WO2018120575A1 - 网页主图识别方法和装置 - Google Patents

网页主图识别方法和装置 Download PDF

Info

Publication number
WO2018120575A1
WO2018120575A1 PCT/CN2017/083544 CN2017083544W WO2018120575A1 WO 2018120575 A1 WO2018120575 A1 WO 2018120575A1 CN 2017083544 W CN2017083544 W CN 2017083544W WO 2018120575 A1 WO2018120575 A1 WO 2018120575A1
Authority
WO
WIPO (PCT)
Prior art keywords
picture
candidate
main
image
webpage
Prior art date
Application number
PCT/CN2017/083544
Other languages
English (en)
French (fr)
Inventor
秦首科
韩友
徐培治
邱学忠
马小林
Original Assignee
百度在线网络技术（北京）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 百度在线网络技术（北京）有限公司 filed Critical 百度在线网络技术（北京）有限公司
Priority to EP17888273.4A priority Critical patent/EP3564833B1/en
Priority to JP2019552321A priority patent/JP6838167B2/ja
Priority to US16/474,842 priority patent/US10963690B2/en
Publication of WO2018120575A1 publication Critical patent/WO2018120575A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/51—Indexing; Data structures therefor; Storage structures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/53—Querying
    • G06F16/538—Presentation of query results
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/5866—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/951—Indexing; Web crawling techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0276—Advertisement creation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/40—Document-oriented image-based pattern recognition
    • G06V30/41—Analysis of document content
    • G06V30/413—Classification of content, e.g. text, photographs or tables
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
    • G06V20/597—Recognising the driver's state or behaviour, e.g. attention or drowsiness

Definitions

  • the present disclosure generally relates to the field of Internet technologies, and in particular, to a webpage main map identification method and apparatus.
  • the pictures in the webpage can intuitively convey the content contained in the user; and, compared with the text display, the image display in the webpage can provide more information, which is more convenient for the netizens to intuitively and quickly select the content of interest.
  • the picture quality and information is significantly higher than the main picture of other pictures, which best reflects the theme of the page. Therefore, in practical applications, in order to improve the click rate of commercial promotion, the advertiser's picture material can be fully explored, and the main picture of the theme of the prominent web page is displayed in the commercial promotion.
  • a main picture recognition scheme in the prior art that can capture key pictures in a webpage: a DOM (Document Object Model) structure of a webpage according to a webpage address; and a central node of the webpage according to a DOM structure of the webpage Regularly match the picture at the center node and its sibling node, format filtering and attribute filtering the regular matching picture (select the picture that meets the specified height and width), output the picture that meets the filtering condition; The key picture of the page taken.
  • a DOM Document Object Model
  • the inventors of the present invention have found that the main image captured by the existing webpage main image recognition scheme cannot accurately express the theme of the webpage, thereby resulting in inefficient webpage information transmission.
  • an embodiment of the present application provides a method for identifying a web page main image, including:
  • the original picture of the candidate main picture is cropped according to the information theme of the candidate main picture to obtain a corresponding picture composition
  • the picture composition corresponding to the matched candidate main picture is identified as the main picture of the web page.
  • the embodiment of the present application further provides a webpage main map identification apparatus, including:
  • An attribute screening unit configured to filter out candidate main maps based on page attributes of each picture in the webpage
  • the composition cropping unit is configured to crop the original image of the candidate main image according to the information theme of the candidate main image, to obtain a corresponding image composition
  • a topic matching unit configured to determine a candidate master image that matches an information topic with a theme of the web page
  • the main picture recognition unit is configured to identify the picture composition corresponding to the matched candidate main picture as the main picture of the webpage.
  • the embodiment of the present application further provides a computing device, including one or more processors and a memory, where the memory includes instructions executable by the processor to enable the processor to perform the embodiments of the present application.
  • the webpage main map identification scheme provided by the embodiment of the present application, after filtering the candidate main map according to the page attribute, by cropping the original image of the candidate main image to obtain a picture composition capable of highlighting the information theme; the information theme and the webpage
  • the theme of the matching picture is the main picture of the webpage, which not only can express the theme of the webpage, but also effectively highlight the theme of the webpage and improve the efficiency of webpage information transmission. Further, filtering the candidate main image by the picture type can ensure the high quality requirement of the final determined webpage main image.
  • FIG. 1 illustrates an exemplary system architecture in which embodiments of the present application may be applied
  • FIG. 2 illustrates an exemplary flowchart of a web page master map identification method in accordance with an embodiment of the present application
  • FIG. 3 illustrates an exemplary flowchart of a webpage main map identification method according to another embodiment of the present application
  • FIG. 4 is a block diagram showing an exemplary structure of a webpage main map recognizing apparatus according to an embodiment of the present application.
  • FIG. 5 is a block diagram showing an exemplary structure of a webpage main map recognizing apparatus according to another embodiment of the present application.
  • FIG. 6 shows a block diagram of a computer system suitable for implementing the embodiments of the present application.
  • FIG. 1 illustrates an exemplary system architecture 100 in which embodiments of the present application may be applied.
  • system architecture 100 can include terminal devices 101, 102, network 103, and servers 104, 105, 106, and 107.
  • the network 103 is used to provide a medium for communication links between the terminal devices 101, 102 and the servers 104, 105, 106, 107.
  • Network 103 may include various types of connections, such as wired, wireless communication links, fiber optic cables, and the like.
  • the user 110 can interact with the servers 104, 105, 106, 107 over the network 103 using the terminal devices 101, 102 to access various services, such as browsing web pages, downloading data, and the like.
  • the terminal devices 101, 102 can be various electronic devices including, but not limited to, personal computers, smart phones, smart televisions, tablets, personal digital assistants, e-book readers, and the like.
  • the servers 104, 105, 106, 107 may be servers that provide various services.
  • the server can provide services in response to a user's service request. It can be understood that one server can provide one or more services, and the same service can also be provided by multiple servers.
  • terminal devices, networks, and servers in Figure 1 is merely illustrative. Depending on the implementation needs, there can be any number of terminal devices, networks, and servers.
  • the webpage main map identification scheme provided by the prior art has a problem that the theme of the webpage cannot be accurately expressed. In this way, not only the efficiency of the transmission of the advertiser's webpage information is lowered, but also the user cannot access the webpage accurately and quickly, and the user experience is poor.
  • the embodiment of the present application provides a webpage main map identification scheme, which can filter the original main image according to the page attribute, and then cut the original image of the candidate main image to obtain a more prominent information theme.
  • the composition of the picture; the picture composition matching the information theme and the theme of the webpage as the main picture of the webpage can effectively highlight the theme of the webpage, thereby improving the transmission efficiency of the webpage information and improving the user experience.
  • FIG. 2 an exemplary flowchart of a web page master map identification method in accordance with one embodiment of the present application is shown.
  • the method shown in FIG. 2 can be performed by the terminal device in FIG. 1, or by a server.
  • step 210 the candidate main map is filtered based on the page attributes of each picture in the web page.
  • the inventor of the present invention analyzes the page structure of the webpage and finds that the most important image (ie, the main image) that best expresses the main image of the webpage is generally higher and more special in the webpage; and compared with other images,
  • the main picture has a larger area and a higher definition.
  • the page information of the webpage may be first captured, and all the images included in the webpage may be identified; and the page attribute of the image is obtained for each image included in the webpage.
  • the page attribute of the picture may include at least one of the following: a page position and a screen ratio.
  • the page position refers to the position of the image in the web page; the screen ratio refers to the proportion of the area of the image in the area of the entire web screen.
  • the position of the image in the webpage can be rendered by the browser to calculate the coordinate information of the image in the webpage.
  • the candidate main picture in the webpage may be filtered based on the page attribute; the selected candidate main picture is one or more.
  • a picture in the webpage that satisfies any of the following conditions may be filtered as a candidate main picture:
  • the difference between the page position of the picture and the preset center position is less than the first threshold; the screen ratio of the picture exceeds the second threshold.
  • the first threshold and the second threshold are all empirical values set by those skilled in the art.
  • a picture that satisfies both of the above conditions may be used as a candidate main picture in the screening process of the candidate main picture; and any of the conditions or the above two conditions are not satisfied.
  • the picture is used as a non-candidate master picture.
  • the page attribute may further include: a picture area, a picture definition, and the like.
  • the image sharpness and the picture area may be used for preliminary filtering: removing the picture whose picture clarity is lower than the third threshold and the picture area is smaller than the fourth threshold; and then, for the remaining after filtering
  • the picture compares the page position with the preset center position, and the screen's screen ratio calculation; selects the candidate main picture according to the page position and the screen ratio.
  • the third threshold and the fourth threshold may be set by a person skilled in the art according to experience.
  • step 220 the original picture of the candidate main picture is cropped according to the information theme of the candidate main picture to obtain a corresponding picture composition.
  • each candidate main image obtained by the step 210 may be firstly captured, and the corresponding original image is captured; then, the first model tool based on the deep learning is used to identify the information theme of the candidate main image; A model tool that crops the original image of the candidate main image to obtain a picture composition of the highlighted information topic.
  • the first model tool is a pre-trained machine learning model based on deep learning.
  • the sample picture may be collected in advance, the information subject of the sample picture is manually determined, and the picture composition highlighting the information subject in the sample picture is cropped; then, the machine learning model is used to learn the correspondence between the sample picture and the information topic, and the information topic
  • the first model tool is obtained by the correspondence between the picture composition and the correspondence between the sample picture and the picture composition.
  • the training of the first model tool may also adopt other technical means commonly used in the art, and will not be described in detail herein.
  • step 230 a candidate master map in which the information subject matches the subject of the web page is determined.
  • the information theme of each candidate main graph needs to be compared with the theme of the webpage, and the candidate main map matching the theme of the webpage is determined.
  • the information subject of the candidate main picture may be extracted by using a picture subject extraction technology commonly used by those skilled in the art; or the information subject of the candidate main picture recognized by the first model tool may be directly obtained.
  • bidding words for web pages, and these bidding words are generally concentrated on a certain topic.
  • the subject matter of the web page can be determined based on the bidding words that the user has previously configured for the web page.
  • step 240 the picture composition corresponding to the matched candidate main picture is identified as the main picture of the web page.
  • the embodiments of the present application are applicable to the generation of search result content.
  • the search keyword input by the user may be used as the query subject; after receiving the query subject, the main map matching the information subject with the query subject may be searched; and then the matching main map will be found. Display as search result content.
  • a scheme for cropping the candidate main image according to the information theme of the candidate main image is proposed.
  • a reasonable composition of the candidate main image highlighting theme can be obtained, and then, after matching with the theme of the webpage, the main image of the theme of the webpage can be effectively highlighted, the transmission efficiency of the webpage information can be improved, and the user experience can be improved.
  • the identified main map of the webpage is promoted and displayed, the click rate of the promotion will be greatly improved.
  • step 230 may be performed to determine a candidate main map that matches the information subject with the topic of the webpage; and then, step 220 is performed, for the candidate main map matching the theme of the webpage, according to the candidate main map.
  • the information subject, the original picture of the candidate main picture is cropped, and the corresponding picture composition is obtained; then, in step 240, the picture composition corresponding to the matched candidate main picture is identified as the main picture of the webpage.
  • FIG. 3 illustrates an exemplary flowchart of a web page main map identification method according to another embodiment of the present application.
  • the method shown in FIG. 3 can be performed at the terminal device or server in FIG. 1.
  • step 310 based on the page attributes of each picture in the web page, Select the candidate master map.
  • Step 310 is similar to step 210 and will not be described here.
  • step 320 the original picture of the candidate main picture is cropped according to the information theme of the candidate main picture to obtain a corresponding picture composition.
  • Step 320 is similar to step 220, and details are not described herein again.
  • step 330 the picture type of the candidate main picture is acquired, and the candidate main picture of the specified picture type is filtered.
  • the second model tool based on deep learning may be used to classify the original pictures of the candidate main picture to determine the picture type of the candidate main picture.
  • the candidate main image filtered by step 210 is filtered by preset filtering conditions. For example, filtering the candidate master image for the specified image type.
  • the specified picture type includes at least one of the following: a texture type and a two-dimensional code type. In actual applications, you can also set the image type that needs to be filtered according to actual needs.
  • the second model tool is a pre-trained classification model based on deep learning.
  • the sample picture may be collected in advance, and the type description of the sample picture may be manually determined; then, the collected sample picture and the corresponding type description are used as training samples, and the classification model is trained to obtain a second model tool.
  • the training of the second model tool may also adopt other technical means commonly used in the art, and will not be described in detail herein.
  • step 340 a candidate master map in which the information subject matches the subject of the web page is determined.
  • step 350 the picture composition corresponding to the matched candidate main picture is identified as the main picture of the web page.
  • the search keyword input by the user may be used as a query subject; after receiving the query subject, the main map matching the information subject with the query subject may be searched; and then the matching main map is displayed as the search result content for display.
  • a scheme for cropping the candidate main image according to the information theme of the candidate main image is proposed.
  • a reasonable composition of the candidate main image highlighting theme can be obtained, and then, after matching with the theme of the webpage, the main image of the theme of the webpage can be effectively highlighted, the transmission efficiency of the webpage information can be improved, and the user experience can be improved.
  • the embodiment of FIG. 3 increases the scheme for filtering the candidate main image by the picture type, and can filter the low quality picture, thereby ensuring the high quality requirement of the final determined main picture of the webpage. To enhance the user experience.
  • step 320 and step 330 in embodiment 3 may be performed in an exchange order or simultaneously. Additionally or alternatively, certain steps may be omitted, multiple steps being combined into one step, and/or one step being broken down into multiple steps.
  • FIG. 4 an exemplary block diagram of a web page master map identification device 400 in accordance with one embodiment of the present application is shown.
  • the web page main map identifying apparatus 400 may include an attribute screening unit 401, a composition cropping unit 402, a topic matching unit 403, and a main map determining unit 404.
  • the attribute filtering unit 401 is configured to filter out the candidate main map based on page attributes of each picture in the webpage.
  • the page attribute includes at least one of the following: page position, screen ratio.
  • the attribute screening unit 401 is configured to filter a picture in the webpage that satisfies any of the following conditions as a candidate main picture: a difference between a page position of the picture and a preset center position is less than a first threshold; a screen ratio of the picture Exceeded the second threshold.
  • the composition cropping unit 402 is configured to crop the original image of the candidate main image according to the information theme of the candidate main image to obtain a corresponding picture composition.
  • the composition cropping unit 402 is configured to identify an information theme of the candidate main image by using the first model tool based on deep learning; and crop the original image of the candidate main image by using the first model tool based on deep learning, Get a picture composition of a prominent information topic.
  • the topic matching unit 403 is configured to determine a candidate master map in which the information topic matches the topic of the web page.
  • the theme of the webpage is determined according to the bidding words configured by the user for the webpage.
  • the main map recognition unit 404 is configured to compose the picture corresponding to the matched candidate main picture Recognized as the main map of the web page.
  • the webpage main image recognition apparatus 400 may further include: a main map display unit.
  • the main image display unit is configured to receive the main theme of the information topic and the query theme after receiving the query theme, and display the matching main image.
  • FIG. 5 an exemplary block diagram of a web page master map identification device 500 in accordance with another embodiment of the present application is shown.
  • the webpage main map identifying apparatus 500 may include an attribute screening unit 501, a composition cropping unit 502, a candidate filtering unit 503, a topic matching unit 504, and a topic determining unit 505.
  • the attribute filtering unit 501 is configured to filter out the candidate main map based on page attributes of each picture in the webpage.
  • the page attribute includes at least one of the following: page position, screen ratio.
  • the attribute screening unit 501 is configured to filter a picture in the webpage that satisfies any of the following conditions as a candidate main picture: a difference between a page position of the picture and a preset center position is less than a first threshold; a screen ratio of the picture Exceeded the second threshold.
  • the composition cropping unit 502 is configured to crop the original image of the candidate main image according to the information theme of the candidate main image to obtain a corresponding picture composition.
  • the composition cropping unit 502 is configured to identify an information theme of the candidate main image by using the first model tool based on depth learning; and crop the original image of the candidate main image by using the first model tool based on deep learning, Get a picture composition of a prominent information topic.
  • the candidate filtering unit 503 is configured to acquire a picture type of the candidate main picture; and filter the candidate main picture of the specified picture type, wherein the specified picture type includes at least one of the following: a texture type, a two-dimensional code type.
  • the candidate filtering unit 503 is configured to classify the original picture of the candidate main picture by using the second model tool based on deep learning, and determine the picture type of the candidate main picture.
  • the topic matching unit 504 is configured to determine that the information topic matches the theme of the webpage Candidate master map.
  • the theme of the webpage is determined according to the bidding words configured by the user for the webpage.
  • the main map recognition unit 505 is configured to recognize the picture composition corresponding to the matched candidate main picture as the main picture of the web page.
  • the webpage main image recognition apparatus 500 further includes: a main map display unit.
  • the main image display unit is configured to receive the main theme of the information topic and the query theme after receiving the query theme, and display the matching main image.
  • the embodiment of the present application further provides a computing device, including one or more processors and a memory, where the memory includes instructions executable by the processor to cause the processor to perform the webpage main image recognition provided by the embodiment of the present application. method.
  • FIG. 6 a block diagram of a computing device 600 suitable for implementing embodiments of the present application is shown.
  • computer system 600 includes a central processing unit (CPU) 601 that can be loaded into a program in random access memory (RAM) 603 according to a program stored in read only memory (ROM) 602 or from storage portion 608. And perform various appropriate actions and processes.
  • RAM random access memory
  • ROM read only memory
  • RAM random access memory
  • various programs and data required for the operation of the system 600 are also stored.
  • the CPU 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604.
  • An input/output (I/O) interface 605 is also coupled to bus 604.
  • the following components are connected to the I/O interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), and the like, and a storage portion 608 including a hard disk or the like. And a communication portion 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the Internet.
  • Driver 610 is also coupled to I/O interface 605 as needed.
  • a removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory or the like, is mounted on the drive 610 as needed so that a computer program read therefrom is installed into the storage portion 608 as needed.
  • an embodiment of the present disclosure includes a computer program product comprising a computer program tangibly embodied on a machine readable medium, the computer program comprising program code for performing the methods of FIGS. 2-3.
  • the computer program can be downloaded and installed from the network via communication portion 609, and/or installed from removable media 611.
  • each block of the flowchart or block diagram can represent a module or unit, a program segment, or a portion of code, the module or unit, program segment, or a portion of code comprising one or more An executable instruction that implements the specified logic functions.
  • the functions noted in the blocks may also occur in a different order than that illustrated in the drawings. For example, two successively represented blocks may in fact be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts can be implemented in a dedicated hardware-based system that performs the specified function or operation. Or it can be implemented by a combination of dedicated hardware and computer instructions.
  • the units or modules described in the embodiments of the present application may be implemented by software or by hardware.
  • the described unit or module can also be provided in the processor.
  • the names of these units or modules do not in any way constitute a limitation on the unit or module itself.
  • the present application further provides a computer readable storage medium storing a computer program, which may be a computer readable storage medium included in the computing device described in the above embodiments; It may be a computer readable storage medium that exists alone and is not assembled into the device.
  • the computer readable storage medium stores one or more programs that are used by one or more processors to perform the formula input methods described in this application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Business, Economics & Management (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Accounting & Taxation (AREA)
  • Library & Information Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Signal Processing (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Information Transfer Between Computers (AREA)
  • Image Analysis (AREA)
  • User Interface Of Digital Computer (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种网页主图识别方法和装置，该方法包括：基于网页中各图片的页面属性，筛选出候选主图（210）；根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图（220）；以及确定出信息主题与所述网页的主题匹配的候选主图（230）；以及将匹配的候选主图所对应的图片构图识别为所述网页的主图（240）。该方法可识别出有效突出网页的主题的主图，提升网页信息的传递效率，提高用户体验

Description

网页主图识别方法和装置 技术领域
本公开一般涉及互联网技术领域，具体涉及一种网页主图识别方法和装置。
背景技术
随着互联网技术的发展，图文并茂形态的网页越来越多。商业推广中，网页中的图片可以直观地向用户传达所包含的内容；而且，相比文字展示，网页中的图片展示可提供更多的信息，更利于网民直观快速选定感兴趣的内容，尤其是图片质量和信息明显高于其他图片的网页主图，最能体现网页的主题。因此，实际应用中，为了提高商业推广的点击率，可以充分挖掘广告主的图片物料，在商业推广中展示突出网页的主题的主图。
目前，现有技术中存在一种主图识别方案可抓取网页中的关键图片：根据网页地址获取网页的DOM(Document Object Model，文档对象模型)结构；根据网页的DOM结构定位网页的中心节点；正则匹配中心节点及其兄弟节点处的图片，对正则匹配出的图片进行格式过滤和属性过滤(选出符合指定高度和宽度的图片)，输出符合过滤条件的图片；将输出的图片作为抓取到的网页的关键图片。
然而，本发明的发明人发现，通过现有网页主图识别方案抓取的主图无法准确表达网页的主题，从而导致网页信息传递效率低。
发明内容
鉴于现有技术中的上述缺陷或不足，期望提供一种能够挖掘出有效突出网页的主题的主图的方案，从而提升网页信息的传递效率，提高用户体验。
第一方面，本申请实施例提供了一种网页主图识别方法，包括：
基于网页中各图片的页面属性，筛选出候选主图；
根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图；
确定出信息主题与所述网页的主题匹配的候选主图；以及
将匹配的候选主图所对应的图片构图识别为所述网页的主图。
第二方面，本申请实施例还提供了一种网页主图识别装置，包括：
属性筛选单元，配置用于基于网页中各图片的页面属性，筛选出候选主图；
构图裁剪单元，配置用于根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图；
主题匹配单元，配置用于确定出信息主题与所述网页的主题匹配的候选主图；以及
主图识别单元，配置用于将匹配的候选主图所对应的图片构图识别为所述网页的主图。
第三方面，本申请实施例还提供了一种计算设备，包括一个或多个处理器以及存储器，所述存储器包含可由所述处理器执行的指令以使得所述处理器执行本申请实施例提供的网页主图识别方法。
本申请实施例提供的网页主图识别方案，在根据页面属性筛选出候选主图后，通过对候选主图的原始图片的裁剪，以得到能够更加突出信息主题的图片构图；将信息主题与网页的主题匹配的图片构图作为网页的主图，不仅能够表达网页的主题，还可有效突出网页的主题，提升网页信息的传递效率。进一步地，通过图片类型对候选主图进行过滤，可保障最终确定的网页主图的高质量需求。
附图说明
通过阅读参照以下附图所作的对非限制性实施例所作的详细描述，本申请的其它特征、目的和优点将会变得更明显：
图1示出了其中可以应用本申请实施例的示例性系统架构；
图2示出了根据本申请实施例的网页主图识别方法的示例性流程图；
图3示出了根据本申请另一实施例的网页主图识别方法的示例性流程图；
图4示出了根据本申请一个实施例的网页主图识别装置的示例性结构框图；
图5出了根据本申请另一个实施例的网页主图识别装置的示例性结构框图；以及
图6示出了适于用来实现本申请实施例的计算机系统的结构示意图。
具体实施方式
下面结合附图和实施例对本申请作进一步的详细说明。可以理解的是，此处所描述的具体实施例仅仅用于解释相关发明，而非对该发明的限定。另外还需要说明的是，为了便于描述，附图中仅示出了与发明相关的部分。
需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本申请。
请参考图1，其示出了可以应用本申请实施例的示例性系统架构100。
如图1所示，系统架构100可以包括终端设备101、102、网络103和服务器104、105、106和107。网络103用以在终端设备101、102和服务器104、105、106、107之间提供通信链路的介质。网络103可以包括各种连接类型，例如有线、无线通信链路或者光纤电缆等等。
用户110可以使用终端设备101、102通过网络103与服务器104、105、106、107交互，以访问各种服务，例如浏览网页、下载数据等。
终端设备101、102可以是各种电子设备，包括但不限于个人电脑、智能手机、智能电视、平板电脑、个人数字助理、电子书阅读器等等。
服务器104、105、106、107可以是提供各种服务的服务器。服务器可以响应于用户的服务请求而提供服务。可以理解，一个服务器可以提供一种或多种服务，同一种服务也可以由多个服务器来提供。
应该理解，图1中的终端设备、网络和服务器的数目仅仅是示意性的。根据实现需要，可以具有任意数目的终端设备、网络和服务器。
如背景技术中所提到的，现有技术提供的网页主图识别方案存在无法准确表达网页的主题的问题。这样，不仅会导致广告主网页信息的传递效率降低，而且将会导致用户无法准确快速地访问网页，用户体验差。
鉴于现有技术的上述缺陷，本申请实施例提供了一种网页主图识别方案，根据页面属性筛选出候选主图后，通过对候选主图的原始图片的裁剪，以得到能够更加突出信息主题的图片构图；将信息主题与网页的主题匹配的图片构图作为网页的主图，可有效突出网页的主题，从而提升网页信息的传递效率，提高用户体验。下面将结合流程图来描述本申请实施例的方法。
参考图2，其示出了根据本申请一个实施例的网页主图识别方法的示例性流程图。图2所示的方法可以在图1中的终端设备执行，或服务器执行。
如图2所示，在步骤210中，基于网页中各图片的页面属性，筛选出候选主图。
本发明的发明人对网页的页面结构进行分析后发现，最能表达网页主图的最主要的图片(即主图)在网页中的位置通常比较靠前且特殊；且相较于其他图片，主图的面积较大，清晰度较高。
因此，本申请实施例中，可以先抓取网页的页面信息，识别出网页中所包含的所有图片；针对网页中包含的每个图片，获取该图片的页面属性。其中，图片的页面属性可以包括如下至少一项：页面位置、屏幕占比。
页面位置指的是图片在网页中所处的位置；屏幕占比指的是图片的面积在整个网页屏幕的面积中所占比例。图片在网页中所处的位置，可通过浏览器渲染，真实计算出图片在网页中的坐标信息。
获取到网页中各图片的页面属性之后，可以基于页面属性，筛选出网页中的候选主图；筛选出的候选主图为一个或多个。
具体地，可以将网页中满足以下任一条件的图片筛选为候选主图： 图片的页面位置与预设中心位置之间的差值小于第一阈值；图片的屏幕占比超过第二阈值。其中，第一阈值、第二阈值均是本领域技术人员设置的经验值。
或者，为了提高最终识别的主图的准确率，可在候选主图的筛选过程中将同时满足上述两个条件的图片作为候选主图；而满足其中任一条件或不满足上述两个条件的图片作为非候选主图。
可选地，页面属性可以进一步包括：图片面积、图片清晰度等。
这样，在筛选候选主图的过程中，可以先利用图片清晰度和图片面积做初步过滤：去除图片清晰度低于第三阈值和图片面积小于第四阈值的图片；继而，针对过滤后剩余的图片进行页面位置与预设中心位置的比较，以及图片的屏幕占比的计算；根据页面位置和屏幕占比筛选候选主图。其中，第三阈值、第四阈值可由本领域技术人员根据经验进行设置。
接着，在步骤220中，根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图。
本申请实施例中，可先针对步骤210筛选得到的每个候选主图，抓取对应的原始图片；继而，利用基于深度学习的第一模型工具，识别候选主图的信息主题；并利用第一模型工具，对候选主图的原始图片进行裁剪，以得到突出信息主题的图片构图。
其中，第一模型工具是基于深度学习的预先训练好的机器学习模型。例如，可以预先收集样本图片，人工确定样本图片的信息主题，并裁剪出样本图片中突出该信息主题的图片构图；继而，利用机器学习模型学习样本图片与信息主题之间的对应关系、信息主题与图片构图之间的对应关系、样本图片与图片构图之间的对应关系，得到第一模型工具。实际应用中，第一模型工具的训练，也可采用本领域常用的其他技术手段，在此不再详述。
接着，在步骤230中，确定出信息主题与网页的主题匹配的候选主图。
为了筛选出突出网页的主题的主图，需要将各候选主图的信息主题与网页的主题进行比对，确定出与网页的主题匹配的候选主图。
候选主图的信息主题可采用本领域技术人员常用的图片主题提取技术手段进行提取；或者也可以直接获取第一模型工具识别出的候选主图的信息主题。
考虑到实际应用中，广告主通常会对网页配置竞价词，而这些竞价词一般会集中在某一个主题上。因此，网页的主题可根据用户预先为该网页配置的竞价词而确定。
最后，在步骤240中，将匹配的候选主图所对应的图片构图识别为网页的主图。
进一步地，本申请实施例可应用于搜索结果内容的生成。例如，在用户输入检索关键词后，可将用户输入的检索关键词作为查询主题；接收到查询主题后，可查找信息主题与该查询主题匹配的主图；继而，将查找到匹配的主图作为搜索结果内容进行展示。
从上面描述可以看出，在本申请的一些实施例中，针对现有网页主题不突出的问题，提出了根据候选主图的信息主题，对候选主图进行裁剪的方案。这样做可以得到候选主图突出主题的合理构图，继而在与网页的主题进行匹配后，可以得到有效突出网页的主题的主图，提升网页信息的传递效率，提升用户体验。而且，在实际应用中，若识别出的网页主图进行推广展示，将大大提高推广的点击率。
应当注意，尽管在附图中以特定顺序描述了本发明方法的操作，但是，这并非要求或者暗示必须按照该特定顺序来执行这些操作，或是必须执行全部所示的操作才能实现期望的结果。相反，流程图中描绘的步骤可以改变部分执行顺序。例如，在执行步骤210后，可以先执行步骤230，确定出信息主题与网页的主题匹配的候选主图；之后，执行步骤220，针对与网页的主题匹配的候选主图，根据该候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图；继而通过步骤240将匹配的候选主图所对应的图片构图识别为网页的主图。
图3示出了根据本申请另一实施例的网页主图识别方法的示例性流程图。图3所示的方法可以在图1中的终端设备或服务器执行。
如图3所示，在步骤310中，基于网页中各图片的页面属性，筛 选出候选主图。
步骤310与步骤210类似，此处不再赘述。
在步骤320中，根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图。
步骤320与步骤220类似，此处不再赘述。
在步骤330中，获取候选主图的图片类型，过滤指定的图片类型的候选主图。
图3所示实施例中，为了过滤一些低质量的图片，可以利用基于深度学习的第二模型工具，对候选主图的原始图片进行分类，确定出候选主图的图片类型。继而，通过预设的过滤条件对步骤210筛选出的候选主图进行过滤。比如，过滤指定的图片类型的候选主图。其中，指定的图片类型包括如下至少一项：纹理类型、二维码类型。实际应用中，还可以根据实际需求，设置需要过滤指定的图片类型。
其中，第二模型工具是基于深度学习的预先训练好的分类模型。例如，可以预先收集样本图片，人工确定样本图片的类型描述；继而，将收集的样本图片和对应的类型描述作为训练样本，对分类模型进行训练，得到第二模型工具。实际应用中，第二模型工具的训练，也可采用本领域常用的其他技术手段，在此不再详述。
接着，在步骤340中，确定出信息主题与网页的主题匹配的候选主图。
最后，在步骤350中，将匹配的候选主图所对应的图片构图识别为网页的主图。
进一步地，可将用户输入的检索关键词作为查询主题；接收到查询主题后，可查找信息主题与该查询主题匹配的主图；继而，将查找到匹配的主图作为搜索结果内容进行展示。
从上面描述可以看出，在本申请的一些实施例中，针对现有网页主题不突出的问题，提出了根据候选主图的信息主题，对候选主图进行裁剪的方案。这样做可以得到候选主图突出主题的合理构图，继而在与网页的主题进行匹配后，可以得到有效突出网页的主题的主图，提升网页信息的传递效率，提升用户体验。
而且，图3的实施例相比于图2的实施例，增加了通过图片类型对候选主图进行过滤的方案，可过滤低质量的图片，以此保障最终确定的网页主图的高质量需求，提升用户体验。
应当注意，尽管在附图中以特定顺序描述了本发明方法的操作，但是，这并非要求或者暗示必须按照该特定顺序来执行这些操作，或是必须执行全部所示的操作才能实现期望的结果。相反，流程图中描绘的步骤可以改变执行顺序。例如，实施例3中的步骤320和步骤330可以交换顺序或同时进行。附加地或备选地，可以省略某些步骤，将多个步骤合并为一个步骤执行，和/或将一个步骤分解为多个步骤执行。
进一步参考图4，其示出了根据本申请一个实施例的网页主图识别装置400的示例性结构框图。
如图4所示，网页主图识别装置400可以包括：属性筛选单元401、构图裁剪单元402、主题匹配单元403和主图确定单元404。
其中，属性筛选单元401配置用于基于网页中各图片的页面属性，筛选出候选主图。
页面属性包括如下至少一项：页面位置、屏幕占比。
具体地，属性筛选单元401配置用于将网页中满足以下任一条件的图片筛选为候选主图：图片的页面位置与预设中心位置之间的差值小于第一阈值；图片的屏幕占比超过第二阈值。
构图裁剪单元402配置用于根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图。
具体地，构图裁剪单元402配置用于利用基于深度学习的第一模型工具，识别候选主图的信息主题；以及利用基于深度学习的第一模型工具，对候选主图的原始图片进行裁剪，以得到突出信息主题的图片构图。
主题匹配单元403配置用于确定出信息主题与网页的主题匹配的候选主图。
其中，网页的主题根据用户为该网页配置的竞价词而确定。
主图识别单元404配置用于将匹配的候选主图所对应的图片构图 识别为网页的主图。
可选地，网页主图识别装置400中还可以包括：主图展示单元。
主图展示单元配置用于接收到查询主题后，查找信息主题与查询主题匹配的主图，将匹配的主图进行展示。
应当理解，网页主图识别装置400中记载的诸单元与图2描述的方法中的各个步骤相对应。由此，上文针对方法描述的操作和特征同样适用于网页主图识别装置400及其中包含的单元，在此不再赘述。
进一步参考图5，其示出了根据本申请另一个实施例的网页主图识别装置500的示例性结构框图。
如图5所示，网页主图识别装置500可以包括：属性筛选单元501、构图裁剪单元502、候选过滤单元503、主题匹配单元504和主题确定单元505。
其中，属性筛选单元501配置用于基于网页中各图片的页面属性，筛选出候选主图。页面属性包括如下至少一项：页面位置、屏幕占比。
具体地，属性筛选单元501配置用于将网页中满足以下任一条件的图片筛选为候选主图：图片的页面位置与预设中心位置之间的差值小于第一阈值；图片的屏幕占比超过第二阈值。
构图裁剪单元502配置用于根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图。
具体地，构图裁剪单元502配置用于利用基于深度学习的第一模型工具，识别候选主图的信息主题；以及利用基于深度学习的第一模型工具，对候选主图的原始图片进行裁剪，以得到突出信息主题的图片构图。
候选过滤单元503配置用于获取候选主图的图片类型；以及过滤指定的图片类型的候选主图，其中，指定的图片类型包括如下至少一项：纹理类型、二维码类型。
具体地，候选过滤单元503配置用于利用基于深度学习的第二模型工具，对候选主图的原始图片进行分类，确定出候选主图的图片类型。
主题匹配单元504配置用于确定出信息主题与网页的主题匹配的 候选主图。其中，网页的主题根据用户为该网页配置的竞价词而确定。
主图识别单元505配置用于将匹配的候选主图所对应的图片构图识别为网页的主图。
可选地，网页主图识别装置500中还可以包括：主图展示单元。
主图展示单元配置用于接收到查询主题后，查找信息主题与查询主题匹配的主图，将匹配的主图进行展示。
应当理解，网页主图识别装置500中记载的诸单元与参考图3描述的方法中的各个步骤相对应。由此，上文针对方法描述的操作和特征同样适用于网页主图识别装置500及其中包含的单元，在此不再赘述。
进一步地，本申请实施例还提供了一种计算设备，包括一个或多个处理器以及存储器；其中，存储器包含可由处理器执行的指令以使得处理器执行本申请实施例提供的网页主图识别方法。
下面参考图6，其示出了适于用来实现本申请实施例的计算设备600的结构示意图。
如图6所示，计算机系统600包括中央处理单元(CPU)601，其可以根据存储在只读存储器(ROM)602中的程序或者从存储部分608加载到随机访问存储器(RAM)603中的程序而执行各种适当的动作和处理。在RAM 603中，还存储有系统600操作所需的各种程序和数据。CPU 601、ROM 602以及RAM 603通过总线604彼此相连。输入/输出(I/O)接口605也连接至总线604。
以下部件连接至I/O接口605：包括键盘、鼠标等的输入部分606；包括诸如阴极射线管(CRT)、液晶显示器(LCD)等以及扬声器等的输出部分607；包括硬盘等的存储部分608；以及包括诸如LAN卡、调制解调器等的网络接口卡的通信部分609。通信部分609经由诸如因特网的网络执行通信处理。驱动器610也根据需要连接至I/O接口605。可拆卸介质611，诸如磁盘、光盘、磁光盘、半导体存储器等等，根据需要安装在驱动器610上，以便于从其上读出的计算机程序根据需要被安装入存储部分608。
特别地，根据本公开的实施例，上文参考图2-图3描述的过程可 以被实现为计算机软件程序。例如，本公开的实施例包括一种计算机程序产品，其包括有形地包含在机器可读介质上的计算机程序，所述计算机程序包含用于执行图2-图3的方法的程序代码。在这样的实施例中，该计算机程序可以通过通信部分609从网络上被下载和安装，和/或从可拆卸介质611被安装。
附图中的流程图和框图，图示了按照本发明各种实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块或单元、程序段、或代码的一部分，所述模块或单元、程序段、或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注意，在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个接连地表示的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或操作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
描述于本申请实施例中所涉及到的单元或模块可以通过软件的方式实现，也可以通过硬件的方式来实现。所描述的单元或模块也可以设置在处理器中。这些单元或模块的名称在某种情况下并不构成对该单元或模块本身的限定。
作为另一方面，本申请还提供了一种存储有计算机程序的计算机可读存储介质，该计算机可读存储介质可以是上述实施例中所述计算设备中所包含的计算机可读存储介质；也可以是单独存在，未装配入设备中的计算机可读存储介质。计算机可读存储介质存储有一个或者一个以上程序，所述程序被一个或者一个以上的处理器用来执行描述于本申请的公式输入方法。
以上描述仅为本申请的较佳实施例以及对所运用技术原理的说明。本领域技术人员应当理解，本申请中所涉及的发明范围，并不限于上述技术特征的特定组合而成的技术方案，同时也应涵盖在不脱离所述发明构思的情况下，由上述技术特征或其等同特征进行任意组合 而形成的其它技术方案。例如上述特征与本申请中公开的(但不限于)具有类似功能的技术特征进行互相替换而形成的技术方案。

Claims (17)

  1. 一种网页主图识别方法，其特征在于，包括：
    基于网页中各图片的页面属性，筛选出候选主图；
    根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图；
    确定出信息主题与所述网页的主题匹配的候选主图；以及
    将匹配的候选主图所对应的图片构图识别为所述网页的主图。
  2. 根据权利要求1所述的识别方法，其特征在于，所述页面属性包括如下至少一项：页面位置、屏幕占比。
  3. 根据权利要求2所述的识别方法，其特征在于，所述根据网页中各图片的页面属性，筛选出候选主图，包括：将网页中满足以下任一条件的图片筛选为候选主图：
    图片的页面位置与预设中心位置之间的差值小于第一阈值；
    图片的屏幕占比超过第二阈值。
  4. 根据权利要求1所述的识别方法，其特征在于，所述根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图，包括：
    利用基于深度学习的第一模型工具，识别候选主图的信息主题；以及
    利用所述第一模型工具，对候选主图的原始图片进行裁剪，以得到突出所述信息主题的图片构图。
  5. 根据权利要求1所述的识别方法，其特征在于，所述网页的主题根据用户为该网页配置的竞价词而确定。
  6. 根据权利要求1所述的识别方法，其特征在于，所述筛选出候选主图之后，还包括：
    获取所述候选主图的图片类型；以及
    过滤指定的图片类型的候选主图，其中，所述指定的图片类型包括如下至少一项：纹理类型、二维码类型。
  7. 根据权利要求6所述的识别方法，其特征在于，所述获取所述 候选主图的图片类型，包括：
    利用基于深度学习的第二模型工具，对所述候选主图的原始图片进行分类，确定出所述候选主图的图片类型。
  8. 根据权利要求1-7任一所述的识别方法，其特征在于，还包括：
    接收到查询主题后，查找信息主题与所述查询主题匹配的主图；以及
    将匹配的主图进行展示。
  9. 一种网页主图识别装置，其特征在于，包括：
    属性筛选单元，配置用于基于网页中各图片的页面属性，筛选出候选主图；
    构图裁剪单元，配置用于根据候选主图的信息主题，对候选主图的原始图片进行裁剪，得到对应的图片构图；
    主题匹配单元，配置用于确定出信息主题与所述网页的主题匹配的候选主图；以及
    主图识别单元，配置用于将匹配的候选主图所对应的图片构图识别为所述网页的主图。
  10. 根据权利要求9所述的识别装置，其特征在于，所述页面属性包括如下至少一项：页面位置、屏幕占比。
  11. 根据权利要求10所述的识别装置，其特征在于，
    所述属性筛选单元配置用于将网页中满足以下任一条件的图片筛选为候选主图：
    图片的页面位置与预设中心位置之间的差值小于第一阈值；
    图片的屏幕占比超过第二阈值。
  12. 根据权利要求9所述的识别装置，其特征在于，
    所述构图裁剪单元配置用于利用基于深度学习的第一模型工具，识别候选主图的信息主题；以及利用基于深度学习的第一模型工具，对候选主图的原始图片进行裁剪，以得到突出所述信息主题的图片构图。
  13. 根据权利要求9所述的识别装置，其特征在于，所述网页的主题根据用户为该网页配置的竞价词而确定。
  14. 根据权利要求9所述的识别装置，其特征在于，还包括：
    候选过滤单元，配置用于获取所述候选主图的图片类型；以及过滤指定的图片类型的候选主图，其中，所述指定的图片类型包括如下至少一项：纹理类型、二维码类型。
  15. 根据权利要求14所述的识别装置，其特征在于，所述候选过滤单元配置用于利用基于深度学习的第二模型工具，对所述候选主图的原始图片进行分类，确定出所述候选主图的图片类型。
  16. 根据权利要求9-15任一所述的识别装置，其特征在于，还包括：
    主图展示单元，配置用于接收到查询主题后，查找信息主题与所述查询主题匹配的主图，将匹配的主图进行展示。
  17. 一种计算设备，包括一个或多个处理器以及存储器，其特征在于：
    所述存储器包含可由所述处理器执行的指令以使得所述处理器执行权利要求1-8任一所述的方法。
PCT/CN2017/083544 2016-12-30 2017-05-09 网页主图识别方法和装置 WO2018120575A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP17888273.4A EP3564833B1 (en) 2016-12-30 2017-05-09 Method and device for identifying main picture in web page
JP2019552321A JP6838167B2 (ja) 2016-12-30 2017-05-09 ウェブページメイン画像認識方法及び装置
US16/474,842 US10963690B2 (en) 2016-12-30 2017-05-09 Method for identifying main picture in web page

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611259514.4 2016-12-30
CN201611259514.4A CN108268488B (zh) 2016-12-30 2016-12-30 网页主图识别方法和装置

Publications (1)

Publication Number Publication Date
WO2018120575A1 true WO2018120575A1 (zh) 2018-07-05

Family

ID=62706746

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/083544 WO2018120575A1 (zh) 2016-12-30 2017-05-09 网页主图识别方法和装置

Country Status (5)

Country Link
US (1) US10963690B2 (zh)
EP (1) EP3564833B1 (zh)
JP (1) JP6838167B2 (zh)
CN (1) CN108268488B (zh)
WO (1) WO2018120575A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113742553A (zh) * 2021-09-03 2021-12-03 上海哔哩哔哩科技有限公司 数据处理方法及装置

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110209854B (zh) * 2019-05-06 2021-08-31 无线生活(北京)信息技术有限公司 图片确定方法及装置
CN112766997A (zh) * 2019-11-01 2021-05-07 百度在线网络技术（北京）有限公司 图片投放方法、装置、设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101990064A (zh) * 2009-07-29 2011-03-23 索尼公司 控制设备、操作设置方法和程序
CN103824311A (zh) * 2013-11-29 2014-05-28 北京奇虎科技有限公司 聚合图像的生成方法及设备
CN104063489A (zh) * 2014-07-04 2014-09-24 百度在线网络技术（北京）有限公司 一种确定网页图片相关度及显示检索结果的方法及装置
CN104123363A (zh) * 2014-07-21 2014-10-29 北京奇虎科技有限公司 网页主图提取方法及装置
CN105959330A (zh) * 2016-07-20 2016-09-21 广东世纪网通信设备股份有限公司 虚假链接拦截方法、装置以及系统

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB0116113D0 (en) 2001-06-30 2001-08-22 Hewlett Packard Co Tilt correction of electronic images
GB2378340A (en) 2001-07-31 2003-02-05 Hewlett Packard Co Generation of an image bounded by a frame or of overlapping images
JP2005062991A (ja) * 2003-08-08 2005-03-10 Komatsu Ltd Ｗｅｂページ閲覧装置
WO2008141430A1 (en) * 2007-05-17 2008-11-27 Fat Free Mobile Inc. Method and system for continuation of browsing sessions between devices
US20100073398A1 (en) 2008-09-22 2010-03-25 Microsoft Corporation Visual summarization of web pages
JP5165021B2 (ja) 2010-05-11 2013-03-21 ヤフー株式会社 カテゴリ処理装置及び方法
JP5388988B2 (ja) 2010-10-26 2014-01-15 ヤフー株式会社 広告選択装置、方法及びプログラム
US9589184B1 (en) * 2012-08-16 2017-03-07 Groupon, Inc. Method, apparatus, and computer program product for classification of documents
US20140122233A1 (en) 2012-10-31 2014-05-01 Microsoft Corporation User-advertiser bargaining in search results
US9536001B2 (en) 2012-11-13 2017-01-03 Microsoft Technology Licensing, Llc Intent-based presentation of search results
US20140244405A1 (en) 2013-02-28 2014-08-28 OneSpot, Inc. Automatic Generation of Digital Advertisements
CN104142945B (zh) * 2013-05-08 2018-05-04 阿里巴巴集团控股有限公司 一种基于查询词的检索方法及装置
CN105630780A (zh) * 2014-10-27 2016-06-01 小米科技有限责任公司 网页信息处理方法及装置
CN104504649A (zh) * 2014-12-30 2015-04-08 百度在线网络技术（北京）有限公司 图片的裁剪方法和装置
CN105095911B (zh) * 2015-07-31 2019-02-12 小米科技有限责任公司 敏感图片识别方法、装置以及服务器
CN105718950B (zh) * 2016-01-22 2018-11-27 天津中科智能识别产业技术研究院有限公司 一种基于结构约束的半监督多视角聚类方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101990064A (zh) * 2009-07-29 2011-03-23 索尼公司 控制设备、操作设置方法和程序
CN103824311A (zh) * 2013-11-29 2014-05-28 北京奇虎科技有限公司 聚合图像的生成方法及设备
CN104063489A (zh) * 2014-07-04 2014-09-24 百度在线网络技术（北京）有限公司 一种确定网页图片相关度及显示检索结果的方法及装置
CN104123363A (zh) * 2014-07-21 2014-10-29 北京奇虎科技有限公司 网页主图提取方法及装置
CN105959330A (zh) * 2016-07-20 2016-09-21 广东世纪网通信设备股份有限公司 虚假链接拦截方法、装置以及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3564833A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113742553A (zh) * 2021-09-03 2021-12-03 上海哔哩哔哩科技有限公司 数据处理方法及装置
CN113742553B (zh) * 2021-09-03 2024-03-19 上海哔哩哔哩科技有限公司 数据处理方法及装置

Also Published As

Publication number Publication date
EP3564833A4 (en) 2020-08-19
US10963690B2 (en) 2021-03-30
EP3564833A1 (en) 2019-11-06
US20190332859A1 (en) 2019-10-31
CN108268488B (zh) 2022-04-19
EP3564833B1 (en) 2023-02-22
JP6838167B2 (ja) 2021-03-03
CN108268488A (zh) 2018-07-10
JP2020502710A (ja) 2020-01-23

Similar Documents

Publication Publication Date Title
US20210256320A1 (en) Machine learning artificialintelligence system for identifying vehicles
Raja et al. Color object detection based image retrieval using ROI segmentation with multi-feature method
CN110334346B (zh) 一种pdf文件的信息抽取方法和装置
WO2017080220A1 (zh) 知识数据的处理方法和装置
US9817904B2 (en) Method and system for generating augmented product specifications
CN112559800B (zh) 用于处理视频的方法、装置、电子设备、介质和产品
CN109918513B (zh) 图像处理方法、装置、服务器及存储介质
CN113382279B (zh) 直播推荐方法、装置、设备、存储介质以及计算机程序产品
JP2014524058A (ja) ウェブ情報マイニングを用いたビデオ内製品アノテーション
WO2016018683A1 (en) Image based search to identify objects in documents
JP6419969B2 (ja) 画像の提示情報を提供するための方法及び機器
CN113255501B (zh) 生成表格识别模型的方法、设备、介质及程序产品
WO2018120575A1 (zh) 网页主图识别方法和装置
WO2021114634A1 (zh) 文本标注方法、设备及存储介质
US10606884B1 (en) Techniques for generating representative images
CN110929647B (zh) 一种文本检测方法、装置、设备及存储介质
CN112528610A (zh) 一种数据标注方法、装置、电子设备及存储介质
CN114445833B (zh) 文本识别方法、装置、电子设备和存储介质
KR102444172B1 (ko) 영상 빅 데이터의 지능적 마이닝 방법과 처리 시스템
CN113742485A (zh) 一种处理文本的方法和装置
CN111597368A (zh) 一种数据处理方法及装置
CN115205555B (zh) 确定相似图像的方法、训练方法、信息确定方法及设备
CN116578763B (zh) 基于生成式ai认知模型的多源信息展览系统
CN111881778B (zh) 文本检测的方法、装置、设备和计算机可读介质
CN114979742B (zh) 视频处理方法、装置、设备及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17888273

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2019552321

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2017888273

Country of ref document: EP

Effective date: 20190730