CN109101964A - Determine the method, equipment and storage medium in head and the tail region in multimedia file - Google Patents

Determine the method, equipment and storage medium in head and the tail region in multimedia file Download PDF

Info

Publication number
CN109101964A
CN109101964A CN201810928792.7A CN201810928792A CN109101964A CN 109101964 A CN109101964 A CN 109101964A CN 201810928792 A CN201810928792 A CN 201810928792A CN 109101964 A CN109101964 A CN 109101964A
Authority
CN
China
Prior art keywords
multimedia file
head
tail region
location information
region
Prior art date
Legal status (The legal status 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 status listed.)
Granted
Application number
CN201810928792.7A
Other languages
Chinese (zh)
Other versions
CN109101964B (en
Inventor
左洪涛
刘阿海
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen Co Ltd
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 Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Priority to CN201810928792.7A priority Critical patent/CN109101964B/en
Publication of CN109101964A publication Critical patent/CN109101964A/en
Application granted granted Critical
Publication of CN109101964B publication Critical patent/CN109101964B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/22Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/40Scenes; Scene-specific elements in video content
    • G06V20/46Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames

Abstract

The embodiment of the present application provides method, equipment and the storage medium in head and the tail region in a kind of determining multimedia file.This method comprises: obtaining multimedia file;Multimedia file is pre-processed, the first location information in head and the tail region is obtained;It calls machine learning model to handle first location information, obtains the second location information in head and the tail region, the precision of the second position is higher than the precision of first position.In the embodiment of the present application, first multimedia file is pre-processed, obtain first position of the head and the tail region in multimedia file, first position is further processed by machine learning model later, obtain the higher second position of precision, and the higher second position of precision is determined as to the position in the head and the tail region of multimedia file, to improve the accuracy rate of the position in the head and the tail region of determining multimedia file.

Description

Determine the method, equipment and storage medium in head and the tail region in multimedia file
Technical field
The invention relates to machine learning techniques field, in particular to head and the tail region in a kind of determining multimedia file Method, equipment and storage medium.
Background technique
It can also include some other information in multimedia file other than formal content.It is electricity with multimedia file For collection of drama, before the formal content of collection of TV plays is played, it will usually play some head contents, such as the collection of TV plays Creator's information, the collection of choice specimens segment of a upper collection of drama etc..Terminal can skip above-mentioned head content, and directly play TV The formal content of collection of drama.
Head content is skipped in terminal realization, and the precondition for directly playing the formal content of collection of TV plays is to be accurately positioned Head and the tail region in collection of TV plays, namely starting position and the end position of formal content.In the related technology, server is to TV Broadcasting frame included by collection of drama carries out image recognition one by one, if in a certain broadcasting frame including the keywords such as " xth collection ", " beginning " When, then the broadcasting frame is determined as to the starting position of formal content.
In the related technology, determine that the accuracy rate of the position in the head and the tail region of multimedia file is lower.
Summary of the invention
The embodiment of the present application provides method, equipment and the storage medium in head and the tail region in a kind of determining multimedia file, It can be used for solving the problems, such as determining that the accuracy rate of the position in the head and the tail region of multimedia file is lower in the related technology.
On the one hand, the embodiment of the present application provides a kind of method in head and the tail region in determining multimedia file, this method packet It includes:
Obtain multimedia file;
The multimedia file is pre-processed, the first location information in head and the tail region, the head and the tail region packet are obtained The start region and/or end region of the multimedia file are included, the first location information in the head and the tail region is used to indicate institute State first position of the head and the tail region in the multimedia file;
It calls machine learning model to handle the first location information, obtains the second position in the head and the tail region Information, the machine learning model obtain neural metwork training using multiple groups training sample data, the multiple groups training Every group of training sample data in sample data include the first location information in the head and the tail region of sample multimedia file, Yi Jisuo The actual position information in the head and the tail region of sample multimedia file is stated, the second location information in the head and the tail region is used to indicate institute The second position of the head and the tail region in the multimedia file is stated, the precision of the second position is higher than the essence of the first position Degree.
On the other hand, the embodiment of the present application provides a kind of device in head and the tail region in determining multimedia file, the device Include:
File acquisition module, for obtaining multimedia file;
First processing module obtains the first position letter in head and the tail region for pre-processing to the multimedia file Breath, the head and the tail region include the start region and/or end region of the multimedia file, and first of the head and the tail region Confidence breath is used to indicate first position of the head and the tail region in the multimedia file;
Second processing module obtains described for calling machine learning model to handle the first location information The second location information in head and the tail region, the machine learning model are to be obtained using multiple groups training sample data to neural metwork training It arrives, every group of training sample data in the multiple groups training sample data include the of the head and the tail region of sample multimedia file The actual position information in the head and the tail region of one location information and the sample multimedia file, the second of the head and the tail region Location information is used to indicate the second position of the head and the tail region in the multimedia file, and the precision of the second position is high Precision in the first position.
In another aspect, the embodiment of the present application provides a kind of computer equipment, the computer equipment include processor and Memory, is stored at least one instruction, at least one section of program, code set or instruction set in the memory, and described at least one Item instruction, at least one section of program, the code set or the instruction set are loaded by the processor and are executed to realize above-mentioned side The method in head and the tail region in determination multimedia file described in face.
Another aspect, the embodiment of the present application provide a kind of computer readable storage medium, the computer-readable storage Be stored at least one instruction, at least one section of program, code set or instruction set in medium, at least one instruction, it is described extremely Few one section of program, the code set or instruction set are loaded as processor and are executed to realize determination multimedia described in above-mentioned aspect The method in head and the tail region in file.
Another aspect provides a kind of computer program product, when the computer program product is performed, is used to hold The method in head and the tail region in determination multimedia file described in the above-mentioned aspect of row.
Technical solution bring beneficial effect provided by the embodiments of the present application includes at least:
By first pre-processing to multimedia file, first position of the head and the tail region in multimedia file is obtained, it First position is further processed by machine learning model afterwards, obtains the higher second position of precision, and more by precision The high second position is determined as the position in the head and the tail region of multimedia file, and then improves the head and the tail region for determining multimedia file Position accuracy rate.
Detailed description of the invention
Fig. 1 is the schematic diagram for the implementation environment that the application one embodiment provides;
Fig. 2 is the schematic diagram for the implementation environment that another embodiment of the application provides;
Fig. 3 is the flow chart of the method in head and the tail region in the determination multimedia file of the application one embodiment offer;
Fig. 4 is a kind of schematic diagram for application scenarios that the application one embodiment provides;
Fig. 5 is the schematic diagram for another application scenarios that the application one embodiment provides;
Fig. 6 is a kind of block diagram for application scenarios that the application one embodiment provides;
Fig. 7 is the block diagram of the device in head and the tail region in the determination multimedia file of the application one embodiment offer;
Fig. 8 is the block diagram of the device in head and the tail region in the determination multimedia file of the application one embodiment offer;
Fig. 9 is the block diagram for the computer equipment that the application one embodiment provides.
Specific embodiment
To keep the purposes, technical schemes and advantages of the application clearer, below in conjunction with attached drawing to the application embodiment party Formula is described in further detail.
Fig. 1 is the schematic diagram of the implementation environment shown in the application one embodiment.The implementation environment includes at least one end End 11 and server 12.
Terminal 11 has the function of playing multimedia file, and the function of data interaction is carried out with server 12.Terminal 11 can be mobile phone, tablet computer, E-book reader, multimedia play equipment, wearable device, pocket computer on knee Equal electronic equipments.Optionally, specified application is installed in terminal 11, above-mentioned broadcasting multimedia text is realized by the application program The function of part, and the function with the progress data interaction of server 12.Above-mentioned specified application can be video class application journey Sequence, music class application program, k sing class application program, and the embodiment of the present application is not construed as limiting this.
Server 12 has the function that head and the tail region is determined in multimedia file, and carries out data friendship with terminal 11 Mutual function.Server 12 can be the corresponding background server of the specified application installed in terminal 11.Server 12 can To be a server, it is also possible to a server cluster, can also be a cloud computing service center.
Establishing between at least one terminal 11 and server 12 has communication connection.Above-mentioned communication connection can be based on wired network Network or wireless network are established.
For the application, another implements the schematic diagram of the implementation environment exemplified to Fig. 2.In the implementation environment, server 12 It is a server cluster, which includes media stream server 121, matchmaker provides server 122 and position determines service Device 123.Above-mentioned media stream server 121, matchmaker provide server 122 and position determines that server 123 can be separately provided, can also To integrate.
Media stream server 121 is used for storing multimedia, and media stream server 121 is also used to terminal 11 and position It sets determining server 123 and multimedia file is provided.
Matchmaker provide server 122 be used for storing multimedia relevant information, such as multimedia file title, play when Position in multimedia file of length, creator, head and the tail region, the address of media stream server 121 etc..Matchmaker provides server 122 are also used to provide the relevant information of above-mentioned multimedia file to terminal 11.
Position determines server 123 for realizing the function in head and the tail region is determined in multimedia file.Optionally, position Setting determining server 123 includes machine learning model and image analysis engine, by the two cooperative cooperating, and then is realized in more matchmakers The function in head and the tail region is determined in body file.Position determines that server 123 is also used to provide server 122 to matchmaker and provides multimedia The second location information in the head and the tail region of file.
Media stream server 121 and matchmaker provide server 122, position determines to establish have communication connection between server 123.Matchmaker Money server 122 and position, which determine also to establish between server 123, communication connection.Above-mentioned communication connection can be based on wired network Network or wireless network are established.
Technical solution provided by the embodiments of the present application obtains head and the tail region by first pre-processing to multimedia file First position in multimedia file is later further processed first position by machine learning model, obtains essence The higher second position is spent, and the higher second position of precision is determined as to the position in the head and the tail region of multimedia file, is compared In in the related technology only with image set otherwise come determine multimedia file head and the tail region position, the application implement The technical solution that example provides, can be improved the accuracy rate of the position in the head and the tail region of determining multimedia file.
Technical solution provided by the embodiments of the present application can be applied to the recording of the playing task, song of multimedia file In task, both application scenarios are explained respectively below.
(1) collection of TV plays is played
Server determines starting position and the end position of the formal content of collection of TV plays, subsequent when a certain user asks When seeking the broadcasting collection of TV plays, server judges whether the user account number used by a user is VIP (Very Important Person) account number sends skip instruction to the user terminal of the user, to indicate user if the user account number is VIP account number Terminal skips head and run-out when playing collection of TV plays, only plays the formal content of the collection of TV plays.
(2) recording song.
User's triggering terminal starts recording song, the accompaniment of the terminal plays song, and shows and skip in recording interface Prelude option, when this skips prelude option when the user clicks, the formal content for the song that terminal is determined according to server is opened Beginning position, and continue to play from the starting position, quick recording song is allowed users to, is finished without equal preludes Play song again afterwards.
Fig. 3 is the flow chart of the method that head and the tail region is determined in multimedia file shown in the application one embodiment. This method can be executed by server, which can be server 12 shown in FIG. 1.This method includes following several steps It is rapid:
Step 301, multimedia file is obtained.
Multimedia file refers to the file including audio-frequency information and/or video information.In the embodiment of the present application, multimedia File is divided according to content-form comprising collection of TV plays, film, song MV etc..Server can be in multimedia File is authored when personnel upload and obtains the multimedia file.More matchmakers can also be obtained when multimedia file is requested by a user Body file, the embodiment of the present application are not construed as limiting the acquisition opportunity of multimedia file.
Step 302, multimedia file is pre-processed, obtains the first location information in head and the tail region.
Head and the tail region includes the start region and/or end region of multimedia file.
The start region of multimedia file refer to be played to since multimedia file it is formal included by multimedia file Content starts the part between broadcasting.Since the time that multimedia file starts to play is fixed as 00:00, it is thus determined that multimedia The position of the start region of content that is to say that formal content included by determining multimedia file starts the position played, can be with It is a timestamp, is also possible to serial number of a certain broadcasting frame in multimedia file.For collection of TV plays or film, Start region is also referred to as head.For song, start region is also referred to as prelude.
The end region of multimedia file refers to that formal end of text included by multimedia file is played to multimedia text Part terminates the part between broadcasting.Since the time that multimedia file starts to play is fixed as the playing duration of multimedia file, It is thus determined that the position of the end region of multimedia content that is to say that formal end of text included by determining multimedia file is broadcast The position put can be a timestamp, be also possible to serial number of a certain broadcasting frame in multimedia file.For TV play For collection or film, end region is also referred to as run-out.
The first location information in head and the tail region is used to indicate first position of the head and the tail region in multimedia file.First The precision set is lower, can also be referred to as rough position, refers to the position for being greater than predetermined probabilities as the probability of physical location.In advance If probability can be set according to actual needs, the embodiment of the present application is not construed as limiting this.The first position in head and the tail region can wrap Multiple broadcasting frames are included, above-mentioned multiple broadcasting frames can be continuous, can also be discontinuous, the embodiment of the present application is not construed as limiting this.It is first The first location information in tail region domain can use the corresponding timestamp of above-mentioned multiple broadcasting frames to indicate, can also be using multimedia text The serial number of frame is played included by part to indicate.For example, the first location information in head and the tail region is used to indicate multimedia file The position that formal content starts is the 40th second to the 47th second.For another example the first location information in head and the tail region is used to indicate more matchmakers The position that the formal content of body file starts is the 26th frame to the 30th frame.
Optionally, it before pre-processing multimedia file, first needs to be decoded multimedia file, obtain more Data flow in media file.Optionally.Image analysis engine is installed in server, is realized by the image analysis engine to more The decoding of media file and subsequent pretreatment work.
Optionally, step 302 may include following several sub-steps:
Step 302a, extracts target fragment from multimedia file.
Step 302b, pre-processes target fragment, obtains first location information.
Target fragment includes the head and the tail region of multimedia file.The playing duration of target fragment is less than broadcasting for multimedia file Put duration.Since the position in multimedia file of start region and/or end region of multimedia file is relatively fixed, The target fragment that greater probability includes above-mentioned start region and/or end region can be selected in multimedia file, and Target fragment is pre-processed, to determine the first location information of start region and/or end region.Pass through above-mentioned side Formula can pre-process to avoid to entire multimedia file, to reduce workload, be lifted in multimedia file and determine head The efficiency in tail region domain.
Illustratively, the playing duration of multimedia file is 45 minutes, and target fragment is 10 minutes, can be multimedia File start play first 10 minutes or it is 10 minutes latter.
Step 303, it calls machine learning model to handle the first location information, obtains the second of head and the tail region Location information.
Machine learning model obtains neural metwork training using multiple groups training sample data.
Every group of training sample data in multiple groups training sample data include the of the head and the tail region of sample multimedia file The actual position information in the head and the tail region of one location information and sample multimedia file.
The acquisition modes of the first location information in the head and the tail region of sample multimedia file can refer to step 302.The mistake Journey is specific as follows: server pre-processes sample multimedia file, obtains the of the head and the tail region of sample multimedia file One location information.Optionally, server extracts target fragment from sample multimedia file, later to the target fragment extracted It is pre-processed, obtains first location information.
In one possible implementation, broadcasting of the server to the target fragment extracted from sample multimedia file Frame carries out image recognition, obtains playing the corresponding image recognition result of frame, and server determines the according to above-mentioned image recognition result One target plays frame and the second target plays frame, and first object is played frame and/or the second target broadcasting frame in the more matchmakers of sample Position in body file is determined as the first location information in the head and the tail region of sample multimedia file.Wherein, first object plays Frame is the broadcasting frame in image recognition result including designated key word, and it is that image recognition result is used to indicate that the second target, which plays frame, In the broadcasting frame that specified region includes specified word information.
In alternatively possible implementation, server calculates the target fragment institute extracted from sample multimedia file Including the corresponding similarity of each image group, by similarity be less than similarity thresholding image group in the multimedia file Position, be determined as the first location information in the head and the tail region of sample multimedia file.Wherein, each image group includes adjacent Two broadcasting frames.
What the actual position information in the head and the tail region of sample multimedia file can be obtained by the way of manually marking.? That is, related technical personnel determine the physical location in the head and the tail region of multimedia file when watching sample multimedia file, go forward side by side Row record.
Neural network can be convolutional neural networks (Convolutional Neural Network, CNN), artificial neuron Network (Artificial Neural Network, ANN), deep neural network (Deep Neural Networks, DNN) etc., The embodiment of the present application is not construed as limiting this.
The machine learning algorithm used when training machine learning model can be back-propagation algorithm (Back- Propagation, BP), faster region convolutional neural networks faster RCNN (fasterRegions with Convolutional Neural Network, faster RCNN) algorithm etc., the embodiment of the present application is not construed as limiting this.
Optionally, machine learning model includes: an input layer, at least one hidden layer and an output layer.Input layer The first location information in tail region domain headed by input data, the output of output layer is the result is that the corresponding second of the first location information Confidence breath.Determination process is as follows: the first location information in head and the tail region being input to the input layer of tagsort model, by feature The hidden layer of disaggregated model carries out feature extraction to features described above data, and the feature extracted is combined and is abstracted, finally The corresponding second location information of the first location information is exported by output layer.In addition, in the embodiment of the present application, to the tool of hidden layer Body structure is not construed as limiting.In general, the number of plies of neural network is more, and effect is better but the calculating time also can be longer, in reality In, in combination with required precision, the neural network of the appropriate number of plies is designed.
The second location information in head and the tail region is used to indicate the second position of the head and the tail region in multimedia file.Second The precision set is greater than the precision of first position, can also be referred to as exact position, refer to the maximum probability as physical location Position.The second position in head and the tail region includes a broadcasting frame.The second location information in head and the tail region can use above-mentioned broadcasting The corresponding timestamp of frame indicates, can also be indicated using the serial number of above-mentioned broadcasting frame.
In conclusion technical solution provided by the embodiments of the present application is obtained by first pre-processing to multimedia file First position is later further located by machine learning model in first position of the head and the tail region in multimedia file Reason, obtains the higher second position of precision, and the higher second position of precision is determined as the head and the tail region of multimedia file Position, and then improve the accuracy rate for determining the position in head and the tail region of multimedia file.
Carrying out pretreated mode to multimedia file may include image recognition, similarity calculation etc..Below to this two Kind mode is illustrated respectively.
1, image recognition mode.
In one possible implementation, target fragment is pre-processed, obtaining first location information may include Following two sub-steps:
Step 302b1 carries out image recognition to broadcasting frame included by target fragment, obtains playing the corresponding image knowledge of frame Other result.
Algorithm used by image recognition can be image recognition algorithm neural network based, the image based on wavelet moment Recognizer, image recognition algorithm based on fractal characteristic etc., the embodiment of the present application is not construed as limiting this.In addition, in this Shen Please be in embodiment, playing the corresponding image recognition result of frame may include text information included by the broadcasting frame.
Step 302b2 determines first location information according to the corresponding image recognition result of frame is played.
In the embodiment of the present application, server plays text information included by frame by identifying, to determine head and the tail area Rough position of the domain in multimedia file.
In one example, step 302b2 may be implemented are as follows: will include the broadcasting of designated key word in image recognition result Frame is determined as first object and plays frame;First object is played into position of the frame in multimedia file, is determined as first position letter Breath.
Designated key word is used to indicate the formal content that will play multimedia file, can be pre- by relevant staff First set.Illustratively, designated key word can be " xth collection ", " beginning ", " Start ", and designated key word may be " this Collected ", " end ", " End ".
When playing the corresponding image recognition result of frame includes the designated keys word such as " xth collection ", " beginning ", " Start ", then It is the start region of multimedia file that the broadcasting frame, which has greater probability, thus server by the broadcasting frame in multimedia file Position is determined as the rough position of start region;When play the corresponding image recognition result of frame include " this collection is complete ", " end ", When the designated keys word such as " End ", then it is the end region of multimedia file that the broadcasting frame, which has greater probability, therefore server should Position of the frame in multimedia file is played, the rough position of end region is determined as.
Optionally, first object is played the corresponding timestamp of frame as the first location information in head and the tail region by server, Optionally, first object is played the serial number of frame as the first location information in head and the tail region by server.
In another example, step 302b2 may be implemented are as follows: according to the corresponding image recognition result of each broadcasting frame, Determine that the second target plays frame;Second target is played into position of the frame in multimedia file, is determined as first location information.
The specified region that the corresponding image recognition result of second target broadcasting frame is used to indicate the second target broadcasting frame includes Specified word information.
During multimedia file is played, for convenience of user to watch, lead on the corresponding picture of multimedia file It often is superimposed with text information, above-mentioned text information can be the lines of virtual portrait, lyrics etc..In the embodiment of the present application, According to the text information of Overlapping display on the corresponding picture of multimedia file, to determine the position in head and the tail region.
Specified region refers to the region for Overlapping display text information, is usually located at the lower half portion of picture, to keep away Exempt from that picture is caused to block.When specified word information can be collection of TV plays or played film, first lines of display, It can be first lyrics in song.
Optionally, the second target is played the corresponding timestamp of frame as the first location information in head and the tail region by server, Optionally, the second target is played the serial number of frame as the first location information in head and the tail region by server.
2, similarity calculation mode
In alternatively possible implementation, target fragment is pre-processed, obtaining first location information can wrap Include following two sub-steps:
Step 302b3 calculates the corresponding similarity of each image group included by target fragment.
Each image group includes two adjacent broadcasting frames.Multimedia file is during played at present, it will usually Prompt formal content that will be played by the picture of specific background, alternatively, formal content will stop playing.It is exemplary Ground, the formal content of multimedia file would generally show the picture of black background before being played.The picture of above-mentioned specific background Usually there is bigger difference in picture corresponding to the formal content with multimedia file, therefore in the embodiment of the present application, pass through It calculates adjacent two and plays the similarity between frame to determine the picture of above-mentioned specific background, and then determine multimedia file Head and the tail region.
Algorithm used by similarity calculation can be similarity calculation algorithm based on cosine similarity, based on it is European away from From similarity calculation algorithm, the similarity calculation algorithm based on Pearson came correlation etc., the embodiment of the present application does not make this It limits.
Similarity is less than position of the image group in multimedia file of similarity thresholding by step 302b4, is determined as the One location information.
Similarity thresholding can be set based on practical experience, and the embodiment of the present application is not construed as limiting this.
Optionally, similarity is less than the corresponding timestamp of image group of similarity thresholding as head and the tail region by server First location information, optionally, serial number of the server using similarity less than the image group of similarity thresholding is as head and the tail region First location information.
After the second location information for determining head and the tail region, server can be by the second position in above-mentioned head and the tail region Information is sent to terminal, so that terminal is capable of providing the service of the only formal content of broadcasting multimedia file.Optionally, in step After 303, it should determine that the method in head and the tail region further included following steps in multimedia file:
Step 401, the playing request that terminal is sent is received.
Playing request plays multimedia file for requesting.Playing request carries the mark and multimedia file of terminal Mark.The terminal is also carried when terminal is equipped with specified application, in playing request in above-mentioned specified application Used in user account number.
Optionally, terminal shows the mark (such as title, preview graph etc.) of multimedia file, when user triggers more matchmakers When the mark of body file, terminal to server sends playing request.
Step 402, whether detection terminal, which has, is skipped permission.
It skips permission and refers to the permission for skipping head and the tail region when playing multimedia file.If terminal, which has, skips permission, The beginning part (such as head, prelude etc.) and ending (example of multimedia file can be skipped when then playing multimedia file Such as run-out), and directly play the formal content of multimedia file.If terminal, which does not have, skips permission, multimedia file is played When need the beginning part to multimedia file to finish, just start the formal content for playing multimedia file, in addition, when more When the formal content of media file finishes, terminal needs the ending for continuing to play multimedia file.
Optionally, server is preserved with the terminal list for skipping permission.Server inquires above-mentioned list, if in list Including the terminal, it is determined that the terminal, which has, skips permission, if in list not including the terminal, it is determined that the terminal, which does not have, jumps Cross permission.If playing request also carries terminal user account number used in above-mentioned specified application, server can also There is the user account number list for skipping permission to save, if in list including the user account number, it is determined that the user account number has Permission is skipped, if in list not including the user account number, it is determined that the user account number, which does not have, skips permission.
Step 403, if it is detected that terminal, which has, skips permission, the second location information in head and the tail region is returned to terminal.
Terminal receives the second location information in head and the tail region, and the position according to indicated by the second location information, to jump Cross the beginning part and the ending of multimedia file.Illustratively, when the second location information of start region be used to indicate it is more In 25 frame, then terminal plays multimedia file since the 26th frame for the starting position of the formal content of media file, works as knot The second location information in beam region is used to indicate the end position of multimedia file in 2018 frame, then terminal is playing extremely Stop playing multimedia file after 2017 frames.
Fig. 4 is the interface schematic diagram for the application scenarios that the application one embodiment provides.Video 1 is watched in user A expectation, and The mark of video 1 is clicked, according to that should send playing request to server corresponding to the click signal of video 1, server exists terminal Detect that with when skipping permission, the head and the tail region of video 1 and video 1 is returned to terminal for user account number used in user A Second location information, terminal skips the start region of video 1 later, and plays since the formal content of video 1.
If it is detected that terminal, which does not have, skips permission, do not execute to the second location information in terminal return head and the tail region Step.
In addition, server can also first determine the second confidence in head and the tail region in other possible implementations When breath, the second location information in above-mentioned head and the tail region is directly sent to terminal later.Terminal is just starting to play multimedia file When show the first inquiry message, which is used to ask whether to skip the beginning part of multimedia file, works as terminal When receiving the confirmation instruction corresponding to the first inquiry message, terminal directly plays the formal content of multimedia file.Terminal exists After the formal content of multimedia file finishes, the second inquiry message is shown, second inquiry message is for asking whether to jump The ending for crossing multimedia file, when terminal receives the confirmation instruction corresponding to the second inquiry message, terminal terminates to broadcast Put multimedia file.
Fig. 5 is the interface schematic diagram for the application scenarios that another embodiment of the application provides.Song 1 is recorded in user B expectation, And the mark of song 1 is clicked, terminal sends playing request, server to server according to the click signal that should correspond to song 1 The second location information in the head and the tail region of song 1 and song 1 is returned to terminal, terminal shows inquiry message, the inquiry later Information is used to ask whether to skip the prelude (namely start region) of song 1, and terminal is receiving the inquiry letter corresponding to song 1 After breath, the start region of song 1 is skipped, and play since the formal content of song 1.
In one specifically example, in conjunction with reference Fig. 6, it illustrates the application scenarios shown in the application one embodiment Schematic diagram.Position determines that server reads multimedia file from media stream server, solves later to multimedia file Code, obtain data flow, position determines that the image analysis engine in server handles data stream, obtain characteristic information ( That is the first location information in head and the tail region), and by features described above information input machine learning model, it is inputted by machine learning model Statistic analysis result (namely the second location information in head and the tail region), finally by above-mentioned image analysis engine by the of head and the tail region Two location informations are stored to database.Subsequent, user provides server to matchmaker by the application program in terminal and requests multimedia text When part, matchmaker provides server and judges whether the user account number used by a user skips permission, if skipping permission, matchmaker provides clothes Business device returns to the second location information in head and the tail region and the address of media stream server, broadcasting in subsequent terminal to application program It puts device and obtains multimedia file from media stream server, and play out.In playing process, which can skip multimedia The head and the tail region of file only plays the formal content of multimedia file.
Before calling deep learning model to handle the first location information in head and the tail region, need to train in advance Machine learning model below explains the training process of machine learning model.One provided based on Fig. 2 embodiment can It selects in embodiment, before step 303, should determine that the method in head and the tail region further included following steps in multimedia file:
Step 501, call original machine learning model to the first location information in the head and the tail region of sample multimedia file It is handled, obtains the corresponding output result of training sample data.
Original machine learning model that is to say unbred machine learning model, the parameter in original machine learning model Usually set at random.Optionally, above-mentioned original machine learning model can be convolutional neural networks (such as alexNet net Network, VGG-16 network, GoogleNet network, Deep Residual Learning (study of depth residual error) network etc., this Shen Please embodiment this is not construed as limiting.
The related introduction of training sample data can refer to step 303, not repeat herein.
The second location information in the head and the tail region of the corresponding output result of training sample data namely sample multimedia file. In the embodiment of the present application, the first location information in the head and the tail region of multimedia file is inputted into original machine learning model, by Original machine learning model exports the corresponding output result of training sample data.
Step 502, by the reality of the corresponding output result of training sample data and the head and the tail region of sample multimedia file Location information compares, and obtains calculating loss.
Calculate the reality that loss is the head and the tail region of the corresponding output result of training sample data and sample multimedia file Difference between location information.
Step 503, detection calculates whether loss is greater than loss thresholding.
Losing thresholding can be according to the required precision actual set to machine learning model.If required precision is higher, damage It is with losing threshold sets smaller;It is with losing threshold sets larger if required precision is lower.
If calculating loss is greater than loss thresholding, the parameter in original machine learning model is adjusted according to loss is calculated It is whole, and above-mentioned steps 501 to 503 are repeated, namely from calling original machine learning model to the head of sample multimedia file The step of first location information in tail region domain is handled, and training sample data corresponding output result is obtained restarts to hold Row.
If calculating loss no more than loss thresholding, namely calculate loss and be less than or equal to loss thresholding, then will calculate and lose Original machine learning model when less than loss thresholding is as the original machine learning model for completing training, namely generates engineering Practise model.
Following is the application Installation practice, can be used for executing the application embodiment of the method.It is real for the application device Undisclosed details in example is applied, the application embodiment of the method is please referred to.
The block diagram of the device in head and the tail region in the determination multimedia file that Fig. 7 provides for the application one embodiment.The dress Setting has the function of realizing in above method example, and the function can also be executed corresponding by hardware realization by hardware Software realization.The apparatus may include: file acquisition module 701, first processing module 702 and Second processing module 703.
File acquisition module 701, for obtaining multimedia file.
First processing module 702 obtains the first position in head and the tail region for pre-processing to the multimedia file Information, the head and the tail region include the start region and/or end region of the multimedia file, and the first of the head and the tail region Location information is used to indicate first position of the head and the tail region in the multimedia file.
Second processing module 703 obtains institute for calling machine learning model to handle the first location information The second location information in head and the tail region is stated, the machine learning model is using multiple groups training sample data to neural metwork training It obtains, every group of training sample data in the multiple groups training sample data include the head and the tail region of sample multimedia file The actual position information in the head and the tail region of first location information and the sample multimedia file, the of the head and the tail region Two location informations are used to indicate the second position of the head and the tail region in the multimedia file, the precision of the second position Higher than the precision of the first position.
In conclusion technical solution provided by the embodiments of the present application is obtained by first pre-processing to multimedia file First position is later further located by machine learning model in first position of the head and the tail region in multimedia file Reason, obtains the higher second position of precision, and the higher second position of precision is determined as the head and the tail region of multimedia file Position, and then improve the accuracy rate for determining the position in head and the tail region of multimedia file.
In the alternative embodiment provided based on embodiment illustrated in fig. 7, the first processing module 701 is used for:
Extract target fragment from the multimedia file, the target fragment includes the head and the tail region, and the mesh The playing duration of standard film section is less than the playing duration of the multimedia file;
The target fragment is pre-processed, the first location information is obtained.
Optionally, the first processing module 701, is used for:
Target fragment, the target fragment packet are extracted from the multimedia file
Image recognition is carried out to broadcasting frame included by the target fragment, obtains the corresponding image recognition of the broadcasting frame As a result;
According to the corresponding image recognition result of the broadcasting frame, the first location information is determined.
Optionally, the first processing module 701, is used for:
It will include the broadcasting frame of designated key word in image recognition result, be determined as first object and play frame;
First object is played into position of the frame in the multimedia file, is determined as the first location information.
Optionally, the first processing module 701, is used for:
According to the corresponding image recognition result of each broadcasting frame, determine that the second target plays frame, second target Playing the corresponding image recognition result of frame and being used to indicate second target to play the specified region of frame includes specified word information;
Second target is played into position of the frame in the multimedia file, is determined as the first location information.
Optionally, the first processing module 701, is used for:
The corresponding similarity of each image group included by the target fragment is calculated, each image group includes adjacent Two broadcasting frames;
Similarity is less than position of the image group of similarity thresholding in the multimedia file, is determined as described first Location information.
In the alternative embodiment provided based on embodiment illustrated in fig. 7, referring to FIG. 8, described device further include: ask Ask receiving module 704, permission detection module 705 and position sending module 707.
Request receiving module 704, for receiving the playing request of terminal transmission, the playing request plays institute for requesting State multimedia file.
Whether permission detection module 705 has for detecting the terminal and skips permission, it is described skip permission and refer to broadcasting The permission in the head and the tail region is skipped when putting the multimedia file.
Position sending module 707, for if it is detected that the terminal skips permission with described, to terminal return The second location information in the head and the tail region.
In the alternative embodiment provided based on embodiment illustrated in fig. 7, referring to FIG. 8, described device further include: mould Type training module 707.
Model training module 707, is used for:
Original machine learning model is called to carry out the first location information in the head and the tail region of the sample multimedia file Processing, obtains the corresponding output result of the training sample data;
By the reality of the corresponding output result of the training sample data and the head and the tail region of the sample multimedia file Location information compares, and obtains calculating loss;
Detect whether the calculating loss is greater than loss thresholding;
If the calculating loss is greater than the loss thresholding, lose according to calculating in the original machine learning model Parameter be adjusted, and from the calling original machine learning model to the of the head and the tail region of the sample multimedia file The step of one location information is handled, and the training sample data corresponding output result is obtained restarts to execute;
If the calculating loss is not more than the loss thresholding, the machine learning model is generated.
Fig. 9 is the structural schematic diagram of computer equipment provided by one embodiment of the present invention.The computer equipment can be Server.The method in head and the tail region in the determination multimedia file that the computer equipment is used to implement to provide in above-described embodiment. Specifically:
Computer equipment 900 is including central processing unit (CPU) 901 including random access memory (RAM) 902 and only Read the system storage 904 of memory (ROM) 903, and the system of connection system storage 904 and central processing unit 901 Bus 905.Computer equipment 900 further includes that the basic input/output of information is transmitted between each device helped in computer System (I/O system) 906, and large capacity for storage program area 913, application program 914 and other program modules 915 are deposited Store up equipment 907.
Basic input/output 906 includes display 908 for showing information and inputs information for user The input equipment 909 of such as mouse, keyboard etc.Wherein display 908 and input equipment 909 are all by being connected to system bus 905 input and output controller 910 is connected to central processing unit 901.Basic input/output 906 can also include defeated Enter o controller 910 for receiving and handling from the defeated of multiple other equipment such as keyboard, mouse or electronic touch pen Enter.Similarly, input and output controller 910 also provides output to display screen, printer or other kinds of output equipment.
Mass-memory unit 907 is connected by being connected to the bulk memory controller (not shown) of system bus 905 To central processing unit 901.Mass-memory unit 907 and its associated computer-readable medium are computer equipment 900 Non-volatile memories are provided.That is, mass-memory unit 907 may include such as hard disk or CD-ROM drive it The computer-readable medium (not shown) of class.
Without loss of generality, computer-readable medium may include computer storage media and communication media.Computer storage Medium includes any of the information such as computer readable instructions, data structure, program module or other data for storage The volatile and non-volatile of method or technique realization, removable and irremovable medium.Computer storage medium include RAM, ROM, EPROM, EEPROM, flash memory or other solid-state storages its technologies, CD-ROM, DVD or other optical storages, cassette, magnetic Band, disk storage or other magnetic storage devices.Certainly, skilled person will appreciate that computer storage medium is not limited to It states several.Above-mentioned system storage 904 and mass-memory unit 907 may be collectively referred to as memory.
According to various embodiments of the present invention, computer equipment 900 can also be arrived by network connections such as internets Remote computer operation on network.Namely computer equipment 900 can be by the network interface that is connected on system bus 905 Unit 911 is connected to network 912, in other words, Network Interface Unit 911 also can be used be connected to other kinds of network or Remote computer system (not shown).
Memory further includes one, and perhaps more than one program one or more than one program are stored in memory In, and be configured to be executed by one or more than one processor.Said one or more than one program include for holding The instruction in head and the tail region in the above-mentioned determining multimedia file of row.
In the exemplary embodiment, a kind of computer readable storage medium is additionally provided, is stored in the storage medium At least one instruction, at least one section of program, code set or instruction set, at least one instruction, at least one section of program, institute Code set or instruction set is stated to be loaded by the processor of electronic equipment and executed to realize the more matchmakers of the determination in above method embodiment The method in head and the tail region in body file.
Optionally, above-mentioned computer readable storage medium can be ROM, random access memory (RAM), CD-ROM, magnetic Band, floppy disk and optical data storage devices etc..
Another aspect provides a kind of computer program product, when the computer program product is performed, is used to hold The method in head and the tail region in determination multimedia file described in the above-mentioned aspect of row.
It should be understood that referenced herein " multiple " refer to two or more."and/or", description association The incidence relation of object indicates may exist three kinds of relationships, for example, A and/or B, can indicate: individualism A exists simultaneously A And B, individualism B these three situations.Character "/" typicallys represent the relationship that forward-backward correlation object is a kind of "or".Make herein " first ", " second " and similar word are not offered as any sequence, quantity or importance, and are used only to distinguish Different component parts.
Above-mentioned the embodiment of the present application serial number is for illustration only, does not represent the advantages or disadvantages of the embodiments.
The above is only the exemplary embodiments of the application, all in spirit herein and original not to limit the application Within then, any modification, equivalent replacement, improvement and so on be should be included within the scope of protection of this application.

Claims (11)

1. a kind of method in head and the tail region in determining multimedia file, which is characterized in that the described method includes:
Obtain multimedia file;
The multimedia file is pre-processed, the first location information in head and the tail region is obtained, the head and the tail region includes institute The start region and/or end region of multimedia file are stated, the first location information in the head and the tail region is used to indicate the head First position of the tail region domain in the multimedia file;
It calls machine learning model to handle the first location information, obtains the second confidence in the head and the tail region Breath, the machine learning model obtain neural metwork training using multiple groups training sample data, the multiple groups training sample Every group of training sample data in notebook data include the first location information in the head and the tail region of sample multimedia file and described The actual position information in the head and the tail region of sample multimedia file, the second location information in the head and the tail region are used to indicate described The second position of the head and the tail region in the multimedia file, the precision of the second position are higher than the essence of the first position Degree.
2. being obtained the method according to claim 1, wherein described pre-process the multimedia file The first location information in head and the tail region, comprising:
Extract target fragment from the multimedia file, the target fragment includes the head and the tail region, and the target patch The playing duration of section is less than the playing duration of the multimedia file;
The target fragment is pre-processed, the first location information is obtained.
3. according to the method described in claim 2, it is characterized in that, described to audio, video data included by the target fragment It is pre-processed, obtains the first location information, comprising:
Image recognition is carried out to broadcasting frame included by the target fragment, obtains the corresponding image recognition knot of the broadcasting frame Fruit;
According to the corresponding image recognition result of the broadcasting frame, the first location information is determined.
4. according to the method described in claim 3, it is characterized in that, described according to the corresponding image recognition knot of the broadcasting frame Fruit determines the first location information, comprising:
It will include the broadcasting frame of designated key word in image recognition result, be determined as first object and play frame;
First object is played into position of the frame in the multimedia file, is determined as the first location information.
5. the method according to included by claim 3, which is characterized in that described according to the corresponding image recognition knot of the broadcasting frame Fruit determines the first location information, comprising:
According to the corresponding image recognition result of each broadcasting frame, determine that the second target plays frame, second target plays It includes specified word information that the corresponding image recognition result of frame, which is used to indicate second target and plays the specified region of frame,;
Second target is played into position of the frame in the multimedia file, is determined as the first location information.
6. according to the method described in claim 2, obtaining institute it is characterized in that, described pre-process the target fragment State first location information, comprising:
Calculate the corresponding similarity of each image group included by the target fragment, each image group includes adjacent two A broadcasting frame;
Similarity is less than position of the image group of similarity thresholding in the multimedia file, is determined as the first position Information.
7. method according to any one of claims 1 to 6, which is characterized in that the calling machine learning model is to described First location information is handled, before obtaining the second location information in the head and the tail region, further includes:
Original machine learning model is called to handle the first location information in the head and the tail region of the sample multimedia file, Obtain the corresponding output result of the training sample data;
By the physical location of the corresponding output result of the training sample data and the head and the tail region of the sample multimedia file Information comparison obtains calculating loss;
Detect whether the calculating loss is greater than loss thresholding;
If the calculating loss is greater than the loss thresholding, lose according to calculating to the ginseng in the original machine learning model Number is adjusted, and from the calling original machine learning model to first of the head and the tail region of the sample multimedia file The step of confidence breath is handled, obtains the training sample data corresponding output result restarts to execute;
If the calculating loss is not more than the loss thresholding, the machine learning model is generated.
8. method according to any one of claims 1 to 6, which is characterized in that the calling machine learning model is to described First location information is handled, after obtaining the second location information in the head and the tail region, further includes:
The playing request that terminal is sent is received, the playing request plays the multimedia file for requesting;
Detect the terminal and whether have and skip permission, it is described skip permission and refer to skip institute when playing the multimedia file State the permission in head and the tail region;
If it is detected that the terminal skips permission with described, the second confidence in the head and the tail region is returned to the terminal Breath.
9. the device in head and the tail region in a kind of determining multimedia file, which is characterized in that described device includes:
File acquisition module, for obtaining multimedia file;
First processing module obtains the first location information in head and the tail region, institute for pre-processing to the multimedia file State the start region and/or end region that head and the tail region includes the multimedia file, the first position letter in the head and the tail region Breath is used to indicate first position of the head and the tail region in the multimedia file;
Second processing module obtains the head and the tail for calling machine learning model to handle the first location information The second location information in region, the machine learning model are to be obtained using multiple groups training sample data to neural metwork training , every group of training sample data in the multiple groups training sample data include the first of the head and the tail region of sample multimedia file The actual position information in the head and the tail region of location information and the sample multimedia file, the second in the head and the tail region Confidence breath is used to indicate the second position of the head and the tail region in the multimedia file, and the precision of the second position is higher than The precision of the first position.
10. a kind of computer equipment, which is characterized in that the computer equipment includes processor and memory, the memory In be stored at least one instruction, at least one section of program, code set or instruction set, at least one instruction, described at least one Duan Chengxu, the code set or instruction set are loaded as the processor and are executed to realize as described in any one of claim 1 to 8 Determination multimedia file in head and the tail region method.
11. a kind of computer readable storage medium, which is characterized in that be stored at least one in the computer readable storage medium Item instruction, at least one section of program, code set or instruction set, at least one instruction, at least one section of program, the code Collection or instruction set are loaded by processor and are executed to realize in determining multimedia file as claimed in any one of claims 1 to 8 The method in head and the tail region.
CN201810928792.7A 2018-08-15 2018-08-15 Method, device and storage medium for determining head and tail areas in multimedia file Active CN109101964B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810928792.7A CN109101964B (en) 2018-08-15 2018-08-15 Method, device and storage medium for determining head and tail areas in multimedia file

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810928792.7A CN109101964B (en) 2018-08-15 2018-08-15 Method, device and storage medium for determining head and tail areas in multimedia file

Publications (2)

Publication Number Publication Date
CN109101964A true CN109101964A (en) 2018-12-28
CN109101964B CN109101964B (en) 2023-04-07

Family

ID=64849872

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810928792.7A Active CN109101964B (en) 2018-08-15 2018-08-15 Method, device and storage medium for determining head and tail areas in multimedia file

Country Status (1)

Country Link
CN (1) CN109101964B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111709764A (en) * 2019-03-18 2020-09-25 腾讯科技(深圳)有限公司 Method and device for determining related parameters of multimedia content and storage medium
CN113032679A (en) * 2021-04-19 2021-06-25 北京新三优秀科技有限公司 Short video processing method, electronic device and computer readable storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103974145A (en) * 2014-04-14 2014-08-06 百度在线网络技术(北京)有限公司 Film leader and/or tail leader recognition method and device for multi-media file
US9418296B1 (en) * 2015-03-17 2016-08-16 Netflix, Inc. Detecting segments of a video program
CN105979373A (en) * 2015-12-03 2016-09-28 乐视致新电子科技(天津)有限公司 Play method and device
US20170019444A1 (en) * 2015-07-14 2017-01-19 Echostar Technologies L.L.C. Media content skipping
CN108228915A (en) * 2018-03-29 2018-06-29 华南理工大学 A kind of video retrieval method based on deep learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103974145A (en) * 2014-04-14 2014-08-06 百度在线网络技术(北京)有限公司 Film leader and/or tail leader recognition method and device for multi-media file
US9418296B1 (en) * 2015-03-17 2016-08-16 Netflix, Inc. Detecting segments of a video program
US20170019444A1 (en) * 2015-07-14 2017-01-19 Echostar Technologies L.L.C. Media content skipping
CN105979373A (en) * 2015-12-03 2016-09-28 乐视致新电子科技(天津)有限公司 Play method and device
CN108228915A (en) * 2018-03-29 2018-06-29 华南理工大学 A kind of video retrieval method based on deep learning

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111709764A (en) * 2019-03-18 2020-09-25 腾讯科技(深圳)有限公司 Method and device for determining related parameters of multimedia content and storage medium
CN111709764B (en) * 2019-03-18 2023-04-18 腾讯科技(深圳)有限公司 Method and device for determining related parameters of multimedia content and storage medium
CN113032679A (en) * 2021-04-19 2021-06-25 北京新三优秀科技有限公司 Short video processing method, electronic device and computer readable storage medium
CN113032679B (en) * 2021-04-19 2023-12-29 北京新三优秀科技有限公司 Short video processing method, electronic equipment and computer readable storage medium

Also Published As

Publication number Publication date
CN109101964B (en) 2023-04-07

Similar Documents

Publication Publication Date Title
CN105657535A (en) Audio recognition method and device
CN110557659B (en) Video recommendation method and device, server and storage medium
CN104394126B (en) Information recommendation method, server, client and system
CN105874451A (en) Methods, systems, and media for presenting supplemental information corresponding to on-demand media content
CN104380717B (en) Information processing unit, information processing method and computer readable storage medium
CN109660823A (en) Video distribution method, apparatus, electronic equipment and storage medium
KR20160132047A (en) Communication event history
CN110347866B (en) Information processing method, information processing device, storage medium and electronic equipment
CN109286848B (en) Terminal video information interaction method and device and storage medium
KR20170140188A (en) Method and apparatus for audio content recognition
CN107872713A (en) Short processing system for video, method and device
US20240062772A1 (en) Audio Playout Report for Ride-Sharing Session
CN109101964A (en) Determine the method, equipment and storage medium in head and the tail region in multimedia file
CN111723235B (en) Music content identification method, device and equipment
CN108881766B (en) Video processing method, device, terminal and storage medium
CN103984699B (en) The method for pushing and device of promotion message
US11874942B1 (en) User access to meeting recordings
US8306992B2 (en) System for determining content topicality, and method and program thereof
CN116739647A (en) Marketing data intelligent analysis method and system
CN109688430A (en) A kind of court trial file playback method, system and storage medium
CN112601129B (en) Video interaction system, method and receiving terminal
CN102473178A (en) Method and computer program product for enabling organization of media objects
CN115866290A (en) Video dotting method, device, equipment and storage medium
CN113782027A (en) Audio processing method and audio processing device
CN113573096A (en) Video processing method, video processing device, electronic equipment and medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant