A photography style recommendation and video clip generation method based on age detection

By using age detection and personalized photography style recommendations, the system automatically captures and edits children's wonderful moments, solving the problems of time-consuming and laborious video editing and style mismatch in existing technologies, and achieving efficient video management.

CN118741241BActive Publication Date: 2024-11-22CHENGDU TOMMI TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411225488.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-03
Publication Date
2024-11-22
Estimated Expiration
2044-09-03

AI Technical Summary

Technical Problem

In existing technologies, it is difficult to accurately capture children's wonderful moments, video editing is time-consuming and laborious with low automation, and the shooting style is not compatible, resulting in difficulties in video management.

Method used

The age of the subject is obtained through age detection, and a personalized photography style is matched for video shooting. Target events are identified and key sections are divided. Time-lapse and slow-motion editing techniques are used to generate edited videos.

Benefits of technology

It enables automatic capture and efficient editing of children's wonderful moments, improves the automation of video editing and adaptability to different shooting styles, and reduces the difficulty of video management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118741241B_ABST
    Figure CN118741241B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of video shooting and processing, and proposes a photography style recommendation and video clip generation method based on age detection, the method comprising: determining a target event of a shooting object, matching the photography style of a target shooting task, executing the shooting of the target shooting task, dividing the shot video into key sections and non-key sections for editing, and generating a clip video. The present invention obtains reference information of the target shooting task, extracts the age value of the shooting object, matches the corresponding photography style for the target shooting task based on the age value, shoots the video of the shooting object based on the personalized photography style, and automatically generates a clip video with a unique style and wonderful moments for the shooting object, and can also realize timely sorting and pushing of the clip video, so as to solve the problems of difficulty in capturing wonderful moments, time-consuming and labor-intensive video editing, low degree of automation, and incompatibility of editing photography style in the existing technology for the collection and editing of children's videos.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of video shooting and processing, and in particular to a method for recommending photography styles and generating video clips based on age detection. Background Art

[0002] Currently, the recording of children's important moments is mainly through active filming by parents or caregivers, but the children's movements cannot be predicted in advance and are fleeting. Therefore, many wonderful moments of the children cannot be recorded in time, causing many regrets.

[0003] To make up for this regret, some caring parents may edit and store the monitoring data from surveillance cameras (cameras installed at home or in the living room where children play, with the main purpose of remotely checking the current status of children to prevent safety accidents). However, the image quality of surveillance cameras is very poor, and parents are also required to edit and store the data. This results in a lot of editing workload for parents. For one day's video, they may need to review 24 hours of video, find meaningful things in the 24 hours of video, and then edit it in a targeted manner. This difficulty makes it difficult for parents to do daily editing and storage.

[0004] In addition, the prior art also has AI robots for children to play with. The robots are equipped with cameras that can track and photograph identified target objects. However, compared with surveillance cameras, this method only improves the quality of photos or videos. Similarly, the video images have not undergone any processing. The video is only useful for the moment when the event occurs. For subsequent time, since the video has no additional labels, markings, etc., the pile of videos will be like dusty items placed in a warehouse, which are wanted but cannot be found.

[0005] Therefore, how to accurately capture the wonderful moments of children and improve the automatic editing and style adaptation capabilities of videos shot for children is a technical problem that needs to be solved urgently. Summary of the invention

[0006] In order to solve the above-mentioned problems of the prior art, the present invention provides a photography style recommendation and video editing generation method based on age detection, aiming to solve the problems existing in the prior art of shooting and editing children's videos, such as difficulty in capturing wonderful moments, time-consuming and labor-intensive video editing, low degree of automation, and inappropriate editing photography style.

[0007] The present invention provides a photography style recommendation and video clip generation method based on age detection, the method comprising:

[0008] Obtain reference information for the target shooting task and extract the age value of the subject;

[0009] Determine a target event of the photographed object according to the age value, and use the target event to match the photography style of the target photography task;

[0010] Performing shooting of a target shooting task based on the photography style to obtain a shot video;

[0011] Identifying a target event in the captured video, and dividing the captured video into a key segment and a non-key segment according to a segment where the target event occurs in the captured video;

[0012] The captured video is edited using the video contents of the key segment and the non-key segment to generate an edited video.

[0013] Optionally, the reference information includes a reference picture, and the reference picture includes a photographed object; the step of acquiring the reference information of the target photographing task and extracting the age value of the photographed object specifically includes:

[0014] Obtaining a reference picture containing a photographed object submitted by a user or captured by a photographing device;

[0015] Identify and extract a first human body image of the photographed object in the reference picture, and construct a first image feature set of the photographed object based on image features in the first human body image;

[0016] The first image feature set is input into a pre-trained age recognition model to perform age recognition to obtain the age value of the photographed object.

[0017] Optionally, the reference information includes a reference text, and the reference text records age calibration information of the photographed object; the step of obtaining the reference information of the target photographing task and extracting the age value of the photographed object specifically includes:

[0018] Obtaining a reference text input by a user that records age calibration information of a photographed subject;

[0019] Extracting the calibration time and the calibration age in the age calibration information, and calculating the age difference between the photographed subject and the calibration age by using the current time and the calibration time;

[0020] Based on the age difference, the age value of the photographed subject is determined.

[0021] Optionally, determining a target event of the photographed subject according to the age value specifically includes:

[0022] According to the age value, matching the target event whose age value is in the corresponding age stage in the age stage and target event relationship comparison table;

[0023] Wherein, the age stage and target event relationship comparison table is configured to record a plurality of first setting information pre-input by the user or extracted from the target event standard age range information;

[0024] Each piece of first setting information includes a group of age stages and at least one target event that have an associated relationship.

[0025] Optionally, the target event specifically includes at least one action of laughing, crying, raising head, kicking feet, showing teeth, crawling, standing, and walking.

[0026] Optionally, using the target event to match the photography style of the target photography task specifically includes:

[0027] Using the target event, matching the photography style corresponding to the target event in a photography style and target event relationship comparison table;

[0028] The photography style and target event relationship comparison table is configured to record a plurality of second setting information corresponding to the photography style selected by the user for each target event from the preset photography style set;

[0029] Each piece of second setting information includes a target event and at least one group of photography styles that have an associated relationship.

[0030] Optionally, the photography style specifically includes: the proportion of presentation time of close-up shots and panoramic shots of the parts corresponding to the target event, the presentation order of close-up shots and panoramic shots of the parts corresponding to the target event, the focus control method for the parts corresponding to the target event, the picture change method for the parts corresponding to the target event, and at least one of the following: a picture blur control method for the parts corresponding to the target event.

[0031] Optionally, performing the shooting of the target shooting task based on the photography style to obtain the shot video specifically includes:

[0032] Sending an initial shooting control signal to the tracking camera and / or the area camera to obtain a second human body image obtained by the tracking camera and / or the area camera performing initial shooting;

[0033] constructing a second image feature set of the photographed object according to the image features in the second human body image, inputting the second image feature set into a pre-trained human body part recognition model for part recognition, and generating image positions of different parts in the second human body image;

[0034] Based on the picture control logic of the photography style and the image positions of different parts in the second human body image, a style photography control signal for a target photography task is generated;

[0035] The style shooting control signal is sent to the tracking camera and / or the area camera, driving the tracking camera and / or the area camera to execute the video shooting action corresponding to the picture control logic of the photography style to obtain the shot video.

[0036] Optionally, identifying a target event in the captured video, and dividing the captured video into a key segment and a non-key segment according to a segment where the target event occurs in the captured video, specifically includes:

[0037] Extracting a third human body image of the subject in the captured video, and constructing a third image feature set of the subject according to image features in the third human body image;

[0038] Inputting the third image feature set into a pre-trained target event action recognition model to perform target event recognition, and generating a segment where the target event occurs in the captured video;

[0039] According to the occurrence section of the target event, the captured video is divided into a key section and a non-key section.

[0040] Optionally, using the video contents of the key segment and the non-key segment to edit the shot video to generate an edited video specifically includes:

[0041] Using a first editing method of time-lapse photography to compress the video playback time of the video content of the non-key section to obtain a first editing video;

[0042] The second editing means of slow playback is used to extend the video duration of the video content of the key section to obtain a second editing video;

[0043] The first clipped video and the second clipped video are spliced ​​together to obtain a clipped video of the target shooting task.

[0044] The beneficial effects of the present invention are as follows: a photography style recommendation and video editing generation method based on age detection is proposed, by obtaining reference information of the target shooting task, extracting the age value of the shooting object, matching the corresponding photography style for the target shooting task based on the age value, shooting the shooting object based on the personalized photography style, and automatically generating a clipped video with a unique style and wonderful moments for the shooting object, and realizing timely sorting and pushing of the clipped video, thereby solving the problems existing in the prior art of collecting and editing children's videos, such as difficulty in capturing wonderful moments, time-consuming and labor-intensive video editing, low degree of automation, and incompatibility of editing photography styles. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1This is a flow chart of the method for recommending photography styles and generating video clips based on age detection provided in this embodiment. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] like Figure 1 As shown, a photography style recommendation and video clip generation method based on age detection, the method comprising:

[0048] S1: Obtain reference information of the target shooting task and extract the age value of the shooting object;

[0049] S2: determining a target event of the photographed object according to the age value, and matching the photography style of the target photographing task by using the target event;

[0050] S3: performing shooting of a target shooting task based on the photography style to obtain a shooting video;

[0051] S4: identifying a target event in the captured video, and dividing the captured video into a key segment and a non-key segment according to a segment where the target event occurs in the captured video;

[0052] S5: Using the video contents of the key segment and the non-key segment, the shot video is edited to generate an edited video.

[0053] It should be noted that the existing technology for the collection and editing of children's videos has problems such as difficulty in capturing wonderful moments, time-consuming and laborious video editing, low degree of automation, and incompatibility of editing photography styles. In order to solve the above problems, this embodiment obtains reference information of the target shooting task, extracts the age value of the shooting object, and uses the age value as a basis to match the corresponding photography style for the target shooting task, thereby driving the tracking camera and / or area camera and other shooting equipment to automatically shoot the shooting object with a personalized photography style based on the target event, which can solve the problem of parents capturing wonderful moments of children of different ages during their growth process, and also provides a reference for the key video content for dividing the key sections in the shooting video during the automatic video editing process. At the same time, after obtaining the shooting video as the editing material, a clip video with a unique style and wonderful moments is automatically generated for the shooting object, which improves the adaptability of the video editing photography style, reduces the difficulty of video editing, and retains a high quality of the editing video on the basis of improving the degree of editing automation.

[0054] In a preferred embodiment, the reference information includes a reference picture, and the reference picture includes a photographed object; the step of obtaining the reference information of the target photographing task and extracting the age value of the photographed object specifically includes:

[0055] S111: Acquire a reference picture including a photographed object submitted by a user or captured by a photographing device;

[0056] S112: Identify and extract a first human body image of the photographed object in the reference image, and construct a first image feature set of the photographed object according to image features in the first human body image;

[0057] S113: Inputting the first image feature set into a pre-trained age recognition model to perform age recognition, and obtaining an age value of the photographed subject.

[0058] In this embodiment, when extracting the age value of the photographed object, a reference picture submitted by the user or collected by a photographing device (which may be a tracking camera and / or a regional camera that performs photographing of the photographed object, or a smart terminal with a photographing function held by the user) can be used to construct a first image feature set using the image features of the photographed object in the reference picture, and then the constructed first image feature set is input into a pre-trained age recognition model for age recognition. Among them, the pre-trained age recognition model can be trained by inputting a human image sample set with age annotations into a deep learning model (such as CNN, convolutional neural network), and the age recognition model is obtained after reaching the number of training times or training convergence. In this way, the age of the photographed object can be recognized by machine learning, with faster recognition speed and higher convenience.

[0059] In a preferred embodiment, the reference information includes a reference text, and the reference text records the age calibration information of the photographed object; the step of obtaining the reference information of the target photographing task and extracting the age value of the photographed object specifically includes:

[0060] S121: Obtaining a reference text input by a user that records age calibration information of a photographed subject;

[0061] S122: extracting the calibration time and the calibration age in the age calibration information, and calculating the age difference between the photographed subject and the calibration age by using the current time and the calibration time;

[0062] S123: Determine the age value of the photographed subject based on the age difference.

[0063] In this embodiment, when extracting the age value of the photographed object, the age difference between the photographed age and the calibrated age can be calculated by using the reference text input by the user and the age calibration information of the photographed object recorded in the reference text, and then the current age value of the photographed object can be deduced based on the age difference (for example, if the age calibration information is "January 1, 2020 - 1 year old" and the current time is "January 1, 2021", it can be deduced that the current age value of the photographed object is 2 years old). Therefore, by pre-storing the age calibration information and extracting the calibration time and calibration age from the age calibration information, the age information of the photographed object can be indirectly identified. Only one setting of the age calibration information is required to realize age recognition for each subsequent shooting task, which has high accuracy and convenience.

[0064] In a preferred embodiment, determining the target event of the photographed object according to the age value specifically includes:

[0065] S21: According to the age value, matching the target event whose age value is in the corresponding age stage in the age stage and target event relationship comparison table;

[0066] Wherein, the age stage and target event relationship comparison table is configured to record a plurality of first setting information pre-input by the user or extracted from the target event standard age range information;

[0067] Each piece of first setting information includes a group of age stages and at least one target event that have an associated relationship.

[0068] In practical applications, the target event specifically includes at least one of: laughing, crying, raising the head, kicking the feet, showing teeth, crawling, standing, and walking.

[0069] In this embodiment, after obtaining the age value of the photographed object, the age value is used to match the target event corresponding to the age value in the age stage and target event relationship comparison table, that is, to determine the desired action of the photographed object at the current age stage, and these desired actions are generated by a number of first setting information pre-entered by the user or extracted from the standard age range information of the target event. In one embodiment, when the user enters the wonderful moment capture setting page of the APP end, he enters the first setting information of "1-3 months old, head raised", and the system extracts a group of age stages (i.e. 1-3 months) and a target event (i.e. head raised) from the first setting information, and then puts the associated ones into the constructed age stage and target event relationship comparison table, and the age stage and target event relationship comparison table is transmitted to the shooting device via a wireless network. After that, if it is detected that the age information of the photographed object is within the range of 1-3 months, the target event of the photographed object in the target shooting task is determined as "head raised". In another embodiment, the APP automatically obtains the target event standard age range information pre-stored by the system, and the target event standard age range information can usually be extracted from the general situation of the target event corresponding to the age of the shooting object (for example: 6 months old, sitting; 8 months old, crawling; 10 months old, standing; 12 months old, walking; etc.), and several first setting information extracted from the target event standard age range information are put into the constructed age stage and target event relationship comparison table, and the age stage and target event relationship comparison table is transmitted to the shooting device via a wireless network. After that, the target event of the shooting object in the target shooting task can be determined by using the association relationship between the age stage and the target event stored in the target event standard age range information. In this way, through user input or calling the pre-stored comparison table, the shooting device is helped to determine the video content that the user wants to capture in the target shooting task, thereby improving the efficiency and accuracy of capturing the wonderful moments of the shooting object.

[0070] In a preferred embodiment, using the target event to match the photography style of the target photography task specifically includes:

[0071] S22: using the target event, matching the photography style corresponding to the target event in a photography style and target event relationship comparison table;

[0072] The photography style and target event relationship comparison table is configured to record a plurality of second setting information corresponding to the photography style selected by the user for each target event from the preset photography style set;

[0073] Each piece of second setting information includes a target event and at least one group of photography styles that have an associated relationship.

[0074] In actual applications, the photography style specifically includes: the proportion of presentation time of close-up shots and panoramic shots of the parts corresponding to the target event, the presentation order of close-up shots and panoramic shots of the parts corresponding to the target event, the focus control method for the parts corresponding to the target event, the picture change method for the parts corresponding to the target event, and at least one of the following: a blurring control method for the parts corresponding to the target event.

[0075] In this embodiment, after obtaining the target event of the shooting object in the target shooting task, the target event is used to match the photography style corresponding to the target event in the photography style and target event relationship comparison table, that is, to determine the desired photography style of the target event in the shooting video, and these desired photography styles are generated by the user selecting a number of second setting information corresponding to the photography style for each target event from the preset photography style set. In actual application, when the user enters the video clip setting page of the APP end, the system first calls the target event set pre-stored or generated according to the user's first setting information, and then selects the corresponding photography style for each target event based on the photography style set provided by the system for each target event in the target event set, based on the user's selection action (click, type or drag) (for example: the user selects the photography style of "focus gradually positioned to the mouth" for the target event "smile"; selects the photography style of "distant view gradually switching to close view" for the target event "raise head"; etc.). In this way, by the user selecting the photography style corresponding to the target event from the preset photography style set, the system is helped to determine the way the user wants to present the target event in the shooting video, thereby improving the uniqueness and selection flexibility of the video shooting style.

[0076] In a preferred embodiment, performing the shooting of the target shooting task based on the photography style to obtain the shot video specifically includes:

[0077] S31: sending an initial shooting control signal to the tracking camera and / or the regional camera to obtain a second human body image obtained by the tracking camera and / or the regional camera performing initial shooting;

[0078] S32: constructing a second image feature set of the photographed object according to the image features in the second human body image, inputting the second image feature set into a pre-trained human body part recognition model for part recognition, and generating image positions of different parts in the second human body image;

[0079] S33: generating a style shooting control signal for a target shooting task based on the picture control logic of the photography style and the image positions of different parts in the second human body image;

[0080] S34: Sending the style shooting control signal to the tracking camera and / or the area camera, driving the tracking camera and / or the area camera to execute the video shooting action corresponding to the picture control logic of the photography style, and obtaining the shot video.

[0081] In this embodiment, the area camera can be configured as an infant camera (a camera installed beside a baby's bed) or any other camera capable of performing area camera photography of the subject, and the tracking camera can be configured as a child camera (a robot tracking camera) or any other camera capable of automatically tracking and photographing the subject. In this way, video photography of the subject can be achieved at all times, avoiding the regret of missing the fleeting wonderful moments of the subject. After obtaining the photography style of the target shooting task, the shooting of the target shooting task can be performed based on the photography style. The shooting process includes two parts. First, an initial shooting control signal is sent to the shooting device to drive the shooting device to perform an initial shooting action, obtain a second human body image of the shooting object, use the second human body image to extract image features and construct a second image feature set, and then use the human body part recognition model to perform part recognition based on the second image feature set. According to the human body parts obtained by recognition, the positions of different parts in the second human body image in the image are generated; after that, the style shooting control signal of the target shooting task is generated using the picture control logic of the photography style and the positions of different parts in the image. The style shooting control signal is used to drive the shooting device to perform a video shooting action corresponding to the photography style (for example, focus movement, lens push-pull, etc. for the corresponding parts of the target event). Therefore, this embodiment realizes the continuous recognition and positioning of human body parts in the picture through the generation of two types of shooting control signals, and performs the shooting of the target shooting task according to the picture control logic of the photography style based on the positioning result, so as to realize the video shooting and acquisition of a unique shooting object, and provide rich materials for video editing.

[0082] In a preferred embodiment, identifying a target event in the captured video and dividing the captured video into a key segment and a non-key segment according to the segment where the target event occurs in the captured video specifically includes:

[0083] S41: extracting a third human body image of the subject in the captured video, and constructing a third image feature set of the subject according to image features in the third human body image;

[0084] S42: inputting the third image feature set into a pre-trained target event action recognition model to perform target event recognition, and generating a segment where the target event occurs in the captured video;

[0085] S43: Dividing the captured video into a key segment and a non-key segment according to the segment where the target event occurs.

[0086] Furthermore, the captured video is edited using the video contents of the key segment and the non-key segment to generate an edited video, which specifically includes:

[0087] S51: using a first editing method of time-lapse photography to compress the video playback time of the video content of the non-key section to obtain a first edited video;

[0088] S52: Extending the video duration of the video content of the key section by using a second editing method of slow playback to obtain a second edited video;

[0089] S53: splicing the first edited video and the second edited video to obtain an edited video of the target shooting task.

[0090] In this embodiment, when performing video editing, firstly, by extracting the third human body image of the shooting object in the shooting video, using the constructed third image feature set and the target event action recognition model, the occurrence segment of the target event of the shooting object in the shooting video is identified and located, and then the key segment and non-key segment of the shooting video are obtained. After that, based on the preset editing method, the video content of the non-key segment and the video content of the key segment with the corresponding photography style are edited and spliced ​​to generate a clipped video of the target video. Exemplarily, the preset editing method includes time-lapse photography and slow playback. After obtaining the video content of the key segment and the video content of the non-key segment, the video content of the key segment is played slowly (for example, the action video content of the expected event of the shooting object is stretched on the time axis for slow motion presentation), and the video content of the non-key segment is time-lapse photography (for example, the 3-hour sleeping content of the child is compressed into 2 minutes of video content for presentation), and finally the video is spliced. Therefore, through the selection of photography style and preset editing method, an edited video of the target shooting task is generated, which focuses on presenting the wonderful moments of the expected actions of the subject, and quickly plays and browses other content, providing users with edited video content that is both unique and effective in video content.

[0091] In a preferred embodiment, after the second editing means of slow playback is used to extend the video content of the key section, the second editing video is obtained, specifically including:

[0092] S521: extracting an action corresponding to a target event of each second clip video, and matching a music set pre-associated with the action in a music database based on the action;

[0093] S522: Obtain a target soundtrack randomly selected from the soundtrack set or selected according to a user's selection action in the soundtrack set, and add the target soundtrack to the second clip video.

[0094] In this embodiment, after the video content of the key section is edited to obtain the second edited video, the corresponding target music can be matched to the second edited video in the music database by extracting the action corresponding to the target event in the second edited video. The matching method can be that the system randomly selects from the music set associated with the corresponding action or selects according to the user's selected action (for example, when the target event is "laughing", the target music is selected as soothing music; when the target event is "kicking the foot", the target music is selected as funny music). In this way, based on the video content of the second edited video, the music that matches the video content is automatically added to obtain a more complete edited video.

[0095] In a preferred embodiment, adding the target soundtrack to the second clip video specifically includes:

[0096] S5221: extracting the action of the shooting object in each video frame of each second clip video, calculating the sum of displacement distances of key points in the action of two consecutive video frames, and taking the latter video frame of the two consecutive video frames with the largest sum of displacement distances as a trigger video frame;

[0097] S5222: extracting the music features of the target music that change along the music playing time, calculating the music feature change value between two consecutive music playing time sampling points, and taking the later one of the two consecutive music playing time sampling points with the largest music feature change value as the trigger time sampling point; wherein the music features include the music volume intensity value and / or the music pitch frequency value;

[0098] S4253: Add the target soundtrack to the second clip video, and match the trigger video frame with the trigger time sampling point by adjusting the adding position of the target soundtrack in the second clip video.

[0099] In this embodiment, when adding the target soundtrack to the second edited video, in order to improve the integration effect of the target soundtrack with the video content of the second edited video, the action changes of the subject in the second edited video (for example, the displacement distance of the key points of the body of the subject in two consecutive video frames) and the music feature changes of the target soundtrack (for example, the music volume intensity value and / or the music pitch frequency value of the target music at two consecutive music playback time sampling points) can be extracted respectively, and the position where the target soundtrack is added to the second edited video can be adjusted by matching the action changes with the music feature changes, so that the matching music node is played when the subject in the final edited video performs the corresponding action, thereby achieving the effect of the music and the target event card point, and improving the presentation effect of the final generated edited video.

[0100] In a preferred embodiment, the method further comprises:

[0101] S54: after obtaining the clipped video of the target shooting task, combining the shooting time of the first video frame in the shot video and the target event corresponding to the video content of the key section in the shot video by text, to generate a storage name for the clipped video;

[0102] S55: storing the clipped video into a storage device, and determining whether the storage device stores a historical clipped video with a storage name containing a target event corresponding to the clipped video;

[0103] S56: If not, add text content representing the first occurrence of the target event to the storage name of the clipped video.

[0104] In this embodiment, in order to improve the management efficiency of the generated clipped videos by the system, after the clipped videos of the target video are generated, the target event corresponding to the shooting time of the first video frame in the shot video and the video content of the key section in the shot video are combined by text, and the storage name of the clipped video is generated (for example, March 8, 2024, laugh). After the clipped videos are stored in the storage device, facing a large number of clipped videos, the user can quickly find the clipped videos corresponding to the target event at the corresponding time. Considering that parents pay special attention to the wonderful moment when the target event first appears in their children, when the clipped videos are stored in the storage device, it can also be determined whether there is a historical clipped video corresponding to the target event in the storage device. If not, it indicates that the target event appears for the first time. At this time, the text content representing the first appearance of the target event is added to the storage name of the clipped video (for example, March 8, 2024, the first laugh). Therefore, by reasonably naming and effectively managing the clipped videos, users are helped to perform retrieval when searching for videos.

[0105] In a feasible embodiment, the method further includes:

[0106] S61: extracting a plurality of clipped videos stored in a storage device within a preset time period;

[0107] S62: Editing video contents of key sections in the plurality of edited videos to generate periodic edited videos corresponding to a preset time period;

[0108] S63 pushes the periodic clipped video to the user terminal.

[0109] In this embodiment, after obtaining the clipped videos, a periodic clipped video push method is provided, that is, at the end of each preset time period, the video contents of key sections in several clipped videos within the preset time period are clipped and merged to generate a periodic clipped video and push it to the user terminal (for example, at 9 o'clock every night, the clipped videos captured and edited by the child in a day are pushed to the parents), helping parents to quickly know the wonderful moments of the child throughout the day in a shorter time.

[0110] In another feasible embodiment, the method further includes:

[0111] S71: extracting a plurality of clipped videos stored in the storage device, the clipped videos having text content representing the first occurrence of a target event in the storage name;

[0112] S72: Editing video contents of key sections in the plurality of edited videos to generate a first edited video of the target event;

[0113] S73: Pushing the first edited video of the target event to the user terminal.

[0114] In this embodiment, after obtaining the clipped video, a method for pushing the clipped video based on the target event is also provided, that is, by obtaining all the clipped videos that occur for the first time in the storage device for editing and streaming, the first clipped video of the target event is obtained and pushed to the user terminal, helping parents recall the precious first actions of their children during their growth process.

[0115] In the description of the embodiments of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "center", "top", "bottom", "top", "bottom", "inside", "outside", "inner side", "outer side" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Among them, "inside" refers to an internal or enclosed area or space. "Periphery" refers to the area surrounding a specific component or a specific area.

[0116] In the description of the embodiments of the present invention, the terms "first", "second", "third", and "fourth" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", and "fourth" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0117] In the description of the embodiments of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "install", "connect", "connect", and "assemble" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0118] In the description of the embodiments of the present invention, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0119] In the description of the embodiments of the present invention, it should be understood that "-" and "~" represent a range between two values, and the range includes the endpoints. For example: "AB" represents a range greater than or equal to A and less than or equal to B. "A~B" represents a range greater than or equal to A and less than or equal to B.

[0120] In the description of the embodiments of the present invention, the term "and / or" herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " herein generally indicates that the associated objects before and after are in an "or" relationship.

[0121] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A photography style recommendation and video clip generation method based on age detection, It is characterized in that The method comprises: Acquire reference information of the target shooting task and extract the age value of the shooting object; the reference information includes a reference picture, and the reference picture includes the shooting object; specifically includes: Obtain a reference picture containing a photographed object submitted by a user or collected by a photographing device; identify and extract a first human image of the photographed object in the reference picture, and construct a first image feature set of the photographed object based on image features in the first human image; input the first image feature set into a pre-trained age recognition model to perform age recognition, and obtain an age value of the photographed object; Determining a target event of the photographed object according to the age value, and matching the photographic style of the target photographing task by using the target event; specifically comprising: According to the age value, the target event whose age value is in the corresponding age stage is matched in the age stage and target event relationship comparison table; wherein the age stage and target event relationship comparison table is configured to record a plurality of first setting information pre-input by the user or extracted from the target event standard age range information; wherein each first setting information includes a group of age stages and at least one target event having an associated relationship; Using the target event, matching the photography style corresponding to the target event in a photography style and target event relationship comparison table; wherein the photography style and target event relationship comparison table is configured to record a plurality of second setting information corresponding to the photography style selected by the user for each target event from a preset photography style set; wherein each second setting information includes a target event and at least one group of photography styles having an associated relationship; Performing shooting of a target shooting task based on the photography style to obtain a shot video; Identifying a target event in the captured video, and dividing the captured video into a key segment and a non-key segment according to a segment where the target event occurs in the captured video; The captured video is edited using the video contents of the key segment and the non-key segment to generate an edited video.

2. The method for recommending photography styles and generating video clips based on age detection according to claim 1, It is characterized in that The reference information includes a reference text, and the reference text records the age calibration information of the photographed object; the step of obtaining the reference information of the target photographing task and extracting the age value of the photographed object specifically includes: Obtaining a reference text input by a user that records age calibration information of a photographed subject; Extracting the calibration time and the calibration age in the age calibration information, and calculating the age difference between the photographed subject and the calibration age by using the current time and the calibration time; Based on the age difference, the age value of the photographed subject is determined.

3. The method for recommending photography styles and generating video clips based on age detection according to claim 1, It is characterized in that The target event specifically includes at least one of: laughing, crying, raising head, kicking feet, showing teeth, crawling, standing, and walking.

4. The method for recommending photography styles and generating video clips based on age detection according to claim 1, It is characterized in that The photography style specifically includes: the proportion of presentation time of close-up shots and panoramic shots of the parts corresponding to the target event, the presentation order of close-up shots and panoramic shots of the parts corresponding to the target event, the focus control method for the parts corresponding to the target event, the picture change method for the parts corresponding to the target event, and at least one of the following: a blurring control method for the parts corresponding to the target event.

5. The method for recommending photography styles and generating video clips based on age detection according to claim 1, It is characterized in that Performing the target shooting task based on the photography style to obtain the shot video specifically includes: Sending an initial shooting control signal to the tracking camera and / or the area camera to obtain a second human body image obtained by the tracking camera and / or the area camera performing initial shooting; constructing a second image feature set of the photographed object according to the image features in the second human body image, inputting the second image feature set into a pre-trained human body part recognition model for part recognition, and generating image positions of different parts in the second human body image; Based on the picture control logic of the photography style and the image positions of different parts in the second human body image, a style photography control signal for a target photography task is generated; The style shooting control signal is sent to the tracking camera and / or the area camera, driving the tracking camera and / or the area camera to execute the video shooting action corresponding to the picture control logic of the photography style to obtain the shot video.

6. The method for recommending photography styles and generating video clips based on age detection according to claim 1, It is characterized in that Identifying a target event in the captured video, and dividing the captured video into a key segment and a non-key segment according to the segment where the target event occurs in the captured video, specifically including: Extracting a third human body image of the subject in the captured video, and constructing a third image feature set of the subject according to image features in the third human body image; Inputting the third image feature set into a pre-trained target event action recognition model to perform target event recognition, and generating a segment where the target event occurs in the captured video; According to the occurrence section of the target event, the captured video is divided into a key section and a non-key section.

7. The method for recommending photography styles and generating video clips based on age detection according to claim 1, It is characterized in that Using the video contents of the key segment and the non-key segment, editing the shot video to generate an edited video specifically includes: Using a first editing method of time-lapse photography to compress the video playback time of the video content of the non-key section to obtain a first editing video; The second editing means of slow playback is used to extend the video duration of the video content of the key section to obtain a second editing video; The first clipped video and the second clipped video are spliced ​​together to obtain a clipped video of the target shooting task.