An advertising material content recognition method and system

Abstract

An advertisement material content recognition method and system provided by the present invention. The advertisement material content recognition method includes: using recognition technology to recognize advertisement materials and outputting multiple recognition results; standardizing the multiple recognition results according to an advertisement resource metadata standard library; fusing the multiple standardized recognition results; and outputting the content information of the advertisement materials according to the fusion result. Based on the advertisement resource metadata standard library, this method outputs a recognition result with a unified standard by fusing multiple recognition results, improving the accuracy of advertisement material content recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of advertisement recognition, and particularly relates to a method and a system for recognizing advertisement material content. Background Art

[0002] Advertisement materials contain rich information, such as the products, brands, corresponding industries, advertisers, spokespersons of advertisements, etc. appearing in the advertisement materials. Obtaining these information in the advertisement materials will help commercial analysis institutions obtain commercial information such as the operation status and market segmentation of enterprises.

[0003] The forms of advertisement materials are generally pictures and videos. To obtain the content in the pictures or videos, in the prior art, advertisement materials are generally recognized through optical character recognition (OCR) and automatic speech recognition (ASR) to obtain information such as the products and brands appearing in the advertisement materials.

[0004] However, simply through optical character recognition (OCR) and automatic speech recognition (ASR), the content information in the advertisement materials cannot be well recognized. On the one hand, there will be a lot of interference information in the recognized content information, which affects the recognition accuracy. For example, some common words such as "apple", "boss", and some simple English letters are also the names of products or brands, which will interfere with the recognition results; on the other hand, there will also be a situation where the recognized products and brands do not match, that is, the recognized products do not belong to the recognized brands, which will lead to inconsistent recognition results. Summary of the Invention

[0005] To solve the technical problems of low accuracy in recognizing advertisement material content and inconsistent recognition results in the prior art, the present invention provides a method for recognizing advertisement material content. Based on an advertisement resource metadata standard library, by fusing multiple recognition results, a standard and unified recognition result is output, improving the accuracy of recognizing advertisement material content.

[0006] The present invention provides a method for recognizing advertisement material content, including:

[0007] S1: Using recognition technology to recognize the advertisement material and output multiple recognition results;

[0008] S2: Standardizing the multiple recognition results according to the advertisement resource metadata standard library;

[0009] S3: Fusing the multiple standardized recognition results;

[0010] S4: Outputting the content information of the advertisement material according to the fusion result of S3.

[0011] Further, before the step S1, it further includes:

[0012] S0: Collect the monitoring information of the advertisement material. When the monitoring information is collected, output the content information of the advertisement material according to the monitoring information and the advertisement resource metadata standard library; when the monitoring information is not collected, execute steps S1 to S4.

[0013] Further, in step S1, the recognition technologies include: character recognition technology, speech recognition technology, logo recognition technology, and face recognition technology.

[0014] Further, in step S2, it specifically includes:

[0015] S21: Standardize the character recognition result and the speech recognition result according to the advertisement resource metadata standard library;

[0016] S22: Standardize the logo recognition result according to the advertisement resource metadata standard library;

[0017] S23: Standardize the face recognition result according to the advertisement resource metadata standard library.

[0018] Further, in step S21, the method for standardizing the character recognition result and the speech recognition result specifically includes:

[0019] S211: Determine the product words and / or brand words in the character recognition result and the speech recognition result whose confidence levels are greater than or equal to the first preset threshold;

[0020] S212: When the character recognition result and the speech recognition result contain product words, take the product words and the brand words to which the product words belong;

[0021] S213: When the character recognition result and the speech recognition result only contain brand words, take the brand words;

[0022] S214: When the character recognition result and the speech recognition result contain blacklist words, filter the blacklist words;

[0023] S215: When the character recognition result and the speech recognition result contain product alias words and / or brand alias words, take the product words corresponding to the product alias words and / or the brand words corresponding to the brand alias words;

[0024] S216: When the character recognition result and the speech recognition result contain multiple product words and / or brand words, determine the product words and / or the brand words according to the phrase relationship, the number of product words, and the number of brand words.

[0025] Further, in the step S216, the method for determining the product word and / or the brand word specifically includes:

[0026] When there is a phrase relationship in the text recognition result and the speech recognition result, take the product word and / or the brand word included in the long-short phrase relationship;

[0027] When there is no phrase relationship in the text recognition result and the speech recognition result, count the occurrences of each product word and / or brand word. If the occurrences of each product word and / or brand word are different, take the product word and / or brand word with the most occurrences;

[0028] If the occurrences of each product word and / or brand word are the same, take the product word and / or brand word with the highest confidence.

[0029] Further, in the step S22, the method for standardizing the logo recognition result specifically includes:

[0030] S221: Map the logo recognition result to the advertising resource metadata standard library for matching;

[0031] S222: When the logo recognition result is completely matched with the advertising resource metadata standard library, take the mapped brand word.

[0032] Further, in the step S23, the method for standardizing the face recognition result specifically includes:

[0033] S231: Determine the spokesperson name in the face recognition result whose confidence is greater than or equal to the second preset threshold;

[0034] S232: Map the spokesperson name to the advertising resource metadata standard library for matching;

[0035] S233: When the spokesperson name is completely matched with the advertising resource metadata standard library, take the mapped spokesperson name and the information of the spokesperson.

[0036] Further, in the step S3, the method for fusing the multiple standardized recognition results specifically includes:

[0037] S31: Determine the trust relationship among the standardized text recognition result, the speech recognition result, and the logo recognition result;

[0038] S32: When the standardized text recognition result, the speech recognition result, and the logo recognition result are inconsistent, fuse the multiple standardized recognition results according to the trust relationship.

[0039] The present invention also provides an advertising material content recognition system, including:

[0040] A recognition unit, configured to recognize the advertising material by using recognition technology and output a plurality of recognition results;

[0041] A processing unit, configured to perform standardized processing on the plurality of recognition results according to an advertising resource metadata standard library;

[0042] A fusion unit, configured to fuse the plurality of recognition results after standardized processing;

[0043] An advertising content output unit, configured to output the content information of the advertising material according to the fusion result of the fusion unit.

[0044] Technical effects or advantages of the present invention:

[0045] (1) An advertising material content recognition method provided by the present invention recognizes an advertising material by using recognition technology, performs standardized processing on a plurality of recognition results according to an advertising resource metadata standard library, fuses the plurality of recognition results after standardized processing, and outputs the content information of the advertising material according to the fusion result. Based on the advertising resource metadata standard library, this method outputs a recognition result with a unified standard by fusing a plurality of recognition results, improving the accuracy of advertising material content recognition.

[0046] (2) An advertising material content recognition method provided by the present invention recognizes an advertising material by using character recognition technology, speech recognition technology, logo recognition technology, and face recognition technology based on an advertising resource metadata standard library. It can not only recognize the product information, brand information, and spokesperson information of the advertising material, but also obtain industry information, advertiser information, etc. by combining the recognized product information, brand information, and spokesperson information, enriching the content information output of the advertising material and facilitating better acquisition of valuable information in the advertising material.

[0047] (3) An advertising material content recognition method provided by the present invention collects monitoring information of the advertising material before recognizing the advertising material by using recognition technology. When the monitoring information is collected, the content information of the advertising material is output according to the monitoring information and the advertising resource metadata standard library, further ensuring the accuracy of the output result. Description of the Drawings

[0048] Figure 1 It is a flowchart of an advertising material content recognition method provided by an embodiment of the present invention;

[0049] Figure 2 It is a flowchart of standardizing and processing a logo recognition result provided by an embodiment of the present invention;

[0050] Figure 3 A flowchart for standardizing and processing the face recognition results provided by an embodiment of the present invention;

[0051] Figure 4 A schematic diagram of an advertisement material content recognition system provided by an embodiment of the present invention. Detailed implementation manners

[0052] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0053] It should be understood by those in the art that in the description of the present invention, the terms "first", "second", "third", etc. are only used for descriptive distinction and cannot be construed as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0054] To solve the technical problems of low accuracy and inconsistent recognition results in the prior art for advertisement material content recognition, the present invention provides an advertisement material content recognition method. Based on an advertisement resource metadata standard library, by fusing multiple recognition results, a standard and unified recognition result is output, improving the accuracy of advertisement material content recognition.

[0055] The technical solutions of the present invention will be described in detail below with reference to specific embodiments and the accompanying drawings of the specification.

[0056] Refer to Figure 1 , Figure 1 A flowchart of an advertisement material content recognition method provided by an embodiment of the present invention. The present invention provides an advertisement material content recognition method, including:

[0057] S0: Collect the monitoring information of the advertisement material. When the monitoring information is collected, according to the monitoring information and the advertisement resource metadata standard library, output the content information of the advertisement material; when the monitoring information is not collected, execute steps S1 to S4. In this embodiment, according to the monitoring information and the advertisement resource metadata standard library, output the content information of the advertisement material, ensuring the accuracy of the output result.

[0058] Among them, the advertising resource metadata standard library is common knowledge in the art. In this embodiment, the advertising resource metadata standard library can also be constructed and modified according to actual needs. The advertising resource metadata standard library includes, but is not limited to, a product word library, a brand word library, a logo material library, a spokesperson name library, a blacklist word library, a product alias word library, a brand alias word library, an industry word library, and an advertiser word library. The monitoring information of the advertising materials is collected by a crawler. If the collected monitoring information can parse out an activity ID, according to the activity ID, the content information of the corresponding advertising material can be obtained from the advertising resource metadata standard library. The content information of the advertising material includes, but is not limited to, product information, brand information, industry information, spokesperson information, and advertiser information.

[0059] S1: Use recognition technology to recognize the advertising material and output multiple recognition results. Among them, the recognition technology includes: text recognition technology, speech recognition technology, logo recognition technology, and face recognition technology. The recognition results include text recognition results, speech recognition results, logo recognition results, and face recognition results.

[0060] In this embodiment, the text recognition technology, speech recognition technology, logo recognition technology, and face recognition technology used are common knowledge in the art. According to actual needs, those skilled in the art can select text recognition technology, speech recognition technology, logo recognition technology, and face recognition technology from different manufacturers. Based on the advertising resource metadata standard library, this embodiment uses text recognition technology, speech recognition technology, logo recognition technology, and face recognition technology to recognize advertising materials, which can not only recognize the product information, brand information, and spokesperson information of advertising materials, but also combine the recognized product information, brand information, and spokesperson information to obtain industry information, advertiser information, etc., enriching the content information output by advertising materials and facilitating better acquisition of valuable information in advertising materials.

[0061] It should be noted that the text recognition technology can recognize the product words and brand words in the advertising material, the speech recognition technology can also recognize the product words and brand words in the advertising material, the logo recognition technology can recognize the brand words in the advertising material, and the face recognition technology can recognize the spokesperson name in the advertising material.

[0062] S2: Standardize multiple recognition results according to the advertising resource metadata standard library. Among them, step S2 further includes:

[0063] S21: Standardize the text recognition results and speech recognition results according to the advertising resource metadata standard library;

[0064] S22: Standardize the logo recognition results according to the advertising resource metadata standard library;

[0065] S23: Standardize the face recognition result according to the advertising resource metadata standard library.

[0066] It should be noted that there is no sequence requirement among steps S21, S22, and S23.

[0067] Since the text recognition result is similar to the speech recognition result, the text recognition result and the speech recognition result are standardized using the same method. Among them, in step S21, the method for standardizing the text recognition result and the speech recognition result is as follows:

[0068] S211: Determine the product words and / or brand words in the text recognition result and the speech recognition result whose confidence levels are greater than or equal to the first preset threshold.

[0069] S212: When the text recognition result and the speech recognition result contain product words, select the product words and the brand words to which the product words belong.

[0070] S213: When the text recognition result and the speech recognition result only contain brand words, select the brand words.

[0071] S214: When the text recognition result and the speech recognition result contain blacklist words, filter out the blacklist words.

[0072] S215: When the text recognition result and the speech recognition result contain product alias words and / or brand alias words, select the product words corresponding to the product alias words and / or the brand words corresponding to the brand alias words.

[0073] S216: When the text recognition result and the speech recognition result contain multiple product words and / or brand words, determine the product words and / or brand words according to the phrase relationship, the frequency of product words, and the frequency of brand words.

[0074] It should be noted that there is no sequence requirement among steps S212, S213, S214, S215, and S216.

[0075] Among them, in step S216, the method for determining the product words and / or the brand words is as follows:

[0076] When the text recognition result and the speech recognition result contain a phrase relationship, select the product words and / or brand words included in the long and short phrase relationships.

[0077] When the text recognition result and the speech recognition result do not contain a phrase relationship, count the frequency of each product word and / or brand word. If the frequency of each product word and / or brand word is different, select the product word and / or brand word with the highest frequency.

[0078] If the number of occurrences of each product word and / or brand word is the same, select the product word and / or brand word with the highest confidence.

[0079] Reference Figure 2 , in step S22, the method for standardizing the logo recognition result is carried out according to the following steps:

[0080] S221: Map the logo recognition result to the advertising resource metadata standard library for matching;

[0081] S222: When the logo recognition result completely matches the advertising resource metadata standard library, take the mapped brand word.

[0082] Reference Figure 3 , in step S23, the method for standardizing the face recognition result is carried out according to the following steps:

[0083] S231: Determine the name of the spokesperson in the face recognition result whose confidence is greater than or equal to the second preset threshold. In this embodiment, the second preset threshold is 95;

[0084] S232: Map the name of the spokesperson to the advertising resource metadata standard library for matching;

[0085] S233: When the name of the spokesperson completely matches the advertising resource metadata standard library, take the mapped name of the spokesperson and the information of the spokesperson.

[0086] S3: Fuse the multiple standardized recognition results. Among them, in step S3, the method for fusing the multiple standardized recognition results is carried out according to the following steps:

[0087] S31: Determine the trust relationship among the text recognition result, the speech recognition result, and the logo recognition result after standardization;

[0088] S32: When there is an inconsistency among the text recognition result, the speech recognition result, and the logo recognition result after standardization, fuse the multiple standardized recognition results according to the trust relationship.

[0089] Since the logo recognition result after standardization only contains the brand word, the logo recognition result after standardization is more targeted than the text recognition result and the speech recognition result after standardization, and the text recognition result after standardization is more accurate than the speech recognition result. Therefore, in this embodiment, the trust relationship among the text recognition result, the speech recognition result, and the logo recognition result after standardization is that the trust degree of the logo recognition result after standardization is greater than the trust degree of the text recognition result after standardization, and the trust degree of the text recognition result after standardization is greater than the trust degree of the speech recognition result after standardization.

[0090] Since the text recognition results after standardization may contain product words and brand words, and the speech recognition results after standardization may also contain product words and brand words, while the logo recognition results after standardization only contain brand words, in step S32, the fusion of multiple recognition results after standardization according to the confidence relationship can be carried out as follows:

[0091] S321: When the text recognition result, speech recognition result, and logo recognition result after standardization are all not empty, if the brand words in the logo recognition result after standardization are inconsistent with those in the text recognition result, take the brand words in the logo recognition result after standardization; if the product words in the text recognition result after standardization are inconsistent with those in the speech recognition result, take the product words in the text recognition result after standardization;

[0092] S322: When the logo recognition result after standardization is empty, if the product words, and / or brand words in the text recognition result after standardization are inconsistent with those in the speech recognition result, take the product words, and / or brand words in the text recognition result after standardization;

[0093] S323: When the text recognition result after standardization is empty, if the brand words in the logo recognition result after standardization are inconsistent with those in the speech recognition result, take the brand words in the logo recognition result after standardization, and do not retain the product words in the speech recognition result after standardization;

[0094] S324: When the speech recognition result after standardization is empty, if the brand words in the logo recognition result after standardization are inconsistent with those in the text recognition result, take the brand words in the logo recognition result after standardization, and do not retain the product words in the text recognition result after standardization;

[0095] S325: When two of the logo recognition result, speech recognition result, and text recognition result after standardization are empty, take the non-empty recognition result after standardization.

[0096] It should be noted that the text recognition result and speech recognition result after standardization being empty means that the recognition result contains neither product words nor brand words. In step S322, the inconsistent situations between the product words and / or brand words in the text recognition result and speech recognition result after standardization include: the text recognition result after standardization contains product words while the product words in the speech recognition result after standardization are empty; the text recognition result after standardization contains brand words while the brand words in the speech recognition result after standardization are empty; the product words in the text recognition result after standardization are empty while the speech recognition result after standardization contains product words; the brand words in the text recognition result after standardization are empty while the speech recognition result after standardization contains brand words; both the text recognition result and the speech recognition result after standardization contain both product words and brand words, and the product words or brand words in the text recognition result and speech recognition result after standardization are inconsistent.

[0097] S4: According to the fusion result of S3, output the content information of the advertising material. Among them, output the product information, brand information, industry information, spokesperson information, advertiser information, etc. of the advertising material.

[0098] An advertising material content recognition method provided in this embodiment uses recognition technology to recognize advertising materials, standardizes multiple recognition results according to the advertising resource metadata standard library, fuses multiple standardized recognition results, and outputs the content information of the advertising material according to the fusion result. Based on the advertising resource metadata standard library, this method outputs a recognition result with unified standards by fusing multiple recognition results, improving the accuracy of advertising material content recognition.

[0099] Next, as an example, specifically introduce how to recognize the content information of advertising materials.

[0100] Use crawlers to not collect the monitoring information of advertising materials, and provide advertising materials according to the following website links;

[0101] Website link of the advertising material:

[0102] https: / / chuangyiku-1255521909.cos.ap-beijing.myqcloud.com / chuangyiku2 / 37ddc82b54371a5c1031c444c40c6922.mp4?q-sign-algorithm=sha1&q-ak=AKIDQdaLvJLdYJrALDMLejfMn14uWs1yxAQi&q-sign-time=1591574088;1749254088&q-key-time=1591574088;1749254088&q-header-list=&q-url-param-list=&q-signature=4c21ac32f008d4b7814cca792d7ce08b811ec4d8

[0103] Use text recognition technology, speech recognition technology, logo recognition technology, and face recognition technology to recognize advertising materials, and output text recognition results, speech recognition results, logo recognition results, and face recognition results;

[0104] The product words in the text recognition results include "Uni-President Iced Black Tea" and "JD Supermarket", the brand words in the text recognition results include "Uni-President" and "JD Supermarket", the product word in the speech recognition result is "Uni-President Iced Black Tea", the brand word in the speech recognition result is "Uni-President", the brand word in the logo recognition result is "Uni-President", and the face recognition result is empty;

[0105] According to the advertising resource metadata standard library, standardize the text recognition results, speech recognition results, logo recognition results, and face recognition results. According to the standardization processing method of the text recognition results, the number of occurrences of "Uni-President Iced Black Tea" in the text recognition results is more than that of "JD Supermarket". Therefore, the product word in the text recognition results is "Uni-President Iced Black Tea". According to the advertising resource metadata standard library, the brand word to which the product word "Uni-President Iced Black Tea" belongs is "Uni-President". According to the standardization processing method of the speech recognition results, the product word in the speech recognition results is "Uni-President Iced Black Tea", and the brand word is "Uni-President". According to the standardization processing method of the logo recognition results, the brand word in the logo recognition results is "Uni-President".

[0106] According to the fusion method, where the text recognition result, speech recognition result, and logo recognition result are all not empty, and the text recognition result is consistent with the speech recognition result, and the text recognition result is consistent with the logo recognition result, the content information of the advertisement material is output, the product information: "Uni-President Iced Black Tea", the brand information: "Uni-President", according to the advertisement resource metadata standard library, the related industry information: "tea beverage" can be output, and the advertiser information: "Uni-President".

[0107] An embodiment of the present invention further provides an advertisement material content recognition system. Refer to Figure 4 , the advertisement material content recognition system includes:

[0108] The recognition unit is used to recognize the advertisement material by using recognition technology and output multiple recognition results;

[0109] The processing unit is used to standardize multiple recognition results according to the advertisement resource metadata standard library;

[0110] The fusion unit is used to fuse multiple standardized recognition results;

[0111] The advertisement content output unit is used to output the content information of the advertisement material according to the fusion result of the fusion unit.

[0112] The above are only the preferred embodiments of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An advertising material content recognition method, characterized in that, it includes: S1: Use recognition technology to recognize the advertising material and output multiple recognition results; S2: Standardize multiple said recognition results according to the advertising resource metadata standard library; S3: Integrate multiple standardized said recognition results; S4: Output the content information of the advertising material according to the integration result of S3; wherein, before the S1 step, it further includes: S0: Collect the monitoring information of the advertising material. When the monitoring information is collected, output the content information of the advertising material according to the monitoring information and the advertising resource metadata standard library; when the monitoring information is not collected, execute the S1-S4 steps; wherein, in the S1 step, the recognition technology includes: character recognition technology, speech recognition technology, logo recognition technology and face recognition technology; wherein, in the S2 step, it specifically includes: S21: Standardize the character recognition result and the speech recognition result according to the advertising resource metadata standard library; S22: Standardize the logo recognition result according to the advertising resource metadata standard library; S23: Standardize the face recognition result according to the advertising resource metadata standard library; wherein, in the S3 step, the method for integrating multiple standardized said recognition results specifically includes: S31: Determine the trust relationship of the standardized character recognition result, the speech recognition result and the logo recognition result; wherein, the trust relationship is that the trust of the standardized logo recognition result is greater than the trust of the standardized character recognition result, and the trust of the standardized character recognition result is greater than the trust of the standardized speech recognition result; S32: When the standardized character recognition result, the speech recognition result and the logo recognition result are inconsistent, integrate multiple standardized said recognition results according to the trust relationship; wherein, in the S32 step, integrating multiple standardized recognition results according to the trust relationship can be carried out according to the following steps: S321: When the standardized character recognition result, speech recognition result and logo recognition result are all not empty, if the brand word in the standardized logo recognition result is inconsistent with the brand word in the character recognition result, take the brand word in the standardized logo recognition result; if the product word in the standardized character recognition result is inconsistent with the product word in the speech recognition result, take the product word in the standardized character recognition result; S322: When the standardized logo recognition result is empty, if the product word and / or brand word in the standardized character recognition result is inconsistent with the product word and / or brand word in the speech recognition result, take the product word and / or brand word in the standardized character recognition result; S323: When the text recognition result after normalization is empty, if the logo recognition result after normalization is inconsistent with the brand word in the speech recognition result, take the brand word in the logo recognition result after normalization and do not retain the product word in the speech recognition result after normalization; S324: When the speech recognition result after normalization is empty, if the logo recognition result after normalization is inconsistent with the brand word in the text recognition result, take the brand word in the logo recognition result after normalization and do not retain the product word in the text recognition result after normalization; S325: When two of the logo recognition result, speech recognition result, and text recognition result after normalization are empty, take the non-empty recognition result after normalization.

2. The advertisement material content recognition method according to claim 1, wherein, in the step S21, the method for normalizing the text recognition result and the speech recognition result specifically includes: S211: Determine the product words and / or brand words in the text recognition result and the speech recognition result whose confidence levels are greater than or equal to the first preset threshold; S212: When the text recognition result and the speech recognition result contain product words, take the product words and the brand words to which the product words belong; S213: When the text recognition result and the speech recognition result only contain brand words, take the brand words; S214: When the text recognition result and the speech recognition result contain blacklist words, filter the blacklist words; S215: When the text recognition result and the speech recognition result contain product alias words and / or brand alias words, take the product words corresponding to the product alias words and / or the brand words corresponding to the brand alias words; S216: When the text recognition result and the speech recognition result contain multiple product words and / or brand words, determine the product words and / or the brand words according to the phrase relationship, the number of product words, and the number of brand words.

3. The advertisement material content recognition method according to claim 2, wherein, in the step S216, the method for determining the product words and / or the brand words specifically includes: When the text recognition result and the speech recognition result contain a phrase relationship, take the product words and / or the brand words included in the long and short phrase relationships; When the text recognition result and the speech recognition result do not contain a phrase relationship, count the number of each product word and / or the number of each brand word. If the number of each product word and / or the number of each brand word is different, take the product word and / or the brand word with a larger number; If the number of each product word and / or the number of each brand word is the same, take the product word and / or the brand word with a higher confidence level.

4. The advertisement material content recognition method according to claim 1, wherein, in the step S22, the method for normalizing the logo recognition result specifically includes: S221: Map the logo recognition result to the advertising resource metadata standard library for matching; S222: When the logo recognition result exactly matches the advertising resource metadata standard library, take the brand word after mapping.

5. The advertising material content recognition method according to claim 1, characterized in that, in the step S23, the method for standardizing the face recognition result specifically includes: S231: Determine the name of the spokesperson in the face recognition result whose confidence level is greater than or equal to the second preset threshold; S232: Map the name of the spokesperson to the advertising resource metadata standard library for matching; S233: When the name of the spokesperson exactly matches the advertising resource metadata standard library, take the name of the spokesperson after mapping and the information of the spokesperson.

6. An advertising material content recognition system, characterized in that, it includes: An identification unit, which is used to identify the advertising material by using identification technology and output multiple identification results; A processing unit, which is used to standardize multiple said identification results according to the advertising resource metadata standard library; A fusion unit, which is used to fuse multiple standardized identification results; An advertising content output unit, which is used to output the content information of the advertising material according to the fusion result of the fusion unit; wherein, before the identification unit uses the identification technology to identify the advertising material, it further includes: Collect the monitoring information of the advertising material. When the monitoring information is collected, output the content information of the advertising material according to the monitoring information and the advertising resource metadata standard library; when the monitoring information is not collected, execute the identification unit, the processing unit, the fusion unit, and the advertising content output unit; wherein, in the identification unit, the identification technology includes: character recognition technology, speech recognition technology, logo recognition technology, and face recognition technology; wherein, in the processing unit, it specifically includes: Standardize the character recognition result and the speech recognition result according to the advertising resource metadata standard library; Standardize the logo recognition result according to the advertising resource metadata standard library; Standardize the face recognition result according to the advertising resource metadata standard library; wherein, in the fusion unit, fusing multiple standardized identification results specifically includes: Determine the trust relationship of the standardized character recognition result, the speech recognition result, and the logo recognition result; wherein, the trust relationship is that the trust of the standardized logo recognition result is greater than the trust of the standardized character recognition result, and the trust of the standardized character recognition result is greater than the trust of the standardized speech recognition result; When the standardized character recognition result, the speech recognition result, and the logo recognition result are inconsistent, fuse multiple standardized identification results according to the trust relationship; wherein, fusing multiple standardized identification results according to the trust relationship is specifically: When the text recognition result, speech recognition result, and logo recognition result after standardization are all not empty, if the brand word in the logo recognition result after standardization is inconsistent with the brand word in the text recognition result, take the brand word in the logo recognition result after standardization; if the product word in the text recognition result after standardization is inconsistent with the product word in the speech recognition result, take the product word in the text recognition result after standardization. When the logo recognition result after standardization is empty, if the product word and / or brand word in the text recognition result after standardization is inconsistent with the product word and / or brand word in the speech recognition result, take the product word and / or brand word in the text recognition result after standardization. When the text recognition result after standardization is empty, if the brand word in the logo recognition result after standardization is inconsistent with the brand word in the speech recognition result, take the brand word in the logo recognition result after standardization and do not retain the product word in the speech recognition result after standardization. S324: When the speech recognition result after standardization is empty, if the brand word in the logo recognition result after standardization is inconsistent with the brand word in the text recognition result, take the brand word in the logo recognition result after standardization and do not retain the product word in the text recognition result after standardization. When two of the logo recognition result, speech recognition result, and text recognition result after standardization are empty, take the non-empty recognition result after standardization.

Images