrian enters the detection area, the pedestrian's moving speed v is obtained. The pedestrian's moving state is divided based on the pedestrian's moving speed v. The moving state includes normal movement, deceleration and attention, and stopping to watch. S2: Obtain the playback time period of advertisement A on the LED screen, filter all pedestrians entering the detection area during the playback time period, and obtain the number N of pedestrians remaining for advertisement A. A The number of pedestrians remaining, N A This refers to the number of pedestrians in the detection area whose movement is slowed down to focus or who stop to watch during the playback period; S3: Number of pedestrians retained based on ad A (N) A Calculate the attractiveness coefficient X of advertisement A, repeat the above steps to calculate the attractiveness coefficient of the remaining advertisements, and sort all advertisements in descending order of attractiveness coefficient; S4: Obtain the preset advertising time periods and calculate the effective traffic coefficient M within each advertising time period. This includes the following steps: Obtain the number of unique visitors (UV) during the advertising period, and calculate the traffic coefficient σ based on the number of unique visitors (UV); Calculate the effective flow coefficient Where λ represents the preset flow coefficient adjustment value, and μ represents the preset correction value; S5: Sort the advertising time slots in descending order of effective traffic coefficient, and push the i-th advertisement in the i-th advertising time slot for playback. Repeat the above steps to push the advertisements in sequence.
2. The method for pushing advertisements for internet sales according to claim 1, characterized in that, In step S1, the detection area is divided according to the method of the square LED screen in the ground area: The detection area is an isosceles trapezoidal region extending outward along the normal direction of the screen from the reference point on the ground, with the center point of the LED screen vertically projected onto it. The near boundary of the isosceles trapezoidal region is parallel to the lower edge of the LED screen and the distance is L, where L is the preset minimum viewing distance. The far boundary is parallel to the upper edge of the LED screen. The distance H between the upper and lower edges of the LED screen is obtained. Let the distance between the far boundary and the near boundary of the isosceles trapezoidal region be η×H, where η represents the preset magnification factor and η>1.
3. The method for pushing advertisements for internet sales according to claim 1, characterized in that, In step S2, the method for classifying the pedestrian's movement state based on the pedestrian's moving speed v is as follows: Obtain the mean v of the pedestrian's moving speed. avg Given the standard deviation s, calculate the threshold proportion R = 2s / v avg ; When a pedestrian enters the detection area, if the decrease in moving speed is less than or equal to R, it is recorded as normal movement; if the decrease in moving speed is greater than R but less than 1, it is recorded as deceleration and attention. If a pedestrian's movement speed becomes 0, it is recorded as stopping to watch.
4. The method for pushing advertisements for internet sales according to claim 1, characterized in that, In step S3, the number N of pedestrians remaining based on advertisement A is... A Method for calculating the attractiveness coefficient X of advertisement A: Obtain the total number N of pedestrians passing through the detection area during the playback period of advertisement A. all Calculate the attraction coefficient X = α1 × (N) A1 / N all )+α2×[(N A -N A1 ) / N all ], where N A1 The number of pedestrians to be slowed down is represented by α1 and α2, which represent the preset first and second weighting coefficients, respectively, with 0 < α1 < α2 < 1.
5. The method for pushing advertisements for internet sales according to claim 4, characterized in that, Obtain the duration T of a single stop for a pedestrian within the detection area. If the duration T is less than the preset minimum duration T... min The pedestrian is marked, and the marked pedestrian is not included in the total number N. all middle.
6. The method for pushing advertisements for internet sales according to claim 1, characterized in that, In step S3, if there are advertisements with the same attractiveness coefficient, the advertisement with the shorter playback time will be placed in the higher position in the ranking.
7. The method for pushing advertisements for internet sales according to claim 1, characterized in that, In step S4, the method for calculating the traffic coefficient σ based on the number of unique visitors (UV) is as follows: Get the order rate Y=I1 / UV, payment rate Z=I2 / I1, and payment completion rate W=I3 / I2 of visitors during the advertising period, where I1 represents the number of users who placed orders, I2 represents the number of users who made payments, I3 represents the number of users who successfully paid, and payment without refund is counted as payment success; Calculate the flow coefficient ,in, .
8. The method for pushing advertisements for internet sales according to claim 1, characterized in that, In step S5, when there are advertising periods with equal effective traffic coefficients, the total page views within each advertising period are obtained, and the advertising period with the higher total page views is placed in the higher sorting position.