WO2021088422A1 - 应用消息的通知方法及装置 - Google Patents

应用消息的通知方法及装置 Download PDF

Info

Publication number
WO2021088422A1
WO2021088422A1 PCT/CN2020/103972 CN2020103972W WO2021088422A1 WO 2021088422 A1 WO2021088422 A1 WO 2021088422A1 CN 2020103972 W CN2020103972 W CN 2020103972W WO 2021088422 A1 WO2021088422 A1 WO 2021088422A1
Authority
WO
WIPO (PCT)
Prior art keywords
notification
prediction model
area
style
screenshot
Prior art date
Application number
PCT/CN2020/103972
Other languages
English (en)
French (fr)
Inventor
赖伟彬
马泽祥
李招昕
周茜
Original Assignee
支付宝(杭州)信息技术有限公司
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 支付宝(杭州)信息技术有限公司 filed Critical 支付宝(杭州)信息技术有限公司
Publication of WO2021088422A1 publication Critical patent/WO2021088422A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
    • G06F3/0484—Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
    • G06F3/04845—Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range for image manipulation, e.g. dragging, rotation, expansion or change of colour
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/451—Execution arrangements for user interfaces
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M1/00—Substation equipment, e.g. for use by subscribers
    • H04M1/72—Mobile telephones; Cordless telephones, i.e. devices for establishing wireless links to base stations without route selection
    • H04M1/724—User interfaces specially adapted for cordless or mobile telephones
    • H04M1/72403—User interfaces specially adapted for cordless or mobile telephones with means for local support of applications that increase the functionality
    • H04M1/7243—User interfaces specially adapted for cordless or mobile telephones with means for local support of applications that increase the functionality with interactive means for internal management of messages

Definitions

  • the embodiments of this specification relate to the field of computer technology, and specifically, to a notification method and device for application messages.
  • mainstream operating systems such as Apple iOS, Android, Aliyun OS, etc.
  • preset notification methods for application messages such as notifications based on desktop icons in the form of corner labels, and notifications in the form of banners in the display interface and many more.
  • the current notification method is relatively fixed and single, and cannot meet the needs of users in various aspects.
  • the notification content for an application message will obscure the content originally displayed on the screen, which easily interferes with the user's operation and affects the user experience.
  • One or more embodiments of this specification provide a notification method for application messages, which combines image processing technology and artificial intelligence technology to quickly analyze the environment of the terminal device, and then give a reasonable notification method to help users enjoy the smooth operation of the terminal. At the same time, no application message is missed.
  • a method for notifying an application message wherein the execution subject of the method is a terminal device, and the method includes: intercepting an interface displayed in the terminal device in response to receiving a notification request of the application message , Obtain several screenshot pictures; Input the several screenshot pictures into the first area prediction model to obtain corresponding first display areas; The first area prediction model is pre-trained based on multiple labeled sample screenshot pictures, The annotation corresponds to an area in the sample screenshot picture that does not contain important content; based on the several first display areas, a target display area is determined for displaying the notification content for the application message.
  • the interface displayed in the terminal device is captured to obtain a number of screenshots, including: in response to receiving the notification request, detecting the terminal device Whether the screen is in the unlocked state; in the case of the unlocked state, take a screenshot of the displayed interface to obtain the several screenshot pictures.
  • inputting the several screenshot pictures into the first region prediction model to obtain the corresponding first display regions includes: extracting the color features and/or texture features of the several screenshot pictures respectively, Obtain several corresponding picture feature vectors; input the several picture feature vectors into the first region prediction model to obtain the several first display regions.
  • capturing the interface displayed in the terminal device to obtain several screenshot pictures includes: continuously capturing the interface displayed within a predetermined period of time to obtain multiple screenshot pictures.
  • Inputting the plurality of screenshot pictures into the first region prediction model to obtain corresponding first display regions includes: inputting the plurality of screenshot pictures into the first region prediction model to obtain the corresponding plurality of screenshot pictures.
  • determining the target display area based on the plurality of first display areas includes: taking the plurality of first display areas as an area sequence and inputting it into a pre-trained second area prediction model to obtain the target display area.
  • the second region prediction model is pre-trained based on the following steps: acquiring a plurality of training samples, wherein each training sample includes a corresponding historical region sequence and a display region marked with a historical target Annotated screenshot pictures; wherein the sequence of historical regions includes a plurality of regions obtained by inputting multiple historical screenshot pictures into the first region prediction model, and the multiple historical screenshot pictures are captured in the annotated screenshot pictures A picture taken within the predetermined period of time before the time; using the multiple training samples to train the second region prediction model.
  • the method further includes: acquiring characteristic information of a predetermined type, the predetermined type including one or more of the following: device information of the terminal device , The application information of the application to which the displayed interface belongs, the operation behavior data generated by operating the terminal device within a predetermined historical period; the characteristic information is input into a pre-trained duration prediction model, and a notification for the application message is obtained duration.
  • the method further includes: inputting the target display area and the notification duration together into pre-training In the style prediction model of, the notification style for the application message is obtained.
  • the notification style belongs to one of the following: a bullet screen style, a simple style of the number of messages, a banner style, a horizontal screen subtitle style, and a horizontal screen couplet style.
  • the method further includes: inputting the target display area into a pre-trained font color prediction model to obtain the font color used to display the notification content.
  • the method further includes: inputting the target display area into a pre-trained multi-task model to obtain the font color, background color, and background for displaying the notification content transparency.
  • the method further includes: displaying the notification content in the target display area for notifying the user that the application message has been received.
  • a notification device for application messages the device is integrated in a terminal device, and the device includes: a screen capture unit configured to respond to receiving a notification request for an application message, The interface is intercepted to obtain several screenshots; the area prediction unit is configured to input the several screenshots into the first area prediction model to obtain corresponding first display areas; the first area prediction model is based on multiple strips The marked sample screenshot pictures are pre-trained, and the annotation corresponds to an area that does not contain important content in the sample screenshot picture; the target area determining unit is configured to determine a target display area based on the plurality of first display areas for displaying the target The notification content of the application message.
  • a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method described in the first aspect.
  • a computing device including a memory and a processor, characterized in that executable code is stored in the memory, and when the processor executes the executable code, the implementation described in the first aspect is implemented.
  • Using the above-mentioned method and device disclosed in the embodiments of this specification can make the message notification method fit the environment of the terminal device, so that the user will not be disturbed without missing any message, and enjoy a smooth experience of using the terminal.
  • Fig. 1 shows a schematic diagram of a message notification decision flow according to an embodiment
  • Figure 2 shows a schematic diagram of a message notification decision flow according to another embodiment
  • Fig. 3 shows a schematic flowchart of a method for notifying application messages according to an embodiment
  • FIG. 4 shows a schematic diagram of notification in a barrage style according to an example
  • FIG. 5 shows a schematic diagram of notification in a simple style of the number of messages according to an example
  • Fig. 6 shows a schematic diagram of a banner-style notification according to an example
  • FIG. 7 shows a schematic diagram of notification of a horizontal subtitle style according to an example
  • FIG. 8 shows a schematic diagram of a notification in a horizontal screen couplet style according to an example
  • Fig. 9 shows a schematic diagram of a message notification according to an example
  • FIG. 10 shows a schematic diagram of message notification according to another example
  • Fig. 11 shows a flowchart of a message notification method according to an embodiment
  • Fig. 12 shows a structural diagram of an application message notification device according to an embodiment.
  • the notification methods preset in major operating systems usually display the notification content at a certain fixed position of the screen (for example, the top of the screen or the center of the screen), which is likely to cause interference to the user. For example, if the user is viewing a picture, and the message notification displayed at this time obscures the key content in the picture, this will result in a decrease in user experience. For another example, if the user is operating through the buttons displayed in the interface, and the message notification displayed at this time obscures the buttons displayed in the interface, the user accidentally clicks on the message notification and jumps to another interface, and cannot continue. This will seriously affect the user experience.
  • the embodiment of this specification provides a notification method for application messages, which combines image processing technology and artificial intelligence AI technology to quickly analyze the environment of the user terminal (that is, the terminal device used by the user), and then give a reasonable notification method. Help users not miss any important application message while enjoying the smooth operation of the terminal.
  • FIG. 1 shows a schematic diagram of a message notification decision flow according to an embodiment.
  • a screenshot operation may be performed on a terminal device (hereinafter or referred to as a terminal) to obtain a screenshot picture 110.
  • the screenshot picture reflects the display content in the interface.
  • the pre-trained first area prediction model 120 is used to identify the first target area 130 that does not include important content in the screenshot picture as the target display area, and then the notification content 140 for the above-mentioned certain application message is displayed in the target display area. In this way, the interference of the message notification to the user can be reduced.
  • the inventor also considers that when the user uses the mobile terminal, the content of the terminal interface often changes dynamically. Specifically, when a user performs an interface operation, the content of the interface changes based on the operation instruction received by the terminal. For example, when a user browses a website, he slides up or down to view different parts of the webpage content. In addition, when users view video pictures and other impact data, even if they do not perform interface operations, the content played in the interface will continue to change. For example, when a user is chasing a drama, the video frame played in the terminal is constantly changing.
  • FIG. 2 shows a schematic diagram of a message notification decision flow according to another embodiment.
  • the terminal device in the case where it is necessary to notify the user that a certain application message has been received, the terminal device can be continuously screen-captured within a predetermined period of time (such as 1s) to obtain a predetermined number (such as 3) Multiple screenshots of, including picture 211, picture 212, and picture 213, for example.
  • the pre-trained first area prediction model 120 is used to identify multiple first display areas that do not include important content in multiple screenshot pictures, for example, including area 231, area 232, and area 233.
  • the multiple first display areas are input into the pre-trained second area prediction model 240 as area sequences to obtain the target display area 250, and the notification content 260 is displayed therein. In this way, the interference of the message notification to the user can be reduced more effectively.
  • a number of screenshots are obtained by screenshot operations on the terminal interface, and then image processing technology and AI technology are used to process a number of screenshots, which can be predicted
  • a target display area that does not contain important interface content is used to display notification content for application messages, thereby effectively reducing interference to users, ensuring the smoothness of users' use of the terminal, and not missing any application messages.
  • FIG. 3 shows a schematic flowchart of a method for notifying application messages according to an embodiment.
  • the execution subject of the method may be a terminal device, including a mobile phone, a wearable device, a tablet, a computer, and so on. More specifically, the execution subject may be an operating system (OS for short) or a system plug-in of the terminal device. As shown in FIG.
  • OS operating system
  • the method includes the following steps: step S310, in response to receiving the notification request of the application message, intercept the interface displayed in the terminal device to obtain several screenshots; step S320, combine the several The screenshot pictures are respectively input into the pre-trained first area prediction model to obtain several corresponding first display areas; step S330, based on the several first display areas, determine a target display area for displaying notifications for the application message content.
  • step S310 in response to receiving the notification request of the application message, the interface displayed in the terminal device is intercepted to obtain several screenshots.
  • the OS of the device sends a notification request for the application message, and when the OS determines that the certain application has the notification authority (generally preset by the user), the application message is notified.
  • the notification authority generally preset by the user
  • Dingding APP receives a conversation message from another Dingding user sent to the current user from the application server, it determines that the interface of the terminal device is not the conversation interface corresponding to the conversation message, and then sends the conversation message to the OS.
  • the notification request for the terminal device to notify the session message. In this way, the terminal device can receive the notification request for the application message.
  • the interface displayed in the terminal device is intercepted to obtain several screenshots.
  • the terminal considering that if the terminal is in the locked screen state, it means that the user is not currently using the terminal, so the message notification will not interfere with the user's operation of the terminal, and the existing notification method can be used for message notification.
  • the screen of the terminal is unlocked, the user is likely to be using the terminal. At this time, it can be known that the method provided in this manual is used to determine the message notification mode.
  • the screen of the terminal device in response to receiving the notification request, it is detected whether the screen of the terminal device is in an unlocked state. Further, in a specific embodiment, in the unlocked state, the displayed interface is captured to obtain the several screenshot pictures. In another specific embodiment, in the case of being in the locked state, the message notification is performed according to the existing notification mode.
  • the specific number of several screenshot pictures can be a predetermined value.
  • this step may include: intercepting the interface displayed in the terminal device to obtain a screenshot picture, that is, the predetermined value is 1.
  • this step may include: intercepting the interface displayed within a predetermined time period to obtain multiple screenshot pictures, wherein the number of the multiple screenshot pictures corresponds to the predetermined value.
  • the displayed interface may be captured at a predetermined time interval (such as 1S) within the predetermined time period (such as 3S) to obtain multiple (such as 4) screenshots.
  • the specific number of several screenshot pictures can also be determined based on the mapping relationship between the preset application category and the number of screenshot pictures.
  • the application categories can include e-books, office, social, video, etc.
  • the number of pictures corresponding to the first category can be one, and the number of pictures corresponding to the last three categories can be For 3 sheets.
  • the application category of the application to which the interface displayed in the terminal device belongs is determined, and then based on the mapping relationship between the application category and the number of screenshots The number of pictures that need to be captured, and then take a screenshot of the interface displayed on the terminal to obtain a screenshot of the number of pictures.
  • step S320 the several screenshot pictures are respectively input into the pre-trained first region prediction model to obtain several corresponding first display regions.
  • the above-mentioned first region prediction model may be based on a target detection algorithm, a target instance segmentation algorithm, or a target key point detection algorithm. In a specific embodiment, it may be based on one or more of the following algorithms: R-CNN, SPP-NET, Faster-RCNN, R-FCN, and Mask R-CNN.
  • the above-mentioned first region prediction model is pre-trained based on a plurality of sample screenshot pictures with annotations, and the annotation corresponds to an area that does not contain important content in the sample screenshot pictures.
  • the annotation may specifically be the annotation box in the sample screenshot picture, and the area within the annotation box is the above-mentioned area that does not contain important content.
  • the important content differs depending on the screenshots of the sample.
  • the annotator will judge according to some agreed standards.
  • the important content can include characters, text and operation buttons, etc.
  • the annotation box cannot be in the center of the screen. Each picture There can only be one of the label boxes and so on. In this way, the first region prediction model obtained by pre-training can be obtained.
  • this step may include: extracting color features and/or texture features of the several screenshot pictures respectively to obtain several corresponding picture feature vectors; and then inputting the several picture feature vectors into the first In an area prediction model, the plurality of first display areas are obtained.
  • the color features may include HSV (or HSB) features.
  • HSV is a color space created by A.R.Smith in 1978. It includes the intuitive characteristics of colors such as Hue, Saturation and Brightness. It is also called Hexcone Model.
  • the color features may include RGB features. RGB is a spatial color model, which obtains various colors by changing the three color channels of Red, Green, and Blue and superimposing them with each other.
  • the texture feature may include HOG features.
  • the Histogram of Oriented Gradient (HOG) feature is a feature descriptor used for object detection in computer vision and image processing. It composes features by calculating and counting the histogram of the gradient direction of the local area of the image. .
  • this step may include: inputting a screenshot obtained by the screenshot into the first region prediction model to obtain the corresponding first display region. In another embodiment, this step may include: inputting multiple captured screenshots into the first region prediction model to obtain multiple corresponding first display regions.
  • step S330 based on the several first display areas, a target display area is determined for displaying the notification content for the application message.
  • the above-mentioned several first display areas are specifically one first display area, and thus the one first display area can be directly used as the above-mentioned target display area.
  • the above-mentioned several first display areas are specifically multiple first display areas, so the multiple first display areas can be used as a sequence of areas and input into a pre-trained second area prediction model to obtain the target Display area.
  • the plurality of first display areas are arranged in sequence based on the corresponding interception time, and then constitute the above-mentioned area sequence.
  • the second region prediction model is pre-trained based on the following steps: first, a plurality of training samples are obtained, wherein each training sample includes a corresponding historical region sequence and a label labeled with a historical target display region Screenshot pictures; wherein the historical region sequence includes multiple regions obtained by inputting multiple historical screenshot pictures into the first region prediction model, and the multiple historical screenshot pictures are the above-mentioned before the moment of the capture of the marked screenshot picture Pictures taken within a predetermined period of time. Then, the multiple training samples are used to train the second region prediction model. Using the second region prediction model trained in this way, the target display region suitable for displaying notification content can be predicted according to the environment where the terminal is located in the next stage.
  • a certain first display area can also be randomly selected from a number of first display areas as the target display area.
  • the target display area may be filtered based on a preset blacklist strategy.
  • the preset bottom display area for example, the top area of the screen
  • the target display area is still used to display the notification content.
  • the aforementioned blacklist strategy may include: not allowing the target display area to be located in the center of the screen.
  • the central area of the screen can be preset. When the ratio between the overlapping area of the target display area and the central area and the area of the central area is greater than a predetermined threshold, it is determined that the target display area is in the center of the screen. Then the target display area is updated with the bottom display area.
  • the aforementioned blacklist policy may include: not allowing the target display area to be located in an area frequently operated by the user.
  • the user's operation behavior data in a predetermined historical period can be collected to determine the user's operation area, and then determine whether the target display area is located in the user's operation area.
  • the user's operating data within the last 5 minutes can be collected, including the touch points to the screen during the operation within the 5 minutes, and the number of touches or the duration of each touch point, and the number of touches is greater than the predetermined
  • the number of times threshold (such as 3 times) and the contacts whose pressing duration exceeds the predetermined time threshold (such as 4s) are regarded as frequent contacts, and then the smallest area covering the frequent contacts is determined as the frequent operation area. Further, in the case where there is an overlap between the target display area and the frequent operation area, it is determined that the target display area is located in an area frequently operated by the user, and then the target display area is updated with the bottom display area.
  • the target display area can be determined for displaying notification content for application messages. It is understandable that the display of the notification content not only relates to the display position of the notification content on the screen, but also relates to the display duration of the notification content (or notification duration, stay duration), the font, font size, font color, and color of the notification content. Background color, background color transparency, font transparency, etc.
  • the content other than the display position described above may be preset, so that the notification content is displayed according to the preset other content and the target display area. In another embodiment, some of the other contents can also be selectively determined.
  • the aforementioned notification duration for the notification content may also be determined.
  • the method may further include: first, obtaining characteristic information of a predetermined type, wherein the predetermined type includes one or more of the following: The device information of the terminal device, the application information of the application to which the displayed interface belongs, and the operation behavior data generated by operating the terminal device within a predetermined historical period; and then input the characteristic information into a pre-trained duration prediction model To get the notification duration for the application message.
  • the device information may include the screen resolution, CPU information, screen size, and basic hardware information of the terminal device.
  • the application information may include application categories, such as social, browser, e-book, game, or video.
  • the description of the operation behavior data can refer to the relevant description in the foregoing embodiment, and will not be repeated.
  • the duration prediction model can use Logistic Regression (LR) algorithm, Random Forest (Random Forest) algorithm, Gradient Boosting Descision Tree (GBDT) algorithm and XGBOOST Algorithm etc.
  • LR Logistic Regression
  • Random Forest Random Forest
  • GBDT Gradient Boosting Descision Tree
  • XGBOOST Algorithm XGBOOST Algorithm
  • the collected feature information can be input into a pre-trained duration prediction model to obtain the notification duration for the application message.
  • the mapping relationship between the application category (referring to the application category of the application to which the terminal display interface belongs) and the notification duration can be established in advance, so that the corresponding notification duration is determined based on the collected application category.
  • the mapping relationship includes: the notification duration corresponding to the browser category and the e-book category is 2s, the notification duration corresponding to the social category is 1.5S, and the notification duration corresponding to the game category and the video category is 1S.
  • the collected application category on the terminal display interface is a social category (such as Dingding)
  • the corresponding notification duration is determined to be 1.5s. In this way, the application category corresponding to the terminal display interface can be collected, and then the corresponding notification duration can be determined based on the pre-established mapping relationship.
  • the method for determining the notification duration is introduced. Further, it is also possible to determine the notification style for the application message based on the determined target display area and the notification duration described above.
  • the notification style is one of a variety of alternative notification styles. The following first introduces the alternative notification styles.
  • the design elements corresponding to the notification style may include notification items of the notification content (such as the number of messages or message content of the application message), text direction, font (such as Song Ti or Microsoft Yahei), font size, Word count threshold (such as 10 or 20) and dynamic effects (such as moving from right to left, moving from top to bottom, etc.). Based on this, multiple alternative notification styles can be pre-designed by the staff.
  • multiple alternative notification styles may include: barrage style, simple style of message number, banner style, horizontal screen subtitle style, horizontal screen couplet style.
  • barrage style simple style of message number
  • banner style banner style
  • horizontal screen subtitle style horizontal screen couplet style.
  • FIG. 4 shows a schematic diagram of a notification in a barrage style according to an example.
  • the notification content 410 in the barrage style will move from right to left.
  • FIG. 5 shows a schematic diagram of notification of the number of messages in a simple style according to an example.
  • the notification content 510 in the simple style may include the number of new messages that need to be notified.
  • FIG. 6 shows a schematic diagram of a notification in a banner style according to an example.
  • the notification content 610 in the banner style can be slid into the target display from either side of the screen (such as upper or lower or left or right) Area.
  • FIG. 7 shows a schematic diagram of a notification in a horizontal subtitle style according to an example.
  • the notification content 710 in the horizontal subtitle style may be like a normal video subtitle, and stays still and stays for a certain period of time after it appears.
  • Figure 8 shows a schematic diagram of a notification in a horizontal screen couplet style according to an example. See Figure 8, where the notification content 810 can be displayed on one side in the target display area. In addition, when the number of words in the notification content 820 exceeds the number of characters displayed on one side The threshold value (such as 10) can display the remaining part of the notification content 820 in a display area that is symmetrical about the vertical axis 830 under the horizontal screen in the target display area.
  • the pre-trained style prediction model can be used for prediction.
  • the above-mentioned target display area and the notification duration may be jointly input into the style prediction model to obtain the notification style for the application message.
  • the style prediction model is based on a classification algorithm, which specifically includes a support vector machine SVM algorithm, a decision tree algorithm, a Bayesian classification algorithm, and so on.
  • the style prediction model can be pre-trained based on the following steps: First, multiple labeled samples are obtained.
  • each labeled sample may include a corresponding The target display area and the notification style label can be understood, where the notification style label corresponds to one of the aforementioned alternative notification styles; then, a plurality of label samples are used to train the style prediction model.
  • the font color is one of the design elements of the notification style, that is, the determination of the notification style mentioned above means that the font color is determined accordingly.
  • the font color does not belong to the design element of the notification style. In this case, the font color can be determined additionally.
  • the method may further include: inputting the target display area into a pre-trained font color prediction model to obtain the font color used to display the notification content.
  • the font color prediction model may use LR algorithm, random forest algorithm, etc.
  • the method may further include: first obtaining the color feature of the target display area, and then using a hash algorithm to calculate the hash value of the color feature, based on the pre-calculated Multiple hash values corresponding to multiple candidate font colors, select a hash value that is different from the hash value of the color feature from the multiple hash values, and select the candidate font color corresponding to this hash value Determine the font color of the notification content.
  • the determined font color for displaying the notification content can be different from the color of the original screen content in the target display area or have a larger contrast, which is convenient for the user to view the notification content.
  • the background color and background transparency of the notification content can be set to no background by default in a specific embodiment.
  • both the background color and the background transparency may belong to one of the design elements of the notification style, that is, the determination of the above notification style means that both are determined accordingly.
  • the two do not belong to the design elements of the notification style. In this case, the two can be determined additionally. It needs to be understood that from the perspective of visual effects, font color, background color, and background transparency are usually related. Based on this, a multi-task model can be used to determine these three at the same time.
  • the method may further include: inputting the target display area into a pre-trained multitasking model to obtain the font color, background color, and background transparency used to display the notification content .
  • the multi-task model may use classification algorithms, specifically including support vector machine SVM algorithm, decision tree algorithm, Bayesian classification algorithm, and so on.
  • the multi-task model may include a classification model for three different classification tasks, and specifically may include a first classification model for predicting font colors, a second classification model for predicting background colors, and a second classification model for predicting background transparency. The third classification model. It should be noted that the use process for the multi-task model is similar to the training process, and will not be repeated here.
  • the multitasking model described above can be used to determine the font color, background color, and background transparency.
  • the notification content may have no background by default, and the font color prediction model described above is used to determine the font color.
  • the target display area can be determined for displaying the notification content for the application message.
  • one or more of the following can be determined: notification duration, notification style, font color, background color, and background transparency for the notification content. These determined content can all be used to display the notification content.
  • the method may further include: displaying notification content for the application message in the target display area for notifying the user that the application message has been received.
  • FIG. 9 it shows the notification content displayed in the target display area in a horizontal subtitle style, that is, the pin message: the meeting starts at 10:30, and the notification duration may be 1.5s.
  • a decision on the message notification method is triggered. For example, if an application message A is received at 10:00 in the morning, the above-mentioned method is executed once to determine the message notification method A for the application message A. After that, when the application message B is received at 1:00 pm, the above method is executed again to determine the message notification method B for the application message B.
  • a predetermined time window such as 3S or 5S, etc.
  • the target display area performs message notification.
  • the notification style determined for the first application message is the simple style of the number of messages
  • the number of messages in the notification content will be cumulatively displayed based on the messages received within the time window.
  • a predetermined time window such as 5S
  • the number of messages included in the notification content changes from 1 to 2 and then to 3.
  • using the application message notification method disclosed in the embodiments of this specification can make the message notification method fit the environment of the terminal device, so that the user will not be disturbed without missing any message, and enjoy the benefits of using the terminal. Smooth experience.
  • FIG. 11 shows a flowchart of a message notification method according to an embodiment, and the execution subject of the method may be a terminal device. As shown in FIG. 11, the method flow may include the following steps S1101-step S1114.
  • Step S1101 A notification request of an application message is received.
  • Step S1102 It is detected whether the terminal device is in an unlocked state. In the case of detecting that it is in an unlocked state, in step S1103, a preset lock screen message notification method is used for message notification.
  • steps S1104-step S1106 are executed.
  • step S1104 the interface displayed in the terminal device within a predetermined period of time (such as 2S) is continuously captured to obtain a predetermined number (such as 4) of multiple screenshot pictures.
  • a predetermined period of time such as 2S
  • Step S1105 Input multiple screenshots into the first region prediction model to obtain multiple corresponding first display regions.
  • Step S1106 Input a plurality of first display areas as an area sequence into the second area prediction model to obtain a target display area for displaying notification content for application messages.
  • step S1107 and step S1108 are executed.
  • a predetermined type of characteristic information is obtained, where the predetermined type includes one or more of the following: device information of the terminal device, application information of the application to which the displayed interface belongs, and in a predetermined historical period Operating behavior data generated by internally operating the terminal device.
  • step S1108 input the acquired feature information into a pre-trained duration prediction model to obtain the stay duration for the notification content.
  • Step S1109 Input the target display area and stay time into the pre-trained style prediction model to obtain the notification style for the notification content, where the notification style belongs to one of the following: barrage style, simple style of message number, banner style, Subtitle style and horizontal screen couplet style.
  • Step S1110 It is judged whether the notification style belongs to the first style set.
  • the first style set includes the bullet screen style, the simple style of the number of messages, and the banner style.
  • step S1111 is executed to input the target display area into a pre-trained multi-classification task model to obtain font color, background color, and background brightness.
  • step S1112 the notification content for the application message is displayed based on the determined target display area, notification style, font color, background color, and background brightness.
  • step S1113 is executed to input the target display area into the font color prediction model to obtain the font color.
  • step S1114 based on the determined target display area, notification style, and font color, and without a display background, display the notification content for the application message.
  • using the application message notification method disclosed in the embodiments of this specification can make the message notification method fit the environment of the terminal device, so that the user will not be disturbed without missing any message, and enjoy the benefits of using the terminal. Smooth experience.
  • FIG. 12 shows a structural diagram of an application message notification apparatus according to an embodiment, and the apparatus is integrated in a terminal device.
  • the apparatus 1200 may include: a screenshot unit 1210 configured to, in response to receiving a notification request of an application message, intercept the interface displayed in the terminal device to obtain several screenshot pictures.
  • the area prediction unit 1220 is configured to input the plurality of screenshot pictures into a first area prediction model to obtain a plurality of corresponding first display areas; the first area prediction model is pre-trained based on multiple labeled sample screenshot pictures , The annotation corresponds to an area in the sample screenshot picture that does not contain important content.
  • the target area determining unit 1230 is configured to determine a target display area based on the plurality of first display areas for displaying notification content for the application message.
  • the screen capture unit 1210 is specifically configured to: in response to receiving the notification request, detect whether the screen of the terminal device is in an unlocked state; in the case of an unlocked state, the displayed interface Take a screenshot to obtain the several screenshot pictures.
  • the region prediction unit 1220 is specifically configured to: extract color features and/or texture features of the several screenshot pictures respectively to obtain several corresponding picture feature vectors; and separate the several picture feature vectors Input into the first region prediction model to obtain the plurality of first display regions.
  • the screenshot unit 1210 is specifically configured to continuously capture the interface displayed within a predetermined period of time to obtain multiple screenshot pictures.
  • the area prediction unit 1220 is specifically configured to: input the multiple screenshot pictures into the first area prediction model to obtain corresponding multiple first display areas; wherein the target area determination unit 1230 is specifically configured to: The multiple first display areas are used as a sequence of areas and input into a pre-trained second area prediction model to obtain the target display area.
  • the second region prediction model is pre-trained based on a training unit, and the training unit is configured to: obtain a plurality of training samples, wherein each training sample includes a corresponding historical region sequence and label Annotated screenshot pictures with historical target display areas; wherein the historical area sequence includes a plurality of areas obtained by inputting multiple historical screenshot pictures into the first area prediction model, and the multiple historical screenshot pictures are in the Mark the pictures taken within the predetermined period of time before the time of taking the screenshot pictures. Using the multiple training samples to train the second region prediction model.
  • the apparatus 1200 further includes: a characteristic information acquiring unit 1240 configured to acquire characteristic information of a predetermined type, the predetermined type including one or more of the following: device information of the terminal device, The application information of the application to which the displayed interface belongs, and the operation behavior data generated by operating the terminal device within a predetermined historical period; the duration prediction unit 1250 is configured to input the characteristic information into a pre-trained duration prediction model to obtain The notification duration of the application message.
  • a characteristic information acquiring unit 1240 configured to acquire characteristic information of a predetermined type, the predetermined type including one or more of the following: device information of the terminal device, The application information of the application to which the displayed interface belongs, and the operation behavior data generated by operating the terminal device within a predetermined historical period
  • the duration prediction unit 1250 is configured to input the characteristic information into a pre-trained duration prediction model to obtain The notification duration of the application message.
  • the device 1200 further includes a style prediction unit 1260 configured to input the target display area and the notification duration into a pre-trained style prediction model to obtain the notification style for the application message.
  • a style prediction unit 1260 configured to input the target display area and the notification duration into a pre-trained style prediction model to obtain the notification style for the application message.
  • the notification style belongs to one of the following: a bullet screen style, a simple style of the number of messages, a banner style, a horizontal screen subtitle style, and a horizontal screen couplet style.
  • the device 1200 further includes: a font color prediction unit 1270, configured to input the target display area into a pre-trained font color prediction model to obtain a font color for displaying the notification content.
  • a font color prediction unit 1270 configured to input the target display area into a pre-trained font color prediction model to obtain a font color for displaying the notification content.
  • the device 1200 further includes: a multi-task prediction unit 1280 configured to input the target display area into a pre-trained multi-task model to obtain the font color and background color used to display the notification content And background transparency.
  • a multi-task prediction unit 1280 configured to input the target display area into a pre-trained multi-task model to obtain the font color and background color used to display the notification content And background transparency.
  • the device 1200 further includes: a display unit 1290 configured to display the notification content in the target display area for notifying the user that the application message has been received.
  • using the application message notification device disclosed in the embodiment of this specification can make the message notification method fit the environment of the terminal device, so that the user will not be disturbed without missing any message, and enjoy the benefits of using the terminal. Smooth experience.
  • a computer-readable storage medium on which a computer program is stored.
  • the computer program is executed in a computer, the computer is caused to execute the description described in conjunction with FIG. 3 or FIG. 11 method.
  • a computing device including a memory and a processor, the memory stores executable code, and when the processor executes the executable code, a combination of FIG. 3 or FIG. 11 is implemented. The described method.

Abstract

本说明书实施例提供一种应用消息的通知方法及装置。其中该通知方法的执行主体为终端设备，该方法包括：首先，响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片；接着，将若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域；所述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域；然后，基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。

Description

应用消息的通知方法及装置 技术领域
本说明书实施例涉及计算机技术领域，具体地，涉及一种应用消息的通知方法及装置。
背景技术
目前主流的操作系统，如苹果iOS、安卓、阿里云OS等，都预置有一些针对应用消息的通知方式，如基于桌面图标以角标形式进行通知，在显示界面中以横幅的形式进行通知等等。
然而，目前的通知方式比较固定、单一，无法满足用户多方面的需求。比如说，在用户操作移动设备的过程中，针对应用消息的通知内容会对屏幕中原本显示的内容造成遮挡，容易干扰用户操作，影响用户体验。
因此，需要一种更加合理的消息通知方式，能够满足用户多方面的需求，包括降低对用户的干扰，从而充分提高用户体验。
发明内容
本说明书一个或多个实施例提供一种应用消息的通知方法，结合图像处理技术和人工智能技术，快速分析终端设备所处环境，然后给出合理的通知方式，帮助用户在享受终端流畅操作的同时，不遗漏任何一条的应用消息。
根据第一方面，提供一种应用消息的通知方法，所述方法的执行主体为终端设备，所述方法包括：响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片；将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域；所述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域；基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
在一个实施例中，响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片，包括：响应于接收到所述通知请求，检测所述终端设备的屏幕是否处于解锁状态；在处于解锁状态的情况下，对所述显示的界面进行截取， 得到所述若干截屏图片。
在一个实施例中，将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域，包括：分别对所述若干截屏图片进行颜色特征和/或纹理特征的提取，得到对应的若干图片特征向量；将所述若干图片特征向量分别输入所述第一区域预测模型中，得到所述若干第一显示区域。
在一个实施例中，其中对所述终端设备中显示的界面进行截取，得到若干截屏图片，包括：对预定时长内显示的界面进行连续截取，得到多张截屏图片。其中将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域，包括：将所述多张截屏图片分别输入所述第一区域预测模型中，得到对应的多个第一显示区域。其中基于所述若干第一显示区域，确定目标显示区域，包括：将所述多个第一显示区域作为区域序列，输入预先训练的第二区域预测模型中，得到所述目标显示区域。
进一步地，在一个具体的实施例中，所述第二区域预测模型基于以下步骤而预先训练：获取多个训练样本，其中每个训练样本中包括对应的历史区域序列和标注有历史目标显示区域的标注截屏图片；其中所述历史区域序列包括将多张历史截屏图片分别输入所述第一区域预测模型而得到的多个区域，所述多张历史截屏图片是在所述标注截屏图片的截取时刻之前的所述预定时长内截取的图片；利用所述多个训练样本，训练所述第二区域预测模型。
在一个实施例中，在接收到应用消息的通知请求之后，所述方法还包括：获取预定类型的特征信息，所述预定类型包括以下中的一种或多种：所述终端设备的设备信息，所述显示的界面所属应用的应用信息，在预定历史时段内操作所述终端设备产生的操作行为数据；将所述特征信息输入预先训练的时长预测模型中，得到针对所述应用消息的通知时长。
进一步地，在一个具体的实施例中，在确定目标显示区域，以及，得到针对所述应用消息的通知时长之后，所述方法还包括：将所述目标显示区域和通知时长，共同输入预先训练的样式预测模型中，得到针对所述应用消息的通知样式。
在一个更具体的实施例中，所述通知样式属于以下中的一种：弹幕样式、消息条数简易样式、横幅样式、横屏字幕样式、横屏对联样式。
在一个实施例中，在确定目标显示区域之后，所述方法还包括：将所述目标显示区域输入预先训练的字体颜色预测模型中，得到用于显示所述通知内容的字体颜色。
在一个实施例中，在确定目标显示区域之后，所述方法还包括：将所述目标显示区域输入预先训练的多任务模型中，得到用于显示所述通知内容的字体颜色、背景颜色和背景透明度。
在一个实施例中，在确定目标显示区域之后，所述方法还包括：在所述目标显示区域中显示所述通知内容，用于通知用户已接收到所述应用消息。
根据第二方面，提供一种应用消息的通知装置，所述装置集成于终端设备，所述装置包括：截屏单元，配置为响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片；区域预测单元，配置为将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域；所述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域；目标区域确定单元，配置为基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
根据第三方面，提供了一种计算机可读存储介质，其上存储有计算机程序，当所述计算机程序在计算机中执行时，令计算机执行第一方面所描述的方法。
根据第四方面，提供了一种计算设备，包括存储器和处理器，其特征在于，所述存储器中存储有可执行代码，所述处理器执行所述可执行代码时，实现第一方面所描述的方法。
采用本说明书实施例披露的上述方法及装置，可以使得消息通知方式契合终端设备所处的环境，从而使用户在不遗漏任何一条消息的情况下，不被干扰，享有使用终端的流畅体验。
附图说明
为了更清楚地说明本说明书披露的多个实施例的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本说明书披露的多个实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的附图。
图1示出根据一个实施例的消息通知的决策流程示意图；
图2示出根据另一个实施例的消息通知的决策流程示意图；
图3示出根据一个实施例的应用消息的通知方法流程示意图；
图4示出根据一个例子的弹幕样式的通知示意图；
图5示出根据一个例子的消息条数简易样式的通知示意图；
图6示出根据一个例子的横幅样式的通知示意图；
图7示出根据一个例子的横屏字幕样式的通知示意图；
图8示出根据一个例子的横屏对联样式的通知示意图；
图9示出根据一个例子的消息通知示意图；
图10示出根据另一个例子的消息通知示意图；
图11示出根据一个实施例的消息通知方法流程框图；
图12示出根据一个实施例的应用消息的通知装置结构图。
具体实施方式
下面结合附图，对本说明书披露的多个实施例进行描述。
如前所述，目前针对应用消息的通知方式比较固定，无法满足用户多方面的需求。具体而言，各大操作系统中预置的通知方式，通常是在屏幕的某一固定位置(如，屏幕顶部或屏幕中央)对通知内容进行显示，这很可能对用户造成干扰。举例来说，若用户正在查看一张图片，而此时显示的消息通知遮挡了图片中的关键内容，这将导致用户体验下降。又比如说，若用户正通过界面中显示的按钮进行操作，而此时显示的消息通知遮挡了界面中显示的按钮，导致用户对消息通知误点击而跳转至其他界面，并且，无法继续之前的操作，这将严重影响用户体验。
基于此，本说明书实施例提供一种应用消息的通知方法，结合图像处理技术和人工智能AI技术，快速分析用户终端(即用户使用的终端设备)所处环境，然后给出合理的通知方式，帮助用户在享受终端流畅操作的同时，不遗漏任何一条重要的应用消息。
具体地，图1示出根据一个实施例的消息通知的决策流程示意图。在一个实施例中，参见图1，在需要通知用户已接收到某条应用消息的情况下，可以对终端设备(以下或简称终端)进行截屏操作，得到截屏图片110。可以理解，其中截屏图片反映界面中的显示内容。接着，利用预先训练的第一区域预测模型120识别截屏图片中不包括重要内 容的第一目标区域130，作为目标显示区域，再在目标显示区域中显示针对上述某条应用消息的通知内容140。如此，可以降低消息通知对用户的干扰。
进一步地，发明人还考虑到，在用户使用移动终端的过程中，终端界面的内容往往是动态变化的。具体而言，用户在进行界面操作时，界面内容会基于终端接收到的操作指令而变化。比如说，用户在浏览网站时，会通过上滑或下滑操作，查看网页内容的不同部分。另外，用户在查看视频图片等影响资料时，即使不进行界面操作，界面中播放的内容也会不断发生变化。比如说，用户在追剧时，终端中播放的视频帧是不断变化的。
基于此，在本说明书实施例提供的一种应用消息的通知方法中，还可以通过快速分析用户终端所处环境来预测下一时刻可能的环境，然后给出更加合理的通知方式。
具体地，图2示出根据另一个实施例的消息通知的决策流程示意图。在一个实施例中，参见图2，在需要通知用户已接收到某条应用消息的情况下，可以在预定时长(如1s)内对终端设备进行连续截屏操作，得到预定数量(如3张)的多张截屏图片，例如包括图片211、图片212和图片213。接着，利用预先训练的第一区域预测模型120分别识别多张截屏图片中不包括重要内容的多个第一显示区域，例如包括区域231、区域232和区域233。再将多个第一显示区域作为区域序列输入预先训练的第二区域预测模型240中，得到目标显示区域250，并在其中显示通知内容260。如此，可以更加有效地降低消息通知对用户的干扰。
由上可知，在本说明书实施例提供的一种应用消息的通知方法中，通过对终端界面进行截图操作，得到若干截屏图片，再采用图像处理技术和AI技术对若干截屏图片进行处理，可以预测出不包含重要界面内容的目标显示区域，用于显示针对应用消息的通知内容，进而有效降低对用户的干扰，保证用户对终端使用的流畅度，同时不会遗漏任何一条应用消息。
下面结合实施例，描述上述方法的具体实施步骤。
具体地，图3示出根据一个实施例的应用消息的通知方法流程示意图，所述方法的执行主体可以为终端设备，包括手机、可穿戴设备、平板、电脑等等。更具体地，所述执行主体可以为终端设备的操作系统(operating system，简称OS)或系统插件等。如图3所示，所述方法包括以下步骤：步骤S310，响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片；步骤S320，将所述若干截屏图片分别输入预先训练的第一区域预测模型中，得到对应的若干第一显示区域；步 骤S330，基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
以上步骤具体如下。
首先，在步骤S310，响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片。
需要理解，对于安装在终端设备上的多个应用，通常，当其中某个应用从对应的应用服务器接收到应用消息时，在判断出需要对该应用消息进行通知的情况下，通常会向终端设备的OS发送针对该应用消息的通知请求，在OS判断出该某个应用具有通知权限(一般由用户预先设置)的情况下，对该应用消息进行通知。比如说，当钉钉APP从应用服务器接收到其他钉钉用户发送给当前用户的会话消息时，在判断出终端设备的界面并非与该会话消息对应的会话界面，则向OS发送针对该会话消息的通知请求，以使终端设备对该会话消息进行通知。如此，终端设备可以接收到针对应用消息的通知请求。
基于此，在本步骤中，响应于接收到应用消息的通知请求，对终端设备中显示的界面进行截取，得到若干截屏图片。此外，考虑到若终端处于锁屏状态，意味着用户当前并未使用该终端，因此消息通知并不会干扰到用户对终端的操作，可以使用已有的通知方式进行消息通知。而在终端屏幕处于解锁的状态下，用户很可能正在使用该终端，此时可知采用本说明书提供的方法确定消息通知方式。
具体地，在一个实施例中，响应于接收到所述通知请求，检测所述终端设备的屏幕是否处于解锁状态。进一步地，在一个具体的实施例中，在处于解锁状态的情况下，对所述显示的界面进行截取，得到所述若干截屏图片。在另一个具体的实施例中，在处于锁定状态的情况下，按照已有的通知方式进行消息通知。
另一方面，对于若干截屏图片的具体数量，可以是一个预定值。在一个实施例中，本步骤中可以包括：对所述终端设备中显示的界面进行截取，得到1张截屏图片，也就是预定值为1。在另一个实施例中，本步骤中可以包括：对预定时长内显示的界面进行截取，得到多张截屏图片，其中多张截屏图片的数量对应于上述预定值。在一个具体的实施例中，还可以在所述预定时长(如3S)内，按照预定时间间隔(如1S)对显示的界面进行截取，得到多张(如4张)截屏图片。
另外，对于若干截屏图片的具体数量，还可以基于预先设定的应用类别与截屏图片 数量之间的映射关系而确定。在一个具体的实施例中，其中应用类别可以包括电子书类、办公类、社交类、视频类等，其中第一个类别对应的图片数量可以为1张，后三个类别对应的图片数量可以为3张。基于此，在一个实施例中，响应于接收到应用消息的通知请求，确定终端设备中显示的界面所属应用的应用类别，再基于上述应用类别与截屏图片数量之间的映射关系，确定此时需要截取的图片数量，再对终端显示的界面进行截屏，得到该图片数量的截屏图片。
以上，可以得到若干截屏图片。接着在步骤S320，将所述若干截屏图片分别输入预先训练的第一区域预测模型中，得到对应的若干第一显示区域。
在一个实施例中，上述第一区域预测模型可以基于目标检测算法、目标实例分割算法或目标关键点检测算法。在一个具体的实施例中，可以基于以下算法中的一种或几种：R-CNN、SPP-NET、Faster-RCNN、R-FCN和Mask R-CNN。
在一个实施例中，上述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域。需要说明的是，其中标注具体可以为样本截屏图片中的标注框，标注框内的区域为上述不包含重要内容的区域。其中重要内容因样本截屏图片的不同而不同，由标注人员按照一些约定标准进行判断，比如说，重要内容可以包括人物、文字和操作按钮等，标注框不可以在屏幕的正中央，每张图片的标注框只能有一个等等。如此，可以得到预先训练得到的第一区域预测模型。
在一个实施例中，本步骤可以包括：分别对所述若干截屏图片进行颜色特征和/或纹理特征的提取，得到对应的若干图片特征向量；再将所述若干图片特征向量分别输入所述第一区域预测模型中，得到所述若干第一显示区域。
在一个具体的实施例中，其中颜色特征可以包括HSV(或称HSB)特征。HSV是由A.R.Smith在1978年创建的一种颜色空间，包括色调(Hue)，饱和度(Saturation)和亮度(Value或Brightness)等颜色的直观特性，也称六角椎体模型(Hexcone Model)。在一个具体的实施例中，其中颜色特征可以包括RGB特征。RGB是一种空间颜色模型，是通过对红(Red)、绿(Green)和蓝(Blue)三个颜色通道的变化以及它们相互之间的叠加来得到各式各样的颜色。在一个具体的实施例中，其中纹理特征可以包括HOG特征。方向梯度直方图(Histogram of Oriented Gradient，简称HOG)特征是一种在计算机视觉和图像处理中用来进行物体检测的特征描述子，其通过计算和统计图像局部区域的梯度方向直方图来构成特征。
在一个实施例中，本步骤可以包括：将截取得到的一张截屏图片输入第一区域预测模型中，得到对应的第一显示区域。在另一个实施例中，本步骤可以包括：将截取得到的多种截屏图片输入第一区域预测模型中，得到对应的多个第一显示区域。
以上，利用预先训练得到的第一区域预测模型，可以识别出若干截屏图片对应的若干第一显示区域。然后，在步骤S330，基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
在一个实施例中，上述若干第一显示区域具体为一个第一显示区域，由此可以将该一个第一显示区域直接作为上述目标显示区域。
在一个实施例中，上述若干第一显示区域具体为多个第一显示区域，由此可以将多个第一显示区域作为区域序列，输入预先训练的第二区域预测模型中，得到所述目标显示区域。在一个具体的实施例中，其中多个第一显示区域基于对应的截取时刻而顺序排列，进而组成上述区域序列。
在一个具体的实施例中，其中第二区域预测模型基于以下步骤而预先训练：首先，获取多个训练样本，其中每个训练样本中包括对应的历史区域序列和标注有历史目标显示区域的标注截屏图片；其中历史区域序列包括将多张历史截屏图片分别输入所述第一区域预测模型而得到的多个区域，所述多张历史截屏图片是在所述标注截屏图片的截取时刻之前的上述预定时长内截取的图片。然后，利用所述多个训练样本，训练所述第二区域预测模型。利用如此训练得到的第二区域预测模型，可以根据终端所处环境预测下一阶段所处环境时适合显示通知内容的目标显示区域。
可选地，还可以从若干第一显示区域中随机选取某个第一显示区域作为上述目标显示区域。
以上可以确定出目标显示区域。进一步地，在一个实施例中，还可以基于预先设定的黑名单策略，对目标显示区域进行过滤。在目标显示区域落入黑名单的情况下，使用预设的兜底显示区域(例如屏幕顶部区域)对该目标显示区域进行更新。而在目标显示区域未落入黑名单的情况下，仍使用该目标显示区域显示通知内容。
在一个具体地实施例中，上述黑名单策略可以包括：不允许目标显示区域位于屏幕的正中央。在一个例子中，可以预先设定屏幕的中央区域，当目标显示区域与该中央区域的交叠面积和中央区域的面积之间的比值大于预定阈值，则判定目标显示区域在屏幕的正中央，进而用兜底显示区域对该目标显示区域进行更新。
在另一个具体的实施例中，上述黑名单策略可以包括：不允许目标显示区域位于用户频繁操作的区域。在一个例子中，可以采集用户在预定历史时段内的操作行为数据，用于确定用户的操作区域，再判定目标显示区域是否位于用户的操作区域。在一个具体的例子中，可以采集用户在最近5min内的操作数据，包括这5min内操作时对屏幕的触点，以及每个触点的点触次数或触压时长，将点触次数大于预定次数阈值(如3次)和触压时长超过预定时长阈值(如4s)的触点，作为频繁触点，再确定出覆盖频繁触点的最小区域作为频繁操作区域。进一步地，在目标显示区域与频繁操作区域存在交叠的情况下，判定目标显示区域位于用户频繁操作的区域，进而用兜底显示区域对该目标显示区域进行更新。
以上，可以确定出目标显示区域，用于显示针对应用消息的通知内容。可以理解，对通知内容的显示，不仅涉及到该通知内容在屏幕中的显示位置，还涉及到通知内容的显示时长(或称通知时长、停留时长)，通知内容的字体、字号、字体颜色、背景色、背景色透明度、字体透明度等内容。在一个实施例中，可以预先设定上述除显示位置以外的其他内容，由此按照预先设定的其他内容和目标显示区域对通知内容进行展示。在另一个实施例中，还可以对其他内容中的某些内容进行选择性确定。
在一个具体的实施例中，还可以确定上述针对通知内容的通知时长。
在一个更具体的实施例中，在上述接收到应用消息的通知请求之后，所述方法还可以包括：首先，获取预定类型的特征信息，其中预定类型包括以下中的一种或几种：所述终端设备的设备信息，所述显示的界面所属应用的应用信息，以及，在预定历史时段内操作所述终端设备产生的操作行为数据；再将所述特征信息输入预先训练的时长预测模型中，得到针对所述应用消息的通知时长。
在一个例子中，其中设备信息可以包括终端设备的屏幕分辨率、CPU信息、屏幕尺寸和硬件基础信息等。在一个例子中，其中应用信息可以包括应用类别，具体如社交类、浏览器类、电子书类、游戏类或视频类等。此外需要说明，对操作行为数据的描述可以参见前述实施例中的相关描述，不作赘述。
另一方面，在一个例子中，其中时长预测模型可以采用逻辑回归(Logistic Regression，简称LR)算法，随机森林(Random Forest)算法、梯度提升决策树(Gradient Boosting Descision Tree，简称GBDT)算法和XGBOOST算法等。此外，时长预测模型的使用和训练过程类似，故在此不对时长预测模型的训练进行赘述。
由此，可以将采集的特征信息输入预先训练的时长预测模型中，得到针对所述应用消息的通知时长。
在另一个更具体的实施例中，可以预先建立应用类别(指终端显示界面所属应用的应用类别)和通知时长的映射关系，由此，再基于采集的应用类别，确定对应的通知时长。在一个例子中，假定该映射关系包括：浏览器类和电子书类对应的通知时长为2s、社交类对应的通知时长为1.5S、游戏类和视频类对应的通知时长为1S。基于此，假定采集到的终端显示界面所述应用的类别为社交类(如钉钉)，则将对应的通知时长确定为1.5s。如此，可以采集终端显示界面对应的应用类别，再基于预先建立的映射关系，确定对应的通知时长。
以上，对通知时长的确定方式进行介绍。进一步地，还可以基于上述确定的目标显示区域和通知时长，确定针对应用消息的通知样式。其中通知样式属于备选的多种通知样式中的一种。下面先对备选的多种通知样式进行介绍。在一个具体的实施例中，通知样式对应的设计元素，可以包括通知内容的通知事项(如应用消息的消息条数或消息内容)、文字方向、字体(如宋体或微软雅黑)、字号、字数阈值(如10或20)和动态效果(如从右往左移动、从上向下移动等)。基于此，备选的多种通知样式可以由工作人员预先设计。在一个具体的实施例中，备选的多种通知样式可以包括：弹幕样式、消息条数简易样式、横幅样式、横屏字幕样式、横屏对联样式。下面再结合图4至图8对这些通知样式进行描述，需要理解，图4至图8中示出的通知内容仅用于示意通知样式，其中通知内容的显示位置并不是唯一固定的，而是取决于前述确定出的目标显示区域。
具体地，图4示出根据一个例子的弹幕样式的通知示意图，参见图4，弹幕样式下的通知内容410是会从右向左移动的。图5示出根据一个例子的消息条数简易样式的通知示意图，参见图5，简易样式下的通知内容510可以包括需要通知的新消息条数。图6示出根据一个例子的横幅样式的通知示意图，参见图6，横幅样式下的通知内容610可以从屏幕的任一一侧(如上侧或下侧或左侧或右侧)滑入目标显示区域中。图7示出根据一个例子的横屏字幕样式的通知示意图，参见图7，横屏字幕样式下的通知内容710可以像普通的视频字幕，出现之后静止、停留一定时长。图8示出根据一个例子的横屏对联样式的通知示意图，参见图8，其中通知内容810可以在目标显示区域中进行单侧显示，此外，当通知内容820的字数超过单侧显示的字符数阈值(如10)，可以在目标显示区域关于横屏下的竖直轴830而对称的显示区域，进行通知内容820中剩余部分的显示。
以上对备选的多种通知样式进行介绍。基于此，针对具体的应用消息，具体采用多种通知样式中的哪种，可以利用预先训练的样式预测模型进行预测。在一个具体的实施例中，可以将上述目标显示区域和通知时长，共同输入样式预测模型中，得到针对所述应用消息的通知样式。在一个更具体的实施例中，其中样式预测模型基于分类算法，具体包括支持向量机SVM算法、决策树算法、贝叶斯分类算法等等。在一个更具体的实施例中，其中样式预测模型可以基于以下步骤而预先训练：首先，获取多个标注样本，在一个例子中，可以获取多个标注样本，其中每个标注样本可以包括对应的目标显示区域和通知样式标签，可以理解，其中通知样式标签对应上述备选的多种通知样式中的某一种；接着，利用多个标注样本训练样式预测模型。
以上，可以确定针对应用消息的通知样式。在一个具体的实施例中，字体颜色属于通知样式的设计元素之一，也就是说，上述通知样式的确定意味着字体颜色随之确定。在另一个具体的实施例中，字体颜色不属于通知样式的设计元素，此时，还可以另外再对字体颜色进行确定。
在一个更具体的实施例中，在上述步骤S330之后，所述方法还可以包括：将所述目标显示区域输入预先训练的字体颜色预测模型中，得到用于显示所述通知内容的字体颜色。在一个例子中，其中字体颜色预测模型可以采用LR算法、随机森林算法等。在另一个更具体的实施例中，在上述步骤S330之后，所述方法还可以包括：先获取目标显示区域的颜色特征，再利用哈希算法计算该颜色特征的哈希值，基于预先计算的多个备选字体颜色所对应的多个哈希值，从多个哈希值中选取与该颜色特征的哈希值不相同的哈希值，并将此哈希值对应的备选字体颜色确定为通知内容的字体颜色。如此，可以使得确定的用于显示通知内容的字体颜色与目标显示区域中原屏幕内容的颜色不同或反差较大，便于用户对通知内容的查看。
另外，对于通知内容的背景颜色和背景透明度，在一个具体的实施例中，可以默认设置成无背景。在另一个具体的实施例中，背景颜色和背景透明度两者可以属于通知样式的设计元素之一，也就是说，上述通知样式的确定意味着两者随之确定。在又一个具体的实施例中，两者不属于通知样式的设计元素，此时，还可以另外再对两者进行确定。需要理解，从视觉效果上来看，字体颜色、背景颜色和背景透明度三者通常是具有关联关系的，基于此，可以利用多任务模型，对这三者同时进行确定。
在一个更具体的实施例中，在上述步骤S330之后，所述方法还可以包括：将目标 显示区域输入预先训练的多任务模型中，得到用于显示通知内容的字体颜色、背景颜色和背景透明度。在一个例子中，其中多任务模型可以采用分类算法，具体包括支持向量机SVM算法、决策树算法、贝叶斯分类算法等。在一个例子中，多任务模型可以包括针对三个不同分类任务的分类模型，具体可以包括用于预测字体颜色的第一分类模型，用于预测背景颜色的第二分类模型和用于预测背景透明度的第三分类模型。需要说明的是，针对多任务模型的使用过程和训练过程类似，在此不作赘述。
根据一个具体的例子，在上述确定出的通知样式为弹幕样式、消息条数简易样式和横幅样式中的一个的情况下，可以采用上述多任务模型确定字体颜色、背景颜色和背景透明度。而在上述确定出的通知样式为横屏字幕样式，横屏对联样式的情况下，可以默认通知内容中无背景，并且，采用上述字体颜色预测模型确定字体颜色。
由上可知，通过步骤S310-步骤S330，可以确定出目标显示区域，用于展示针对应用消息的通知内容。此外，还可以确定出以下中的一种或多种：针对通知内容的通知时长、通知样式、字体颜色、背景颜色和背景透明度。这些确定出的内容，均可以用于对通知内容的展示。进一步地，在一个实施例中，在确定以上内容后，所述方法还可以包括：在目标显示区域中显示针对应用消息的通知内容，用于通知用户已接收到该应用消息。在一个例子中，如图9所示，其中示出在目标显示区域中，以横屏字幕样式显示的通知内容，即钉钉消息：会议10:30开始，并且通知时长可以为1.5s。
此外需要说明的是，通常，每接收到一条应用消息，就会触发一次消息通知方式的决策。比如说，上午10:00接收到应用消息A，则执行上述一次上述方法，确定针对该应用消息A的消息通知方式A。之后，下午1:00接收到应用消息B，则再次执行上述方法，确定针对该应用消息B的消息通知方式B。但是，在连续接收到多条应用消息的情况下，通常只会在预定的时间窗口(如3S或5S等)内进行一次决策，然后在该时间窗口内接收到的应用消息都采用该一次决策所确定的消息通知方式。在一个实施例中，假定在第一时刻接收到针对第一应用消息的通知请求，此时，会触发执行上述方法以得到第一目标显示区域，接着在时间窗口内又接收到针对N(N是正整数，如N＝1或5)条应用消息的通知请求，那么，将不会对这N条应用消息另外进行消息通知方式的决策，而是直接利用针对第一应用消息确定出的第一目标显示区域进行消息通知。在一个例子中，假定针对第一应用消息确定出的通知样式是消息条数简易样式，那么会基于在时间窗口内接收到的消息，对通知内容中的消息条数进行累加显示。在一个具体的例子中，如图10所示，在预定时间窗口(如5S)内，通知内容中包括的消息条数由1变化至2 再变化为3。
综上，采用本说明书实施例披露的应用消息的通知方法，可以使得消息通知方式契合终端设备所处的环境，从而使用户在不遗漏任何一条消息的情况下，不被干扰，享有使用终端的流畅体验。
下面再结合一个具体的实施例，对本说明书披露的消息通知方法进行介绍。具体地，图11示出根据一个实施例的消息通知方法流程框图，所述方法的执行主体可以为终端设备。如图11所示，所述方法流程可以包括以下步骤S1101-步骤S1114。
步骤S1101，接收到应用消息的通知请求。
步骤S1102，检测终端设备是否处于解锁状态。在检测到处于未解锁状态的情况下，在步骤S1103，采用预置的锁屏消息通知方式进行消息通知。
在检测到处于解锁状态的情况下，一方面，执行步骤S1104-步骤S1106。
具体地，在步骤S1104，对预定时长(如2S)内终端设备中显示的界面进行连续截取，得到预定数量(如4张)的多张截屏图片。
步骤S1105，将多张截屏图片分别输入第一区域预测模型中，得到对应的多个第一显示区域。
步骤S1106，将多个第一显示区域作为区域序列输入第二区域预测模型中，得到目标显示区域，用于显示针对应用消息的通知内容。
另一方面，执行步骤S1107和步骤S1108。
具体地，在步骤S1107，获取预定类型的特征信息，其中预定类型包括以下中的一种或多种：所述终端设备的设备信息，所述显示的界面所属应用的应用信息，在预定历史时段内操作所述终端设备产生的操作行为数据。
在步骤S1108，将获取的特征信息输入预先训练的时长预测模型中，得到针对通知内容的停留时长。
在步骤S1106，以及步骤S1108之后，执行以下步骤S1109至步骤S1114。
步骤S1109，将目标显示区域和停留时长输入预先训练的样式预测模型中，得到针对通知内容的通知样式，其中通知样式属于以下中的一种：弹幕样式、消息条数简易样式、横幅样式、字幕样式和横屏对联样式。
步骤S1110，判断通知样式是否属于第一样式集合，第一样式集合中包括弹幕样式、消息条数简易样式和横幅样式。
进一步地，一方面，在属于第一样式集合的情况下，执行步骤S1111，将目标显示区域输入预先训练的多分类任务模型中，得到字体颜色、背景颜色和背景亮度。在步骤S1112，基于确定出的目标显示区域、通知样式、字体颜色、背景颜色和背景亮度展示针对应用消息的通知内容。
另一方面，在不属于第一样式集合的情况下，执行步骤S1113，将目标显示区域输入字体颜色预测模型中，得到字体颜色。在步骤S1114，基于确定出的目标显示区域、通知样式和字体颜色，并且，以无展示背景的方式，展示针对应用消息的通知内容。
综上，采用本说明书实施例披露的应用消息的通知方法，可以使得消息通知方式契合终端设备所处的环境，从而使用户在不遗漏任何一条消息的情况下，不被干扰，享有使用终端的流畅体验。
与上述方法相对应的，本说明书实施例还披露一种通知装置。具体地，图12示出根据一个实施例的应用消息的通知装置结构图，所述装置集成于终端设备。如图12所示，所述装置1200可以包括：截屏单元1210，配置为响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片。区域预测单元1220，配置为将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域；所述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域。目标区域确定单元1230，配置为基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
在一个实施例中，所述截屏单元1210具体配置为：响应于接收到所述通知请求，检测所述终端设备的屏幕是否处于解锁状态；在处于解锁状态的情况下，对所述显示的界面进行截取，得到所述若干截屏图片。
在一个实施例中，所述区域预测单元1220具体配置为：分别对所述若干截屏图片进行颜色特征和/或纹理特征的提取，得到对应的若干图片特征向量；将所述若干图片特征向量分别输入所述第一区域预测模型中，得到所述若干第一显示区域。
在一个实施例中，其中所述截屏单元1210具体配置为：对预定时长内显示的界面进行连续截取，得到多张截屏图片。其中所述区域预测单元1220具体配置为：将所述多张截屏图片分别输入所述第一区域预测模型中，得到对应的多个第一显示区域；其 中目标区域确定单元1230具体配置为：将所述多个第一显示区域作为区域序列，输入预先训练的第二区域预测模型中，得到所述目标显示区域。
进一步地，在一个实施例中，所述第二区域预测模型基于训练单元而预先训练，所述训练单元配置为：获取多个训练样本，其中每个训练样本中包括对应的历史区域序列和标注有历史目标显示区域的标注截屏图片；其中所述历史区域序列包括将多张历史截屏图片分别输入所述第一区域预测模型而得到的多个区域，所述多张历史截屏图片是在所述标注截屏图片的截取时刻之前的所述预定时长内截取的图片。利用所述多个训练样本，训练所述第二区域预测模型。
在一个实施例中，所述装置1200还包括：特征信息获取单元1240，配置为获取预定类型的特征信息，所述预定类型包括以下中的一种或多种：所述终端设备的设备信息，所述显示的界面所属应用的应用信息，在预定历史时段内操作所述终端设备产生的操作行为数据；时长预测单元1250，配置为将所述特征信息输入预先训练的时长预测模型中，得到针对所述应用消息的通知时长。
在一个实施例中，所述装置1200还包括：样式预测单元1260，配置为将所述目标显示区域和通知时长，共同输入预先训练的样式预测模型中，得到针对所述应用消息的通知样式。
在一个实施例中，所述通知样式属于以下中的一种：弹幕样式、消息条数简易样式、横幅样式、横屏字幕样式、横屏对联样式。
在一个实施例中，所述装置1200还包括：字体颜色预测单元1270，配置为将所述目标显示区域输入预先训练的字体颜色预测模型中，得到用于显示所述通知内容的字体颜色。
在一个实施例中，所述装置1200还包括：多任务预测单元1280，配置为将所述目标显示区域输入预先训练的多任务模型中，得到用于显示所述通知内容的字体颜色、背景颜色和背景透明度。
在一个实施例中，所述装置1200还包括：显示单元1290，配置为在所述目标显示区域中显示所述通知内容，用于通知用户已接收到所述应用消息。
综上，采用本说明书实施例披露的应用消息的通知装置，可以使得消息通知方式契合终端设备所处的环境，从而使用户在不遗漏任何一条消息的情况下，不被干扰， 享有使用终端的流畅体验。
如上，根据又一方面的实施例，还提供一种计算机可读存储介质，其上存储有计算机程序，当所述计算机程序在计算机中执行时，令计算机执行结合图3或图11所描述的方法。
根据又一方面的实施例，还提供一种计算设备，包括存储器和处理器，所述存储器中存储有可执行代码，所述处理器执行所述可执行代码时，实现结合图3或图11所描述的方法。
本领域技术人员应该可以意识到，在上述一个或多个示例中，本说明书披露的多个实施例所描述的功能可以用硬件、软件、固件或它们的任意组合来实现。当使用软件实现时，可以将这些功能存储在计算机可读介质中或者作为计算机可读介质上的一个或多个指令或代码进行传输。
以上所述的具体实施方式，对本说明书披露的多个实施例的目的、技术方案和有益效果进行了进一步详细说明，所应理解的是，以上所述仅为本说明书披露的多个实施例的具体实施方式而已，并不用于限定本说明书披露的多个实施例的保护范围，凡在本说明书披露的多个实施例的技术方案的基础之上，所做的任何修改、等同替换、改进等，均应包括在本说明书披露的多个实施例的保护范围之内。

Claims (24)

  1. 一种应用消息的通知方法，所述方法的执行主体为终端设备，所述方法包括：
    响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片；
    将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域；所述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域；
    基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
  2. 根据权利要求1所述的方法，其中，响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片，包括：
    响应于接收到所述通知请求，检测所述终端设备的屏幕是否处于解锁状态；
    在处于解锁状态的情况下，对所述显示的界面进行截取，得到所述若干截屏图片。
  3. 根据权利要求1所述的方法，其中，将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域，包括：
    分别对所述若干截屏图片进行颜色特征和/或纹理特征的提取，得到对应的若干图片特征向量；
    将所述若干图片特征向量分别输入所述第一区域预测模型中，得到所述若干第一显示区域。
  4. 根据权利要求1所述的方法，其中对所述终端设备中显示的界面进行截取，得到若干截屏图片，包括：
    对预定时长内显示的界面进行连续截取，得到多张截屏图片；
    其中将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域，包括：
    将所述多张截屏图片分别输入所述第一区域预测模型中，得到对应的多个第一显示区域；
    其中基于所述若干第一显示区域，确定目标显示区域，包括：
    将所述多个第一显示区域作为区域序列，输入预先训练的第二区域预测模型中，得到所述目标显示区域。
  5. 根据权利要求4中所述的方法，其中，所述第二区域预测模型基于以下步骤而 预先训练：
    获取多个训练样本，其中每个训练样本中包括对应的历史区域序列和标注有历史目标显示区域的标注截屏图片；其中所述历史区域序列包括将多张历史截屏图片分别输入所述第一区域预测模型而得到的多个区域，所述多张历史截屏图片是在所述标注截屏图片的截取时刻之前的所述预定时长内截取的图片；
    利用所述多个训练样本，训练所述第二区域预测模型。
  6. 根据权利要求1所述的方法，其中，在接收到应用消息的通知请求之后，所述方法还包括：
    获取预定类型的特征信息，所述预定类型包括以下中的一种或多种：所述终端设备的设备信息，所述显示的界面所属应用的应用信息，在预定历史时段内操作所述终端设备产生的操作行为数据；
    将所述特征信息输入预先训练的时长预测模型中，得到针对所述应用消息的通知时长。
  7. 根据权利要求6所述的方法，其中，在确定目标显示区域，以及，得到针对所述应用消息的通知时长之后，所述方法还包括：
    将所述目标显示区域和通知时长，共同输入预先训练的样式预测模型中，得到针对所述应用消息的通知样式。
  8. 根据权利要求7所述的方法，其中，所述通知样式属于以下中的一种：弹幕样式、消息条数简易样式、横幅样式、横屏字幕样式、横屏对联样式。
  9. 根据权利要求1、6-8中任一项所述的方法，其中，在确定目标显示区域之后，所述方法还包括：
    将所述目标显示区域输入预先训练的字体颜色预测模型中，得到用于显示所述通知内容的字体颜色。
  10. 根据权利要求1、6-8中任一项所述的方法，其中，在确定目标显示区域之后，所述方法还包括：
    将所述目标显示区域输入预先训练的多任务模型中，得到用于显示所述通知内容的字体颜色、背景颜色和背景透明度。
  11. 根据权利要求1所述的方法，其中，在确定目标显示区域之后，所述方法还包括：
    在所述目标显示区域中显示所述通知内容，用于通知用户已接收到所述应用消息。
  12. 一种应用消息的通知装置，所述装置集成于终端设备，所述装置包括：
    截屏单元，配置为响应于接收到应用消息的通知请求，对所述终端设备中显示的界面进行截取，得到若干截屏图片；
    区域预测单元，配置为将所述若干截屏图片分别输入第一区域预测模型中，得到对应的若干第一显示区域；所述第一区域预测模型基于多张带标注的样本截屏图片而预先训练，所述标注对应于样本截屏图片中不包含重要内容的区域；
    目标区域确定单元，配置为基于所述若干第一显示区域，确定目标显示区域，用于显示针对所述应用消息的通知内容。
  13. 根据权利要求12所述的装置，其中，所述截屏单元具体配置为：
    响应于接收到所述通知请求，检测所述终端设备的屏幕是否处于解锁状态；
    在处于解锁状态的情况下，对所述显示的界面进行截取，得到所述若干截屏图片。
  14. 根据权利要求12所述的装置，其中，所述区域预测单元具体配置为：
    分别对所述若干截屏图片进行颜色特征和/或纹理特征的提取，得到对应的若干图片特征向量；
    将所述若干图片特征向量分别输入所述第一区域预测模型中，得到所述若干第一显示区域。
  15. 根据权利要求12所述的装置，其中所述截屏单元具体配置为：
    对预定时长内显示的界面进行连续截取，得到多张截屏图片；
    其中所述区域预测单元具体配置为：
    将所述多张截屏图片分别输入所述第一区域预测模型中，得到对应的多个第一显示区域；
    其中目标区域确定单元具体配置为：
    将所述多个第一显示区域作为区域序列，输入预先训练的第二区域预测模型中，得到所述目标显示区域。
  16. 根据权利要求15中所述的装置，其中，所述第二区域预测模型基于训练单元而预先训练，所述训练单元配置为：
    获取多个训练样本，其中每个训练样本中包括对应的历史区域序列和标注有历史目标显示区域的标注截屏图片；其中所述历史区域序列包括将多张历史截屏图片分别输入所述第一区域预测模型而得到的多个区域，所述多张历史截屏图片是在所述标注截屏图片的截取时刻之前的所述预定时长内截取的图片；
    利用所述多个训练样本，训练所述第二区域预测模型。
  17. 根据权利要求12所述的装置，其中，所述装置还包括：
    特征信息获取单元，配置为获取预定类型的特征信息，所述预定类型包括以下中的一种或多种：所述终端设备的设备信息，所述显示的界面所属应用的应用信息，在预定历史时段内操作所述终端设备产生的操作行为数据；
    时长预测单元，配置为将所述特征信息输入预先训练的时长预测模型中，得到针对所述应用消息的通知时长。
  18. 根据权利要求17所述的装置，其中，所述装置还包括：
    样式预测单元，配置为将所述目标显示区域和通知时长，共同输入预先训练的样式预测模型中，得到针对所述应用消息的通知样式。
  19. 根据权利要求18所述的装置，其中，所述通知样式属于以下中的一种：弹幕样式、消息条数简易样式、横幅样式、横屏字幕样式、横屏对联样式。
  20. 根据权利要求12、17-19中任一项所述的装置，其中，所述装置还包括：
    字体颜色预测单元，配置为将所述目标显示区域输入预先训练的字体颜色预测模型中，得到用于显示所述通知内容的字体颜色。
  21. 根据权利要求12、17-19中任一项所述的装置，其中，所述装置还包括：
    多任务预测单元，配置为将所述目标显示区域输入预先训练的多任务模型中，得到用于显示所述通知内容的字体颜色、背景颜色和背景透明度。
  22. 根据权利要求12所述的装置，其中，所述装置还包括：
    显示单元，配置为在所述目标显示区域中显示所述通知内容，用于通知用户已接收到所述应用消息。
  23. 一种计算机可读存储介质，其上存储有计算机程序，其中，当所述计算机程序在计算机中执行时，令计算机执行权利要求1-11中任一项所述的方法。
  24. 一种计算设备，包括存储器和处理器，其中，所述存储器中存储有可执行代码，所述处理器执行所述可执行代码时，实现权利要求1-11中任一项所述的方法。
PCT/CN2020/103972 2019-11-07 2020-07-24 应用消息的通知方法及装置 WO2021088422A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911083930.7 2019-11-07
CN201911083930.7A CN110865753B (zh) 2019-11-07 2019-11-07 应用消息的通知方法及装置

Publications (1)

Publication Number Publication Date
WO2021088422A1 true WO2021088422A1 (zh) 2021-05-14

Family

ID=69653297

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/103972 WO2021088422A1 (zh) 2019-11-07 2020-07-24 应用消息的通知方法及装置

Country Status (2)

Country Link
CN (1) CN110865753B (zh)
WO (1) WO2021088422A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116048326A (zh) * 2022-07-04 2023-05-02 荣耀终端有限公司 一种消息显示方法及电子设备

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110865753B (zh) * 2019-11-07 2021-01-22 支付宝(杭州)信息技术有限公司 应用消息的通知方法及装置
CN112199005A (zh) * 2020-10-22 2021-01-08 Tcl通讯（宁波）有限公司 内容显示方法、装置、存储介质及移动终端
CN112364196A (zh) * 2020-11-30 2021-02-12 深圳市六度人和科技有限公司 通知信息数据分析方法、装置、电子设备及存储介质
CN116033207A (zh) * 2022-12-09 2023-04-28 北京奇艺世纪科技有限公司 视频标题的生成方法、装置、电子设备及可读存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105867729A (zh) * 2016-03-25 2016-08-17 努比亚技术有限公司 消息显示装置和方法
CN106155471A (zh) * 2015-04-22 2016-11-23 阿里巴巴集团控股有限公司 一种操作界面的显示方法、装置及电子设备
US20180025453A1 (en) * 2016-07-19 2018-01-25 Redmon Jeang LLC Mobile Legal Counsel System and Method
CN107734175A (zh) * 2017-10-25 2018-02-23 维沃移动通信有限公司 一种通知消息的提示方法及移动终端
CN110865753A (zh) * 2019-11-07 2020-03-06 支付宝(杭州)信息技术有限公司 应用消息的通知方法及装置

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103886025B (zh) * 2014-02-22 2017-10-20 小米科技有限责任公司 网页中图片的显示方法和装置
US10628775B2 (en) * 2015-08-07 2020-04-21 Sap Se Sankey diagram graphical user interface customization
CN106502490A (zh) * 2016-09-19 2017-03-15 深圳市金立通信设备有限公司 一种来电提醒方法及终端
US10963525B2 (en) * 2017-07-07 2021-03-30 Avnet, Inc. Artificial intelligence system for providing relevant content queries across unconnected websites via a conversational environment
EP3537317B1 (en) * 2018-03-09 2022-11-16 Tata Consultancy Services Limited System and method for determination of air entrapment in ladles
CN108769821B (zh) * 2018-05-25 2019-03-29 广州虎牙信息科技有限公司 游戏场景描述方法、装置、设备及存储介质
CN109508638A (zh) * 2018-10-11 2019-03-22 平安科技（深圳）有限公司 人脸情绪识别方法、装置、计算机设备及存储介质
CN110147185B (zh) * 2018-11-16 2021-02-26 腾讯科技（深圳）有限公司 消息提示方法、装置、电子装置及存储介质
CN110113636A (zh) * 2019-04-28 2019-08-09 维沃移动通信有限公司 弹幕显示方法、弹幕推送方法、终端设备及服务器
CN110349145B (zh) * 2019-07-09 2022-08-16 京东方科技集团股份有限公司 缺陷检测方法、装置、电子设备以及存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106155471A (zh) * 2015-04-22 2016-11-23 阿里巴巴集团控股有限公司 一种操作界面的显示方法、装置及电子设备
CN105867729A (zh) * 2016-03-25 2016-08-17 努比亚技术有限公司 消息显示装置和方法
US20180025453A1 (en) * 2016-07-19 2018-01-25 Redmon Jeang LLC Mobile Legal Counsel System and Method
CN107734175A (zh) * 2017-10-25 2018-02-23 维沃移动通信有限公司 一种通知消息的提示方法及移动终端
CN110865753A (zh) * 2019-11-07 2020-03-06 支付宝(杭州)信息技术有限公司 应用消息的通知方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116048326A (zh) * 2022-07-04 2023-05-02 荣耀终端有限公司 一种消息显示方法及电子设备
CN116048326B (zh) * 2022-07-04 2023-10-20 荣耀终端有限公司 一种消息显示方法及电子设备

Also Published As

Publication number Publication date
CN110865753B (zh) 2021-01-22
CN110865753A (zh) 2020-03-06

Similar Documents

Publication Publication Date Title
WO2021088422A1 (zh) 应用消息的通知方法及装置
US20210072889A1 (en) Systems and methods for representing data, media, and time using spatial levels of detail in 2d and 3d digital applications
US10499109B2 (en) Method and apparatus for providing combined barrage information
CN109005283B (zh) 显示通知消息的方法、装置、终端及存储介质
US10810698B2 (en) Information processing method and client
US10955985B2 (en) Optimizing an arrangement of content on a display of a user device based on user focus
US20150121301A1 (en) Information processing method and electronic device
US11894021B2 (en) Data processing method and system, storage medium, and computing device
WO2023071861A1 (zh) 数据可视化展示方法、装置、计算机设备和存储介质
JP6206202B2 (ja) 情報処理装置及び情報処理プログラム
CN112135041A (zh) 一种人脸特效的处理方法及装置、存储介质
JP6237135B2 (ja) 情報処理装置及び情報処理プログラム
CN111124111A (zh) 一种处理方法、电子设备
CN107239222A (zh) 一种触摸屏的操控方法及终端设备
CN111986229A (zh) 视频目标检测方法、装置及计算机系统
US20160321968A1 (en) Information processing method and electronic device
CN113268182A (zh) 应用图标的管理方法和电子设备
CN112783594A (zh) 一种消息显示方法、装置及电子设备
CN106021588B (zh) 一种视频鹰眼图呈现方法及装置
CN112148171B (zh) 一种界面切换方法、装置及电子设备
CN111796736B (zh) 应用程序的分享方法、装置和电子设备
CN110737417B (zh) 一种演示设备及其标注线的显示控制方法和装置
CN107872730A (zh) 一种视频中的嵌入内容的获取方法和装置
US11048713B2 (en) System and method for visual exploration of search results in two-mode networks
CN117331632A (zh) 消息提醒方法和装置

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: 20884206

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20884206

Country of ref document: EP

Kind code of ref document: A1