A method and system for accurately pushing personalized advertisements based on digital rental terminals

Abstract

The present invention discloses a method and system for accurate personalized advertising push based on a digital rental terminal, which relates to the field of smart door lock advertising technology, and includes the following steps: pre-collecting the advertising category, user identity information and image information of the advertising delivery scene according to the digital rental terminal, including: associating the digital rental terminal, user identity information and image information, and classifying the user according to the user identity information; performing face recognition on the acquired image information to obtain facial feature information. The present invention integrates advertising into the display screen of the smart door lock, improves the social benefits of the smart door lock, and obtains the user's emotions through real-time and accurate recognition of the user's face, and delivers a list of favorite advertisements of the corresponding user group according to the emotional feedback mechanism, thereby realizing personalized and accurate push adapted to the user group, improving the advertising conversion rate, and at the same time playing a leading role in the market application of smart door locks.

Description

Technical Field

[0001] The present invention relates to the field of smart door lock advertising technology, and in particular to a method and system for accurately pushing personalized advertisements based on a digital terminal for renting a house. Background Art

[0002] Key management for rental housing is a difficult problem. New tenants usually want to change the lock cylinder to increase security. Due to the high mobility of tenants, changing the lock cylinder and key management is a heavy task for the administrator of the rental housing. Therefore, various advanced smart door lock technologies such as biometric access control have emerged.

[0003] The homogeneity of smart door locks in the current market is becoming increasingly serious, and manufacturers can only make some minor innovations in appearance and materials. In terms of networking functions, they basically only provide some functions around door opening services, such as door opening message reporting and remote door opening, which makes it difficult for door lock manufacturers to gain differentiated competitive capabilities.

[0004] At present, traditional advertising is mainly focused on outdoor advertising, community elevator advertising, media advertising and other channels, and has not yet involved new IoT devices such as smart door locks. At the same time, networked smart locks on the market usually include display screens, door opening audio broadcasts, and door opening message push capabilities. These capabilities provide a new way for advertising dissemination, and the precise push of personalized advertising for smart door locks is a technical problem that needs to be solved urgently.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the problems in the related technology, the present invention proposes a method and system for accurately pushing personalized advertisements based on a digital rental terminal to overcome the above-mentioned technical problems existing in the existing related technology.

[0007] The technical solution of the present invention is achieved in this way:

[0008] In one aspect, the present invention:

[0009] A method and system for accurately pushing personalized advertisements based on a digital rental terminal, comprising the following steps:

[0010] Collecting the advertisement category, user identity information and image information of the advertisement delivery scene in advance according to the rental digital terminal, including: associating the rental digital terminal, user identity information and image information, and classifying the users according to the user identity information;

[0011] Performing face recognition on the acquired image information to obtain facial feature information, and identifying user emotions based on the facial feature information to determine whether the current advertisement is a user-satisfactory advertisement, wherein the facial feature information includes: eyebrow tilt value ETD, eye openness value EOD, lip-nose tip height ratio LNHR and mouth state;

[0012] If the current advertisement is satisfactory to the user, an advertisement recommendation list is generated according to the advertisement category of the current advertisement, and is delivered to the digital rental terminals of the same type as the current user.

[0013] The step of collecting the advertisement categories of the advertisement delivery scenarios includes the following steps:

[0014] Classifying the current advertisement includes the following steps:

[0015] Get creative copy and creative pictures for classified ads;

[0016] Extract features from creative texts to determine corresponding text features, and extract features from creative images to determine corresponding image features;

[0017] Determine the industry category of the current advertisement to be classified based on text features and image features.

[0018] The user identity information includes: gender, age, occupation, and rental location.

[0019] The step of performing face recognition on the acquired image information comprises the following steps:

[0020] Perform frame processing on the acquired image information to obtain each frame of image, and pre-process the image to obtain a grayscale image;

[0021] Extract feature points from the grayscale image, including: using a face detector to select a frame containing a face image as a recognition image, and using a face recognition frame to frame the face, and at the same time obtaining the face recognition frame width And face recognition frame height ; Use the trained Dlib face 68 key point detection model to calibrate the key points of the recognition image, expressed as: {1,2,3,4,······,65,66,67,68};

[0022] Obtain facial feature information based on the coordinates of the calibrated key points.

[0023] Wherein, obtaining facial feature information includes the following steps:

[0024] Obtaining the eyebrow inclination value ETD comprises the following steps:

[0025] According to the calibration key points, the left eyebrow calibration point array is obtained, which is expressed as: {18, 19, 20, 21, 22}; the right eyebrow calibration point array is expressed as: {23, 24, 25, 26, 27};

[0026] The left eyebrow calibration point array and the right eyebrow calibration point array are fitted with a straight line according to the horizontal and vertical coordinates of points {22} and {23} respectively, and the slope of the straight line obtained by fitting is used as the eyebrow inclination degree value ETD;

[0027] The eye opening degree value EOD is obtained, which is expressed as:

[0028] in, For face recognition frame height, is the average of the upper and lower orbital distances of the orbital calibration points;

[0029] Among them, get , expressed as:

[0030] Among them, {y 38 ,y 42},{y 39 ,y 41},{y 44 ,y 48} and {y 45 ,y 47} respectively represent four groups of upper and lower orbital key points obtained according to the calibration key points;

[0031] Get the lip-nose tip height ratio LNHR, expressed as:

[0032] in, The nose tip calibration point is the key point of calibration. The nose and upper lip calibration points are the key calibration points;

[0033] Get the mouth status, including:

[0034] The mouth width ratio MWR used to indicate the smiling state is expressed as:

[0035] in, is the face recognition frame width, They are the mouth width calibration points for the calibration key points;

[0036] The mouth height ratio MHR used to indicate the open state is expressed as:

[0037] in, They are the mouth height calibration points for the calibration key points respectively.

[0038] The step of judging whether the current advertisement delivery is a satisfactory advertisement for the user comprises the following steps:

[0039] Pre-determine whether the current user's mouth state is open, and compare the mouth height ratio MHR with the mouth height threshold, where;

[0040] If the mouth height ratio MHR>mouth height threshold, it means that the current user's mouth state is open, and the eyebrow tilt degree value ETD is compared with the eyebrow threshold. If the eyebrow tilt degree value ETD≥eyebrow threshold, it means that the current emotion is angry;

[0041] If the eyebrow inclination value ETD is less than the eyebrow threshold, the lip-nose tip height ratio LNHR is compared with the lip-nose threshold, where;

[0042] If the lip-nose tip height ratio LNHR is less than the lip-nose threshold, it means the current emotion is happy;

[0043] If the lip-nose tip height ratio LNHR ≥ lip-nose threshold, it means the current emotion is surprise;

[0044] If the mouth height ratio MHR ≤ the mouth height threshold, it means that the current user's mouth is not open, then the eyebrow tilt degree value ETD is compared with the eyebrow threshold, wherein, if the eyebrow tilt degree value ETD ≥ the eyebrow threshold, it means that the current emotion is angry;

[0045] If the eyebrow tilt value ETD is less than the eyebrow threshold, the mouth width ratio MWR is compared with the mouth width threshold, where;

[0046] If the mouth width ratio MWR>mouth width threshold, it means the current emotion is happy;

[0047] If the mouth width ratio MWR ≤ mouth width threshold, then compare the eye opening degree value EOD with the eye threshold, where;

[0048] If the eyes openness value EOD>eye threshold, it means the current emotion is surprise;

[0049] If the eye opening degree value EOD ≤ eye threshold, it means that the current emotion is natural;

[0050] The emotion of being marked as happy indicates that the user is satisfied with the current advertising, and the emotion of being marked as angry indicates that the user is dissatisfied with the current advertising.

[0051] The step of delivering the information to the same type of rental digital terminals of the current user includes the following steps:

[0052] The user provides command feedback to the advertisement of the current rental digital terminal, wherein the command feedback includes: like or detail viewing.

[0053] Another aspect of the present invention is:

[0054] A personalized advertising precision push system based on rental digital terminals, including:

[0055] The collection module is used to collect the advertisement category, user identity information and image information of the advertisement delivery scene according to the rental digital terminal;

[0056] A face recognition module is used to perform face recognition on the acquired image information to obtain facial feature information;

[0057] Advertisement placement judgment module, which identifies user emotions based on facial feature information and determines whether the current advertisement is satisfactory to the user;

[0058] The precise delivery module is used to generate an advertisement recommendation list according to the advertisement category of the current advertisement if the current advertisement is satisfactory to the user, and deliver it to the digital rental terminal of the same type as the current user.

[0059] Furthermore, the acquisition module includes: an advertisement classification unit and a user classification unit, wherein;

[0060] The advertisement classification unit is used to obtain the creative text and creative pictures of the advertisement to be classified; extract features from the creative text to determine the corresponding text features, and extract features from the creative pictures to determine the corresponding picture features; determine the industry category of the current advertisement to be classified based on the text features and the picture features;

[0061] The user classification unit is used to classify users according to user identity information.

[0062] Furthermore, the facial feature information includes: eyebrow tilt feature, eye open feature, lip and nose tip height feature and mouth state feature.

[0063] Beneficial effects of the present invention:

[0064] The present invention is based on the precise push of personalized advertisements by digital house rental terminals. By pre-collecting the advertisement category, user identity information and image information of the advertisement delivery scene according to the digital house rental terminals, the digital house rental terminals, user identity information and image information are associated, and the users are classified according to the user identity information. At the same time, face recognition is performed on the acquired image information to acquire facial feature information, and the user's emotion is identified according to the facial feature information to judge whether the current advertisement delivery is a user-satisfied advertisement. If the current advertisement delivery is a user-satisfied advertisement, an advertisement recommendation list is generated according to the advertisement category of the current advertisement delivery, and the advertisement is delivered to the digital house rental terminals of the same type as the current user. Not only the advertisement is integrated into the display screen of the smart door lock, thereby improving the social benefits of the smart door lock, but also the user's emotion is acquired through real-time and precise recognition of the user's face, and a list of advertisements liked by the corresponding group of users is delivered according to the emotion feedback mechanism, thereby realizing personalized and precise push adapted to the user group, improving the advertisement conversion rate, and playing a leading role in the market application of smart door locks. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0066] Figure 1 It is a flow chart of a method for accurately pushing personalized advertisements based on a digital rental terminal according to an embodiment of the present invention;

[0067] Figure 2 68 key point position distribution diagram of a method for accurately pushing personalized advertisements based on a digital rental terminal according to an embodiment of the present invention;

[0068] Figure 3 The present invention is a block diagram of a personalized advertising precision push system based on a digital rental terminal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0070] According to an embodiment of the present invention, a method for accurately pushing personalized advertisements based on a digital terminal for renting a house is provided.

[0071] like Figure 1 As shown, according to an embodiment of the present invention, a method for accurately pushing personalized advertisements based on a rental digital terminal includes the following steps:

[0072] Step S1, collecting the advertisement category, user identity information and image information of the advertisement delivery scene in advance according to the rental digital terminal, which includes: associating the rental digital terminal, user identity information and image information, and classifying the user according to the user identity information;

[0073] For the digital terminal for renting a house, this technical solution can be a digital smart door lock for renting a house. Its advertisement is placed on the display screen of the smart door lock, and the image information can be collected through the camera installed on the smart door lock. Its user identity information is the information that needs to be registered by the smart door lock when renting a house, that is, the user identity information is bound to the digital terminal for renting a house.

[0074] Specifically, user identity information includes: gender, age, occupation, and rental location. Users are classified according to their identity information. For example, users with the same gender, "age ± 4 years old", the same occupation, and rental location within a 10-kilometer radius of the center are classified into the same category.

[0075] In addition, when applied, adaptive adjustments can be made according to the amount of data classified by the user, for example, adaptively adjusting the proportion of classification categories and reducing classification categories, etc., so as to increase the amount of user data after classification.

[0076] Step S2, performing face recognition on the acquired image information, acquiring facial feature information, and identifying user emotions based on the facial feature information to determine whether the current advertisement is a satisfactory advertisement for the user, wherein the facial feature information includes: eyebrow tilt value ETD, eye openness value EOD, lip-nose tip height ratio LNHR and mouth state;

[0077] Wherein, obtaining facial feature information includes the following steps:

[0078] Step S201, performing frame processing on the acquired image information to obtain each frame of image, and preprocessing the image to obtain a grayscale image;

[0079] In this technical solution, the image can be pre-processed using median filtering to filter out interference factors such as random noise in the recognition image, and the image is gray-scaled to obtain a gray-scale image. The gray-scale image obtained after pre-processing reduces irrelevant information, reduces the amount of data, enhances the detectability of facial feature information, and improves the efficiency of recognition operations.

[0080] Step S202, extracting feature points from the grayscale image, including:

[0081] The face detector is used to select a frame containing a face image as the recognition image, and the face is framed by the face recognition frame, and the face recognition frame width is obtained. And face recognition frame height ;

[0082] like Figure 2 As shown, the trained Dlib face 68 key point detection model is used to calibrate the key points of the recognition image, which are expressed as: {1,2,3,4,······,65,66,67,68};

[0083] Obtain facial feature information based on the coordinates of the calibrated key points.

[0084] Among them, obtain facial feature information, specifically:

[0085] Step S203, obtaining the eyebrow inclination value ETD, comprises the following steps:

[0086] According to the calibration key points, the left eyebrow calibration point array is obtained, which is expressed as: {18, 19, 20, 21, 22}; the right eyebrow calibration point array is expressed as: {23, 24, 25, 26, 27};

[0087] The left eyebrow calibration point array and the right eyebrow calibration point array are fitted with a straight line according to the horizontal and vertical coordinates of points {22} and {23} respectively, and the slope of the straight line obtained by fitting is used as the eyebrow inclination degree value ETD;

[0088] When this technical solution is applied, when a person is angry or furious, the facial muscles often pull the eyebrows down and tighten them inward, and the brows are tightly furrowed, and vertical wrinkles appear between the eyebrows. By analyzing the tilt of the front part of the eyebrows, the difference between being angry and not angry can be intuitively compared. Therefore, by setting the eyebrow threshold, the tilt of the front part of the eyebrows can be used as the basis for judging the angry expression.

[0089] Step S204, obtaining the eye opening degree value EOD, expressed as:

[0090] in, For face recognition frame height, is the average of the upper and lower orbital distances of the orbital calibration points;

[0091] Among them, get , expressed as:

[0092] Among them, {y 38 ,y 42},{y 39 ,y 41},{y 44 ,y 48} and {y 45 ,y 47} are respectively represented by four groups of upper and lower eye socket key points obtained according to the calibration key points; when this technical solution is applied, when the expression changes, the eyes will also change accordingly, which is mainly reflected in the shape of the eyes opening, and its change is defined as the degree of eye opening, using the distance between the upper and lower eye sockets Height with face recognition frame The ratio is expressed as the eye opening degree value EOD.

[0093] Step S205, obtaining the lip-nose tip height ratio LNHR, expressed as:

[0094] in, The nose tip calibration point is the key point of calibration. The nose and upper lip calibration points are used to calibrate the key points. In this technical solution, when the nose moves less during the expression change, the "immobile" nose tip and the "moving" upper lip movement are combined, and the distance between the two is combined with the face recognition frame height. The ratio of is used as the lip-nose tip height ratio LNHR to recognize facial expressions.

[0095] Step S206, obtaining the mouth state, which includes:

[0096] The mouth width ratio MWR used to indicate the smiling state is expressed as:

[0097] in, is the face recognition frame width, They are the mouth width calibration points for the calibration key points;

[0098] The mouth height ratio MHR used to indicate the open state is expressed as:

[0099] in, They are the mouth height calibration points for the calibration key points respectively.

[0100] In addition, judging whether the current advertisement delivery is a satisfactory advertisement for the user includes the following steps:

[0101] Pre-determine whether the current user's mouth state is open, and compare the mouth height ratio MHR with the mouth height threshold, where;

[0102] Step S207, if the mouth height ratio MHR>mouth height threshold, it indicates that the current user's mouth state is open, and the eyebrow tilt degree value ETD is compared with the eyebrow threshold, wherein, if the eyebrow tilt degree value ETD≥eyebrow threshold, it indicates that the current emotion is angry;

[0103] If the eyebrow inclination value ETD is less than the eyebrow threshold, the lip-nose tip height ratio LNHR is compared with the lip-nose threshold, where;

[0104] If the lip-nose tip height ratio LNHR is less than the lip-nose threshold, it means the current emotion is happy;

[0105] If the lip-nose tip height ratio LNHR ≥ lip-nose threshold, it means the current emotion is surprise.

[0106] Step S208, if the mouth height ratio MHR ≤ the mouth height threshold, it means that the current user's mouth is not open, then the eyebrow tilt degree value ETD is compared with the eyebrow threshold, wherein, if the eyebrow tilt degree value ETD ≥ the eyebrow threshold, it means that the current emotion is angry;

[0107] If the eyebrow tilt value ETD is less than the eyebrow threshold, the mouth width ratio MWR is compared with the mouth width threshold, where;

[0108] If the mouth width ratio MWR>mouth width threshold, it means the current emotion is happy;

[0109] If the mouth width ratio MWR ≤ mouth width threshold, then compare the eye opening degree value EOD with the eye threshold, where;

[0110] If the eyes openness value EOD>eye threshold, it means the current emotion is surprise;

[0111] If the eyes openness value EOD ≤ the eye threshold, it means that the current emotion is natural.

[0112] According to the technical solution, the above emotions can be divided into angry, happy, surprised and natural, among which the happy emotion can indicate that the user is satisfied with the current advertising, the angry emotion can indicate that the user is dissatisfied with the current advertising, and the surprised and natural emotions indicate that the user has no attitude towards the current advertising.

[0113] In addition, when applying, the relevant values ​​can be referred to as follows: the eyebrow threshold is -0.30~-0.40, the eye threshold is 0.05~0.06, the lip and nose threshold is 0.20~0.30, the mouth width threshold is 0.30~0.40, and the mouth height threshold is 0.05~0.06.

[0114] Step S3: if the current advertisement is satisfactory to the user, an advertisement recommendation list is generated according to the advertisement category of the current advertisement, and the advertisement is delivered to the digital rental terminal of the same type as the current user.

[0115] The step of delivering the information to the same type of rental digital terminals of the current user includes the following steps:

[0116] The user provides command feedback to the advertisement of the current rental digital terminal, wherein the command feedback includes: like or detail viewing.

[0117] This technical solution generates an advertisement recommendation list according to the advertisement category of the current advertisement delivery, including the following steps:

[0118] Categorize ads in advance.

[0119] For the classification of the above-mentioned advertisements, the present technical solution may, for example, obtain the creative text and creative pictures of the advertisements to be classified; perform feature extraction on the creative texts to determine the corresponding text features; perform feature extraction on the creative pictures to determine the corresponding picture features; and determine the industry category of the current advertisement to be classified based on the text features and picture features.

[0120] Specifically, for example, text features can be obtained by using OCR text recognition software, and creative images can be obtained by using image recognition software. For example, if the creative image is a water cup, the outline of the water cup is recognized by image recognition software. Then, the text features and image features of different advertisements are compared respectively, and the current advertisement is classified according to the comparison results.

[0121] In addition, for the above-mentioned advertisement recommendation list, the advertisements that are more similar to the current advertisement are given priority for delivery.

[0122] According to an embodiment of the present invention, a personalized advertisement precision push system based on a digitalized rental terminal is provided.

[0123] like Figure 3 As shown, according to an embodiment of the present invention, a personalized advertising precision push system based on a rental digital terminal includes:

[0124] The collection module is used to collect the advertisement category, user identity information and image information of the advertisement delivery scene according to the rental digital terminal;

[0125] A face recognition module is used to perform face recognition on the acquired image information to obtain facial feature information;

[0126] Advertisement placement judgment module, which identifies user emotions based on facial feature information and determines whether the current advertisement is satisfactory to the user;

[0127] The precise delivery module is used to generate an advertisement recommendation list according to the advertisement category of the current advertisement if the current advertisement is satisfactory to the user, and deliver it to the digital rental terminal of the same type as the current user.

[0128] In addition, the acquisition module includes: an advertisement classification unit and a user classification unit, wherein;

[0129] The advertisement classification unit is used to obtain the creative text and creative pictures of the advertisement to be classified; extract the features of the creative text to determine the corresponding text features, and extract the features of the creative pictures to determine the corresponding picture features; determine the industry category of the current advertisement to be classified based on the text features and picture features;

[0130] The user classification unit is used to classify users according to user identity information.

[0131] In addition, facial feature information includes: eyebrow tilt features, eye open features, lip and nose tip height features, and mouth state features.

[0132] To sum up, with the help of the above-mentioned technical scheme of the present invention, by pre-collecting the advertising category, user identity information and image information of the advertising delivery scene according to the rental digital terminal, the rental digital terminal, user identity information and image information are associated, and the user is classified according to the user identity information, and at the same time, the acquired image information is subjected to face recognition to obtain facial feature information, and the user's emotions are identified according to the facial feature information to determine whether the current advertising delivery is a user-satisfied advertisement. If the current advertising delivery is a user-satisfied advertisement, an advertising recommendation list is generated according to the advertising category of the current advertising delivery, and the advertisement is delivered to the same type of rental digital terminal of the current user. Not only is the advertisement integrated into the display screen of the smart door lock, thereby improving the social benefits of the smart door lock, but also the user's emotions are obtained through real-time and accurate recognition of the user's face, and a list of advertisements liked by the corresponding group of users is delivered according to the emotional feedback mechanism, thereby realizing personalized and accurate push adapted to the user group, improving the advertising conversion rate, and playing a leading role in the market application of smart door locks.

[0133] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. After considering the disclosure of the specification and the examples, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and the examples are only regarded as exemplary, and the true scope and spirit of the present disclosure are indicated by the claims.

[0134] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for accurately pushing personalized advertisements based on a digital rental terminal, comprising the following steps: Collecting the advertisement category, user identity information and image information of the advertisement delivery scene in advance according to the rental digital terminal, including: associating the rental digital terminal, user identity information and image information, and classifying the users according to the user identity information; Performing face recognition on the acquired image information to obtain facial feature information, and identifying user emotions based on the facial feature information to determine whether the current advertisement is a user-satisfactory advertisement, wherein the facial feature information includes: eyebrow tilt value ETD, eye openness value EOD, lip-nose tip height ratio LNHR and mouth state; If the current advertisement is satisfactory to the user, an advertisement recommendation list is generated according to the advertisement category of the current advertisement, and is delivered to the digital rental terminals of the same type as the current user; Obtaining facial feature information includes the following steps: Obtaining the eyebrow inclination value ETD comprises the following steps: According to the calibration key points, the left eyebrow calibration point array is obtained, which is expressed as: {18, 19, 20, 21, 22}; the right eyebrow calibration point array is expressed as: {23, 24, 25, 26, 27}; The left eyebrow calibration point array and the right eyebrow calibration point array are fitted with a straight line according to the horizontal and vertical coordinates of points {22} and {23} respectively, and the slope of the straight line obtained by fitting is used as the eyebrow inclination degree value ETD; The eye opening degree value EOD is obtained, which is expressed as: ; in, For face recognition frame height, is the average of the upper and lower orbital distances of the orbital calibration points; Among them, get , expressed as: ; Among them, {y 38 ,y 42 },{y 39 ,y 41 },{y 44 ,y 48 } and {y 45 ,y 47 } respectively represent four groups of upper and lower orbital key points obtained according to the calibration key points; Get the lip-nose tip height ratio LNHR, expressed as ; in,{ } is the nose tip calibration point of the calibration key point, { } is the nose and upper lip calibration point of the calibration key point; Get the mouth status, including: The mouth width ratio MWR used to indicate the smiling state is expressed as: ; in, is the face recognition frame width, They are the mouth width calibration points for the calibration key points; The mouth height ratio MHR used to indicate the open state is expressed as: in, They are the mouth height calibration points for the calibration key points respectively.

2. The method for accurately pushing personalized advertisements based on a digital terminal for renting a house according to claim 1 is characterized in that: The step of collecting the advertisement category of the advertisement delivery scenario includes the following steps: Classifying the current advertisement includes the following steps: Get creative copy and creative pictures for classified ads; Extract features from creative texts to determine corresponding text features, and extract features from creative images to determine corresponding image features; Determine the industry category of the current advertisement to be classified based on text features and image features.

3. The method for accurately pushing personalized advertisements based on a digital terminal for renting a house according to claim 2 is characterized in that: The user identity information includes: gender, age, occupation, and rental location.

4. The method for accurately pushing personalized advertisements based on a digital terminal for renting a house according to claim 1 is characterized in that: The step of performing face recognition on the acquired image information includes the following steps: Perform frame processing on the acquired image information to obtain each frame of image, and pre-process the image to obtain a grayscale image; Extract feature points from the grayscale image, including: using a face detector to select a frame containing a face image as a recognition image, and using a face recognition frame to frame the face, and at the same time obtaining the face recognition frame width And face recognition frame height ; Use the trained Dlib face 68 key point detection model to calibrate the key points of the recognition image, expressed as: {1,2,3,4······65,66,67,68}; Obtain facial feature information based on the coordinates of the calibrated key points.

5. The method for accurately pushing personalized advertisements based on a digital terminal for renting a house according to claim 1 is characterized in that: The step of judging whether the current advertisement delivery is a satisfactory advertisement for the user comprises the following steps: Pre-determine whether the current user's mouth state is open, and compare the mouth height ratio MHR with the mouth height threshold, where; If the mouth height ratio MHR>mouth height threshold, it means that the current user's mouth state is open, and the eyebrow tilt degree value ETD is compared with the eyebrow threshold. If the eyebrow tilt degree value ETD≥eyebrow threshold, it means that the current emotion is angry; If the eyebrow inclination value ETD is less than the eyebrow threshold, the lip-nose tip height ratio LNHR is compared with the lip-nose threshold, where; If the lip-nose tip height ratio LNHR is less than the lip-nose threshold, it means the current emotion is happy; If the lip-nose tip height ratio LNHR ≥ lip-nose threshold, it means the current emotion is surprise; If the mouth height ratio MHR ≤ the mouth height threshold, it means that the current user's mouth is not open, then the eyebrow tilt degree value ETD is compared with the eyebrow threshold, wherein, if the eyebrow tilt degree value ETD ≥ the eyebrow threshold, it means that the current emotion is angry; If the eyebrow tilt value ETD is less than the eyebrow threshold, the mouth width ratio MWR is compared with the mouth width threshold, where; If the mouth width ratio MWR>mouth width threshold, it means the current emotion is happy; If the mouth width ratio MWR ≤ mouth width threshold, then compare the eye opening degree value EOD with the eye threshold, where; If the eyes openness value EOD>eye threshold, it means the current emotion is surprise; If the eye opening degree value EOD ≤ eye threshold, it means that the current emotion is natural; The emotion of being marked as happy indicates that the user is satisfied with the current advertising, and the emotion of being marked as angry indicates that the user is dissatisfied with the current advertising.

6. The method for accurately pushing personalized advertisements based on a digital terminal for renting a house according to claim 1 is characterized in that: The steps include: The user provides command feedback to the advertisement of the current rental digital terminal, wherein the command feedback includes: like or detail viewing.

7. A system for accurately pushing personalized advertisements based on a digital rental terminal, used in the method for accurately pushing personalized advertisements based on a digital rental terminal according to any one of claims 1 to 6, comprising: The collection module is used to collect the advertisement category, user identity information and image information of the advertisement delivery scene according to the rental digital terminal; A face recognition module is used to perform face recognition on the acquired image information to obtain facial feature information; Advertisement placement judgment module, which identifies user emotions based on facial feature information and determines whether the current advertisement is satisfactory to the user; The precise delivery module is used to generate an advertisement recommendation list according to the advertisement category of the current advertisement if the current advertisement is satisfactory to the user, and deliver it to the digital rental terminal of the same type as the current user.

8. The personalized advertising precision push system based on rental digital terminal according to claim 7 is characterized in that: The acquisition module includes: an advertisement classification unit and a user classification unit, wherein; The advertisement classification unit is used to obtain the creative text and creative pictures of the advertisement to be classified; extract features from the creative text to determine the corresponding text features, and extract features from the creative pictures to determine the corresponding picture features; determine the industry category of the current advertisement to be classified based on the text features and the picture features; The user classification unit is used to classify users according to user identity information.

9. The personalized advertising precision push system based on rental digital terminal according to claim 7 is characterized in that: The facial feature information includes: eyebrow tilt feature, eye open feature, lip and nose tip height feature, and mouth state feature.

Images