CN107833214A - Video definition detection method, device, computing device and computer-readable storage medium - Google Patents
Video definition detection method, device, computing device and computer-readable storage medium Download PDFInfo
- Publication number
- CN107833214A CN107833214A CN201711071515.0A CN201711071515A CN107833214A CN 107833214 A CN107833214 A CN 107833214A CN 201711071515 A CN201711071515 A CN 201711071515A CN 107833214 A CN107833214 A CN 107833214A
- Authority
- CN
- China
- Prior art keywords
- video
- file
- definition
- video file
- detected
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of video definition detection method, device, computing device and computer-readable storage medium, and wherein method is based on trained feature extraction network and trained video definition detection model performs, and method includes:Obtain video file to be detected, video file to be detected is inputted to feature extraction network, obtain the file characteristic of default dimension corresponding with video file to be detected, it dramatically saves on the time, improve the efficiency and accuracy rate of feature extraction, and the influence of false video metamessage can be avoided, file characteristic based on default dimension enters line definition detection using video definition detection model, obtain definition values corresponding to video file, realize file automatic detection, the time required to saving detection, reduce cost of labor, due to the influence of false video metamessage can be avoided, so as to improve the accuracy rate of video definition detection.
Description
Technical field
The present invention relates to technical field of video processing, and in particular to a kind of video definition detection method, device, calculating are set
Standby and computer-readable storage medium.
Background technology
Video definition is an important indicator for influenceing video viewing experience.The unsharp factor of video is caused to include:
Video resolution is too low, video code rate is too low, video darker or lighter, video are excessively shaken.It is flooded with present in internet
A large amount of unsharp low quality videos, existing video definition evaluation method mainly include following several:
(1) manual examination and verification.Whether manual examination and verification video is clear, it is necessary to expend substantial amounts of human cost, and audits video
Required time is longer;
(2) audited based on video metamessage.Video element information includes the information such as resolution ratio, code check, is deposited on internet
The video of false resolution rate and false code check as metamessage largely is being used, the method based on video metamessage is in this type
Video on can judge by accident.
(3) low-level features based on video content.Now just need the low-level features of artificial complex designing and can not profit
With existing massive video data, waste of resource.
The content of the invention
In view of the above problems, it is proposed that the present invention so as to provide one kind overcome above mentioned problem or at least in part solve on
State video definition detection method, device, computing device and the computer-readable storage medium of problem.
According to an aspect of the invention, there is provided a kind of video definition detection method, method is based on trained
Feature extraction network and trained video definition detection model perform, and method includes:
Obtain video file to be detected;
Video file to be detected is inputted to feature extraction network, obtains default dimension corresponding with video file to be detected
File characteristic;
File characteristic based on default dimension enters line definition detection using video definition detection model, obtains video text
Definition values corresponding to part.
Alternatively, video file to be detected is inputted to feature extraction network, obtained corresponding with video file to be detected
The file characteristic of default dimension further comprises:
Decoding process is carried out to video file to be detected, obtains multiple frame of video;
For each frame of video of multiple frame of video, frame of video is inputted to feature extraction network, obtained and the frame of video
The frame feature of corresponding default dimension;
Calculation process is carried out to the frame feature of multiple frame of video using preset algorithm, obtained corresponding with video file to be detected
Default dimension file characteristic.
Alternatively, calculation process is carried out to the frame feature of multiple frame of video using preset algorithm, obtained and video to be detected
The file characteristic that dimension is preset corresponding to file further comprises:
The average value of the characteristic value of the frame feature of multiple frame of video is calculated, is obtained corresponding with video file to be detected default
The file characteristic of dimension.
Alternatively, decoding process is being carried out to video file to be detected, after obtaining multiple frame of video, method also includes:
Take out frame strategy using default at least one frame of video is extracted from multiple frame of video.
Alternatively, video definition detection model training sample used includes:Multiple video files of sample library storage
Sample and video file definition annotation results corresponding with video file sample;Plurality of video file sample includes more
Individual video file positive sample and multiple video file negative samples.
Alternatively, the training process of video definition detection model includes:
Video file sample is inputted to the text for feature extraction network, obtaining default dimension corresponding with video file sample
Part feature;
File characteristic is inputted to video definition detection model and is trained, obtains regard corresponding with video file sample
Frequency file definition testing result;
According to the loss between video file definition testing result and video file definition annotation results, video is obtained
Definition detection model loss function, video definition detection model is updated according to video definition detection model loss function
Weight parameter;
Iteration performs above-mentioned steps, until meeting predetermined convergence condition.
Alternatively, predetermined convergence condition includes:Iterations reaches default iterations;And/or video definition detection
The output valve of model loss function is less than predetermined threshold value.
According to another aspect of the present invention, there is provided a kind of video definition detection means, device are based on trained
Feature extraction network and trained video definition detection model perform, and device includes:
Acquisition module, suitable for obtaining video file to be detected;
Feature extraction module, suitable for video file to be detected is inputted to feature extraction network, obtain and video to be detected
The file characteristic of dimension is preset corresponding to file;
Definition detection module, carried out clearly using video definition detection model suitable for the file characteristic based on default dimension
Clear degree detection, obtains definition values corresponding to video file.
Alternatively, feature extraction module further comprises:
Codec processing unit, suitable for carrying out decoding process to video file to be detected, obtain multiple frame of video;
Fisrt feature extracting unit, suitable for each frame of video for multiple frame of video, frame of video is inputted to feature and taken out
Network is taken, obtains the frame feature of default dimension corresponding with the frame of video;
Operation processing unit, suitable for carrying out calculation process to the frame features of multiple frame of video using preset algorithm, obtain with
The file characteristic of dimension is preset corresponding to video file to be detected.
Alternatively, operation processing unit is further adapted for:The average value of the characteristic value of the frame feature of multiple frame of video is calculated,
Obtain the file characteristic of default dimension corresponding with video file to be detected.
Alternatively, feature extraction module also includes:Frame of video extracting unit, suitable for taking out frame strategy using default and being regarded from multiple
At least one frame of video is extracted in frequency frame.
Alternatively, video definition detection model training sample used includes:Multiple video files of sample library storage
Sample and video file definition annotation results corresponding with video file sample;Plurality of video file sample includes more
Individual video file positive sample and multiple video file negative samples.
Alternatively, device also includes:Video definition detection model training module;
Video definition detection model training module includes:
Second feature extracting unit, suitable for video file sample is inputted to feature extraction network, obtain and video file
The file characteristic of dimension is preset corresponding to sample;
Training unit, it is trained, obtains and video text suitable for file characteristic is inputted to video definition detection model
Video file definition testing result corresponding to part sample;
Updating block, suitable for according between video file definition testing result and video file definition annotation results
Loss, obtains video definition detection model loss function, clear according to video definition detection model loss function more new video
The weight parameter of clear degree detection model;
Video definition detection model training module iteration is run, until meeting predetermined convergence condition.
Alternatively, predetermined convergence condition includes:Iterations reaches default iterations;And/or video definition detection
The output valve of model loss function is less than predetermined threshold value.
According to another aspect of the invention, there is provided a kind of computing device, including:Processor, memory, communication interface and
Communication bus, processor, memory and communication interface complete mutual communication by communication bus;
Memory is used to deposit an at least executable instruction, and executable instruction makes the above-mentioned video definition inspection of computing device
Operated corresponding to survey method.
In accordance with a further aspect of the present invention, there is provided a kind of computer-readable storage medium, be stored with least one in storage medium
Executable instruction, executable instruction make computing device be operated as corresponding to above-mentioned video definition detection method.
According to scheme provided by the invention, the text of trained feature extraction network abstraction video file to be detected is utilized
Part feature, dramatically saves on the time, improve the efficiency of feature extraction, and the accuracy rate of feature extraction, and can keep away
Exempt from the influence of false video metamessage, video file to be detected is carried out using trained video definition detection model clear
Clear degree detection, realizes file automatic detection, without manually checking that complete video file just can determine that whether video file is clear
It is clear, the time required to saving detection, cost of labor is reduced, due to the influence of false video metamessage can be avoided, so as to improve
The accuracy rate of video definition detection, and take full advantage of existing massive video data and carry out video definition detection,
Further improve the accuracy of detection.
Described above is only the general introduction of technical solution of the present invention, in order to better understand the technological means of the present invention,
And can be practiced according to the content of specification, and in order to allow above and other objects of the present invention, feature and advantage can
Become apparent, below especially exemplified by the embodiment of the present invention.
Brief description of the drawings
By reading the detailed description of hereafter preferred embodiment, it is various other the advantages of and benefit it is common for this area
Technical staff will be clear understanding.Accompanying drawing is only used for showing the purpose of preferred embodiment, and is not considered as to the present invention
Limitation.And in whole accompanying drawing, identical part is denoted by the same reference numerals.In the accompanying drawings:
Fig. 1 shows the schematic flow sheet of video definition detection method according to an embodiment of the invention;
Fig. 2 shows the flow signal of video definition detection model training method according to an embodiment of the invention
Figure;
Fig. 3 shows the schematic flow sheet of video definition detection method in accordance with another embodiment of the present invention;
Fig. 4 shows the structure journey schematic diagram of video definition detection means according to an embodiment of the invention;
Fig. 5 shows the structure journey schematic diagram of video definition detection means in accordance with another embodiment of the present invention;
Fig. 6 shows a kind of structural representation of computing device according to an embodiment of the invention.
Embodiment
The exemplary embodiment of the disclosure is more fully described below with reference to accompanying drawings.Although the disclosure is shown in accompanying drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
Limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
Completely it is communicated to those skilled in the art.
Fig. 1 shows the schematic flow sheet of video definition detection method according to an embodiment of the invention.Wherein, side
Method is based on trained feature extraction network and trained video definition detection model performs, as shown in figure 1,
This method comprises the following steps:
Step S100, obtain video file to be detected.
Wherein, video file to be detected is that user wants the video file into line definition detection, video file to be detected
Can be the video file that terminal device is locally stored or the video text that user downloads according to detection demand from network
Part.So that video file to be detected is the video file downloaded from network as an example, from network after downloading video files, Ke Yixiang
User provides an interface, and the interface provides video file upload function, after user clicks on upload button, ejection file upload pair
Frame is talked about, user selects to need the video file to be detected uploaded, clicks on the upload button in dialog box, completes video text to be detected
The upload of part, in this way, to get video file to be detected then follow-up definition detection can be carried out.
Step S101, video file to be detected is inputted to feature extraction network, obtained corresponding with video file to be detected
Default dimension file characteristic.
Feature extraction network obtains by training, specifically, can be based on substantial amounts of video file sample training and roll up
Product neutral net obtains feature extraction network, and the file characteristic of certain dimension can be extracted by extracting network using this feature, wherein,
Convolutional neural networks (Convolutional Neural Network) are a kind of artificial neural networks, and a kind of have supervision
Machine learning model, it is one of representative method of deep learning.
After video file to be detected is got, video file to be detected is inputted to trained feature extraction net
Network, extract network using this feature and carry out file characteristic extraction, obtain the text of default dimension corresponding with video file to be detected
Part feature.In embodiments of the present invention, it is using the advantage of feature extraction network, it is a series of multiple when avoiding artificial extraction feature
Miscellaneous preprocessing process, dramatically saves on the time, improve the efficiency of feature extraction, and also improve the standard of feature extraction
True rate.
Step S102, the file characteristic based on default dimension enter line definition using video definition detection model and detected,
Obtain definition values corresponding to video file.
After step S101 obtains the file characteristic of default dimension, can the file characteristic based on default dimension using regarding
Frequency definition detection model enters line definition detection, obtains definition values corresponding to video file, in embodiments of the present invention, depending on
Frequency definition detection model is two disaggregated models, i.e., whether can obtain video file using the video definition detection model clear
Clear testing result, it can determine whether video file is clear according to definition values corresponding to video file.
The method provided according to the above embodiment of the present invention, is regarded using trained feature extraction network abstraction is to be detected
The file characteristic of frequency file, dramatically saves on the time, improve the efficiency of feature extraction, and the accuracy rate of feature extraction,
And the influence of false video metamessage can be avoided, using trained video definition detection model to video to be detected
File enters line definition detection, file automatic detection is realized, without manually checking that complete video file just can determine that video
Whether file is clear, the time required to saving detection, cost of labor is reduced, due to that can avoid the shadow of false video metamessage
Ring, so as to improve the accuracy rate of video definition detection.
Fig. 2 shows the flow signal of video definition detection model training method according to an embodiment of the invention
Figure.As shown in Fig. 2 this method comprises the following steps:
Step S200, video file sample is inputted to feature extraction network, obtained corresponding with video file sample pre-
If the file characteristic of dimension.
Multiple video file samples are not only stored in the Sample Storehouse of video definition detection model, also stored for and video
Video file definition annotation results corresponding to paper sample, plurality of video file sample include the positive sample of multiple video files
This (i.e. clearly video file) and multiple video file negative samples (i.e. unsharp video file).Those skilled in the art can
The quantity for setting the paper sample stored in Sample Storehouse according to being actually needed, is not limited herein.
In embodiments of the present invention, video file can be divided into clear and unintelligible two class according to definition, can incited somebody to action
Clearly video file is labeled as 1, and unsharp video file is labeled as 0, it is of course also possible to by unsharp video file mark
1 is designated as, clearly video file is labeled as 0, and the annotation results of video file, which affect, subsequently to be obtained corresponding to video file
After definition values, video file whether clearly result is determined according to definition values, for example, setting will clearly video
File mark is 1, unsharp video file be labeled as 0, definition values for [0,0.7) when assert video it is unintelligible, definition
It is worth to assert that video is clear when [0.7,1], it is assumed that definition values corresponding to resulting video file are 0.8, then can be assumed that
Video file is clear.Specifically, the method manually marked can be used to be labeled each video file sample in Sample Storehouse,
Obtain video file definition annotation results corresponding with video file sample.
Specifically, video file sample is extracted from Sample Storehouse, and extracts video text corresponding with the video file sample
Part definition annotation results, video file sample is inputted to feature extraction network, utilize the feature extraction network abstraction video
The file characteristic of paper sample, wherein, feature extraction network can extract the file characteristic of 1024 dimensions, and the file of 1024 dimension is special
Sign more can comprehensively embody content information, sharpness information of file etc., be a kind of generic features.
In alternative embodiment of the present invention, after video file sample is extracted, decoder can be utilized to video text
Part sample carries out decoding process, obtains multiple frame of video, at least one frame of video is extracted from multiple frame of video, by extraction extremely
Each frame in a few frame of video inputs the frame feature for feature extraction network, obtaining 1024 dimension corresponding with the frame of video,
Then, using preset algorithm, for example, being averaging, calculation process is carried out to the frame feature of multiple frame of video, obtained and video file
The file characteristic of 1024 dimensions corresponding to sample.
Step S201, file characteristic is inputted to video definition detection model and is trained, obtained and video file sample
Video file definition testing result corresponding to this.
The video definition detection model trained in the embodiment of the present invention detects suitable for video definition, can only detect
Whether clear video is, is special detection, and cannot be used for other detections, for example whether the detection comprising abnormal information.
After file characteristic corresponding to video file sample is obtained, file characteristic is inputted to video definition detection model
It is trained, obtains video file definition testing result corresponding with paper sample, wherein, it is corresponding with video file sample
Video file definition testing result value is the numerical value in [0,1].
Step S202, according to the damage between video file definition testing result and video file definition annotation results
Lose, obtain video definition detection model loss function, it is clear according to video definition detection model loss function more new video
Spend the weight parameter of detection model.
Wherein, those skilled in the art can set the specific of video definition detection model loss function according to being actually needed
Content, do not limit herein.According to video definition detection model loss function, carry out gradient and decline optimization, regarded to update
The model parameter of frequency definition detection model.
Enter below by multi-layer perception (MLP) (Multi Layer Perceptron, i.e. MLP) of video definition detection model
Row describes in detail, wherein, multi-layer perception (MLP) is the master pattern that two classification are carried out to sample, and video file sample can be carried out
Classification, it is to apply secondly principle of classification carries out video definition detection here, a perceptron has consisting of part:
Weights are inputted, a perceptron can receive multiple inputs, there are a weights in each input, in addition, also one
Individual bias term;
Activation primitive, its conventional activation primitive are Sigmoid (S types activation primitive), are embodied as:
Output.
Multi-layer perception (MLP) includes an input layer, at least one hidden layer and an output layer, wherein, output layer is
Softmax layers, it can realize that video file is classified using softmax layers.
Further, it is also possible to this special regression model, SVMs (SVM), multi-layer perception (MLP) (MLP), random using logic
Forest enters line definition detection, is no longer described in detail here.
Step S203, iteration performs above-mentioned steps, until meeting predetermined convergence condition.
Wherein, those skilled in the art can set predetermined convergence condition according to being actually needed, and not limit herein.For example,
Predetermined convergence condition may include:Iterations reaches default iterations;And/or video definition detection model loss function
Output valve be less than predetermined threshold value.Specifically, can be by judging whether iterations reaches default iterations to judge to be
It is no to meet predetermined convergence condition, whether default threshold can also be less than according to the output valve of video definition detection model loss function
Value judges whether to meet predetermined convergence condition.In step S203, iteration performs the training step of video definition detection model
Suddenly, until meeting predetermined convergence condition, so as to obtain trained video definition detection model.
Fig. 3 shows the schematic flow sheet of video definition detection method in accordance with another embodiment of the present invention.Wherein,
Method is based on trained feature extraction network and trained video definition detection model performs, such as Fig. 3 institutes
Show, this method comprises the following steps:
Step S300, obtain video file to be detected.
Step S301, decoding process is carried out to video file to be detected, obtains multiple frame of video.
Video file is made up of frame of video one by one, in order to extract video file to be detected exactly
File characteristic, it is necessary to be carried out to video file to be detected at decoding before the file characteristic of video file to be detected is extracted
Reason, obtains multiple frame of video.
Step S302, take out frame strategy using default at least one frame of video is extracted from multiple frame of video.
Each video file is made up of many frame of video, is taken out if all carrying out file characteristic to the video file after decoding process
Take, the file characteristic of extraction can be caused very more, can be realized subsequently although all frame of video are all carried out with file characteristic and is extracted
Definition detects, but can cause the waste of resource, while adds feature extraction required time, reduces feature extraction efficiency,
Therefore, the embodiment of the present invention can extract at least one frame of video to carry out subsequent frame feature extraction, example from multiple frame of video
Such as, the frame of video of predetermined number, such as a frame or two frames can be extracted, the quantity of extraction can be set according to being actually needed
It is fixed, in order to lift the accuracy of detection, several frames can be extracted more, be merely illustrative of here, without any restriction effect.Should
Step is optional step, and the frame of video that can also be obtained to decoding process all carries out feature extraction processing.
Step S303, for each frame of video of at least one frame of video, frame of video is inputted to feature extraction network, obtained
To the frame feature of default dimension corresponding with the frame of video.
After extraction obtains at least one frame of video, for each frame of video of at least one frame of video, by frame of video
Input extracts network using this feature and carries out frame feature extraction, for example, can extract to trained feature extraction network
The frame feature of 1024 dimensions.In embodiments of the present invention, it is using the advantage of feature extraction network, when avoiding artificial extraction feature
The preprocessing process of a series of complex, dramatically saves on the time, improve the efficiency of feature extraction, and also improve feature
The accuracy rate of extraction.
Step S304, calculation process is carried out to the frame feature of multiple frame of video using preset algorithm, obtains regarding with to be detected
The file characteristic of dimension is preset corresponding to frequency file.
Specifically, the frame feature of multiple frame of video can be averaging, calculates the spy of the frame feature of multiple frame of video
The average value of value indicative, obtain the file characteristic of default dimension corresponding with video file to be detected.
Step S305, the file characteristic based on default dimension enter line definition using video definition detection model and detected,
Obtain definition values corresponding to video file.
After step S304 obtains the file characteristic of default dimension, can the file characteristic based on default dimension using regarding
Frequency definition detection model enters line definition detection, obtains definition values corresponding to video file, in embodiments of the present invention, depending on
Frequency definition detection model is two disaggregated models, i.e., whether can obtain video file using the video definition detection model clear
Clear testing result, it can determine whether video file is clear according to definition values corresponding to video file.
Wherein, definition values corresponding to resulting video file are represented with the numerical value in [0,1], setting video text
When definition values span corresponding to part is [0.7,1], video is clear, and definition values span is corresponding to video file
[0,0.7) when, video is unintelligible, and after line definition detection is entered, the definition for obtaining the video file to be detected is 0.8, then
It can be assumed that the video file to be detected is clear.Here it is merely illustrative of, without any restriction effect.
The method provided according to the above embodiment of the present invention, is regarded using trained feature extraction network abstraction is to be detected
The file characteristic of frequency file, dramatically saves on the time, improve the efficiency of feature extraction, and the accuracy rate of feature extraction,
And the influence of false video metamessage can be avoided, using trained video definition detection model to video to be detected
File enters line definition detection, file automatic detection is realized, without manually checking that complete video file just can determine that video
Whether file is clear, the time required to saving detection, cost of labor is reduced, due to that can avoid the shadow of false video metamessage
Ring, so as to improve the accuracy rate of video definition detection.
Fig. 4 shows the structure journey schematic diagram of video definition detection means according to an embodiment of the invention.Wherein,
Device is based on trained feature extraction network and trained video definition detection model performs, such as Fig. 4 institutes
Show, the device includes:Acquisition module 400, feature extraction module 410, definition detection module 420.
Acquisition module 400, suitable for obtaining video file to be detected.
Feature extraction module 410, suitable for video file to be detected is inputted to feature extraction network, obtain regarding with to be detected
The file characteristic of dimension is preset corresponding to frequency file.
Definition detection module 420, entered suitable for the file characteristic based on default dimension using video definition detection model
Line definition detects, and obtains definition values corresponding to video file.
The device provided according to the above embodiment of the present invention, is regarded using trained feature extraction network abstraction is to be detected
The file characteristic of frequency file, dramatically saves on the time, improve the efficiency of feature extraction, and the accuracy rate of feature extraction,
And the influence of false video metamessage can be avoided, using trained video definition detection model to video to be detected
File enters line definition detection, file automatic detection is realized, without manually checking that complete video file just can determine that video
Whether file is clear, the time required to saving detection, cost of labor is reduced, due to that can avoid the shadow of false video metamessage
Ring, so as to improve the accuracy rate of video definition detection.
Fig. 5 shows the structure journey schematic diagram of video definition detection means in accordance with another embodiment of the present invention.Its
In, device is based on trained feature extraction network and trained video definition detection model performs, such as Fig. 5
Shown, the device includes:It is acquisition module 500, feature extraction module 510, video definition detection model training module 520, clear
Clear degree detection module 530.
Acquisition module 500, suitable for obtaining video file to be detected.
Feature extraction module 510 further comprises:Codec processing unit 511, suitable for being solved to video file to be detected
Code processing, obtains multiple frame of video;
Frame of video extracting unit 512, suitable for taking out frame strategy using default at least one video being extracted from multiple frame of video
Frame.
Fisrt feature extracting unit 513, suitable for each frame of video for multiple frame of video, frame of video is inputted to feature
Network is extracted, obtains the frame feature of default dimension corresponding with the frame of video;
Operation processing unit 514, suitable for carrying out calculation process to the frame feature of multiple frame of video using preset algorithm, obtain
The file characteristic of default dimension corresponding with video file to be detected.
Wherein, operation processing unit 514 is further adapted for:Calculate being averaged for the characteristic value of the frame feature of multiple frame of video
Value, obtain the file characteristic of default dimension corresponding with video file to be detected.
In embodiments of the present invention, video definition detection model training sample used includes:Sample library storage it is more
Individual video file sample and video file definition annotation results corresponding with video file sample;Plurality of video file
Sample includes multiple video file positive samples and multiple video file negative samples.
Video definition detection model training module 520 includes:Second feature extracting unit 521, suitable for by video file
Sample inputs the file characteristic for feature extraction network, obtaining default dimension corresponding with video file sample;
Training unit 522, it is trained, obtains and video suitable for file characteristic is inputted to video definition detection model
Video file definition testing result corresponding to paper sample;
Updating block 523, suitable for according to video file definition testing result and video file definition annotation results it
Between loss, obtain video definition detection model loss function, according to video definition detection model loss function renewal regard
The weight parameter of frequency definition detection model;
Video definition detection model training module iteration is run, until meeting predetermined convergence condition.
Wherein, predetermined convergence condition includes:Iterations reaches default iterations;And/or video definition detection mould
The output valve of type loss function is less than predetermined threshold value.
Definition detection module 530, entered suitable for the file characteristic based on default dimension using video definition detection model
Line definition detects, and obtains definition values corresponding to video file.
The device provided according to the above embodiment of the present invention, is regarded using trained feature extraction network abstraction is to be detected
The file characteristic of frequency file, dramatically saves on the time, improve the efficiency of feature extraction, and the accuracy rate of feature extraction,
And the influence of false video metamessage can be avoided, using trained video definition detection model to video to be detected
File enters line definition detection, file automatic detection is realized, without manually checking that complete video file just can determine that video
Whether file is clear, the time required to saving detection, cost of labor is reduced, due to that can avoid the shadow of false video metamessage
Ring, so as to improve the accuracy rate of video definition detection.
The embodiment of the present application additionally provides a kind of nonvolatile computer storage media, the computer-readable storage medium storage
There is an at least executable instruction, the computer executable instructions can perform the video definition inspection in above-mentioned any means embodiment
Survey method.
Fig. 6 shows a kind of structural representation of computing device according to an embodiment of the invention, of the invention specific real
Specific implementation of the example not to computing device is applied to limit.
As shown in fig. 6, the computing device can include:Processor (processor) 602, communication interface
(Communications Interface) 604, memory (memory) 606 and communication bus 608.
Wherein:
Processor 602, communication interface 604 and memory 606 complete mutual communication by communication bus 608.
Communication interface 604, for being communicated with the network element of miscellaneous equipment such as client or other servers etc..
Processor 602, for configuration processor 610, it can specifically perform in above-mentioned video definition detection method embodiment
Correlation step.
Specifically, program 610 can include program code, and the program code includes computer-managed instruction.
Processor 602 is probably central processor CPU, or specific integrated circuit ASIC (Application
Specific Integrated Circuit), or it is arranged to implement the integrated electricity of one or more of the embodiment of the present invention
Road.The one or more processors that computing device includes, can be same type of processor, such as one or more CPU;Also may be used
To be different types of processor, such as one or more CPU and one or more ASIC.
Memory 606, for depositing program 610.Memory 606 may include high-speed RAM memory, it is also possible to also include
Nonvolatile memory (non-volatile memory), for example, at least a magnetic disk storage.
Program 610 specifically can be used for so that processor 602 performs the video definition in above-mentioned any means embodiment
Detection method.The specific implementation of each step may refer to the corresponding step in above-mentioned video definition detection embodiment in program 610
Corresponding description in rapid and unit, will not be described here.It is apparent to those skilled in the art that the side for description
Just and succinctly, the specific work process of the equipment of foregoing description and module, it may be referred to corresponding in preceding method embodiment
Journey describes, and will not be repeated here.
Algorithm and display be not inherently related to any certain computer, virtual system or miscellaneous equipment provided herein.
Various general-purpose systems can also be used together with teaching based on this.As described above, required by constructing this kind of system
Structure be obvious.In addition, the present invention is not also directed to any certain programmed language.It should be understood that it can utilize various
Programming language realizes the content of invention described herein, and the description done above to language-specific is to disclose this hair
Bright preferred forms.
In the specification that this place provides, numerous specific details are set forth.It is to be appreciated, however, that the implementation of the present invention
Example can be put into practice in the case of these no details.In some instances, known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this description.
Similarly, it will be appreciated that in order to simplify the disclosure and help to understand one or more of each inventive aspect,
Above in the description to the exemplary embodiment of the present invention, each feature of the invention is grouped together into single implementation sometimes
In example, figure or descriptions thereof.However, the method for the disclosure should be construed to reflect following intention:I.e. required guarantor
The application claims of shield features more more than the feature being expressly recited in each claim.It is more precisely, such as following
Claims reflect as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
Thus the claims for following embodiment are expressly incorporated in the embodiment, wherein each claim is in itself
Separate embodiments all as the present invention.
Those skilled in the art, which are appreciated that, to be carried out adaptively to the module in the equipment in embodiment
Change and they are arranged in one or more equipment different from the embodiment.Can be the module or list in embodiment
Member or component be combined into a module or unit or component, and can be divided into addition multiple submodule or subelement or
Sub-component.In addition at least some in such feature and/or process or unit exclude each other, it can use any
Combination is disclosed to all features disclosed in this specification (including adjoint claim, summary and accompanying drawing) and so to appoint
Where all processes or unit of method or equipment are combined.Unless expressly stated otherwise, this specification (including adjoint power
Profit requires, summary and accompanying drawing) disclosed in each feature can be by providing the alternative features of identical, equivalent or similar purpose come generation
Replace.
In addition, it will be appreciated by those of skill in the art that although some embodiments described herein include other embodiments
In included some features rather than further feature, but the combination of the feature of different embodiments means in of the invention
Within the scope of and form different embodiments.For example, in the following claims, embodiment claimed is appointed
One of meaning mode can use in any combination.
The all parts embodiment of the present invention can be realized with hardware, or to be run on one or more processor
Software module realize, or realized with combinations thereof.It will be understood by those of skill in the art that it can use in practice
Microprocessor or digital signal processor (DSP) are realized in video definition detection device according to embodiments of the present invention
The some or all functions of some or all parts.The present invention is also implemented as being used to perform method as described herein
Some or all equipment or program of device (for example, computer program and computer program product).Such reality
The program of the existing present invention can store on a computer-readable medium, or can have the form of one or more signal.
Such signal can be downloaded from internet website and obtained, and either be provided or in the form of any other on carrier signal
There is provided.
It should be noted that the present invention will be described rather than limits the invention for above-described embodiment, and ability
Field technique personnel can design alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference symbol between bracket should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not
Element or step listed in the claims.Word "a" or "an" before element does not exclude the presence of multiple such
Element.The present invention can be by means of including the hardware of some different elements and being come by means of properly programmed computer real
It is existing.In if the unit claim of equipment for drying is listed, several in these devices can be by same hardware branch
To embody.The use of word first, second, and third does not indicate that any order.These words can be explained and run after fame
Claim.
Claims (10)
1. a kind of video definition detection method, methods described is based on trained feature extraction network and by training
Video definition detection model perform, methods described includes:
Obtain video file to be detected;
The video file to be detected is inputted to feature extraction network, obtained corresponding with the video file to be detected default
The file characteristic of dimension;
File characteristic based on the default dimension enters line definition detection using video definition detection model, obtains video text
Definition values corresponding to part.
2. the method according to claim 11, wherein, it is described to input the video file to be detected to feature extraction net
Network, the file characteristic for obtaining default dimension corresponding with the video file to be detected further comprise:
Decoding process is carried out to the video file to be detected, obtains multiple frame of video;
For each frame of video of multiple frame of video, frame of video is inputted to feature extraction network, obtained corresponding with the frame of video
Default dimension frame feature;
Calculation process is carried out to the frame feature of multiple frame of video using preset algorithm, obtained corresponding with the video file to be detected
Default dimension file characteristic.
3. method according to claim 1 or 2, wherein, it is described that the frame feature of multiple frame of video is entered using preset algorithm
Row calculation process, the file characteristic for obtaining default dimension corresponding with the video file to be detected further comprise:
The average value of the characteristic value of the frame feature of multiple frame of video is calculated, is obtained corresponding with the video file to be detected default
The file characteristic of dimension.
4. according to the method described in claim any one of 1-3, wherein, carried out to the video file to be detected at decoding
Reason, after obtaining multiple frame of video, methods described also includes:
Take out frame strategy using default at least one frame of video is extracted from the multiple frame of video.
5. according to the method described in claim any one of 1-4, wherein, video definition detection model training sample used
Originally include:Multiple video file samples of sample library storage and video file definition corresponding with video file sample mark
As a result;Plurality of video file sample includes multiple video file positive samples and multiple video file negative samples.
6. according to the method described in claim any one of 1-5, wherein, the training process bag of the video definition detection model
Include:
Video file sample is inputted to the file spy for feature extraction network, obtaining default dimension corresponding with video file sample
Sign;
The file characteristic is inputted to video definition detection model and is trained, is obtained corresponding with the video file sample
Video file definition testing result;
According to the loss between the video file definition testing result and the video file definition annotation results, obtain
Video definition detection model loss function, it is clear that the video is updated according to the video definition detection model loss function
Spend the weight parameter of detection model;
Iteration performs above-mentioned steps, until meeting predetermined convergence condition.
7. according to the method described in claim any one of 1-6, wherein, the predetermined convergence condition includes:Iterations reaches
Default iterations;And/or the output valve of the video definition detection model loss function is less than predetermined threshold value.
8. a kind of video definition detection means, described device is based on trained feature extraction network and by training
Video definition detection model perform, described device includes:
Acquisition module, suitable for obtaining video file to be detected;
Feature extraction module, suitable for the video file to be detected is inputted to feature extraction network, obtain with it is described to be detected
The file characteristic of dimension is preset corresponding to video file;
Definition detection module, carried out clearly using video definition detection model suitable for the file characteristic based on the default dimension
Clear degree detection, obtains definition values corresponding to video file.
9. a kind of computing device, including:Processor, memory, communication interface and communication bus, the processor, the storage
Device and the communication interface complete mutual communication by the communication bus;
The memory is used to deposit an at least executable instruction, and the executable instruction makes the computing device such as right will
Ask and operated corresponding to the video definition detection method any one of 1-7.
10. a kind of computer-readable storage medium, an at least executable instruction, the executable instruction are stored with the storage medium
Make operation corresponding to video definition detection method of the computing device as any one of claim 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711071515.0A CN107833214A (en) | 2017-11-03 | 2017-11-03 | Video definition detection method, device, computing device and computer-readable storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711071515.0A CN107833214A (en) | 2017-11-03 | 2017-11-03 | Video definition detection method, device, computing device and computer-readable storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107833214A true CN107833214A (en) | 2018-03-23 |
Family
ID=61654527
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711071515.0A Pending CN107833214A (en) | 2017-11-03 | 2017-11-03 | Video definition detection method, device, computing device and computer-readable storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107833214A (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110113669A (en) * | 2019-06-14 | 2019-08-09 | 北京达佳互联信息技术有限公司 | Obtain method, apparatus, electronic equipment and the storage medium of video data |
CN110278485A (en) * | 2019-07-29 | 2019-09-24 | 北京华雨天成文化传播有限公司 | A kind of method and device for assessing video quality |
CN111836073A (en) * | 2020-07-10 | 2020-10-27 | 腾讯科技(深圳)有限公司 | Method, device and equipment for determining video definition and storage medium |
WO2021129435A1 (en) * | 2019-12-27 | 2021-07-01 | 百果园技术(新加坡)有限公司 | Method for training video definition evaluation model, video recommendation method, and related device |
CN113628286A (en) * | 2021-08-09 | 2021-11-09 | 咪咕视讯科技有限公司 | Video color gamut detection method and device, computing equipment and computer storage medium |
WO2022057789A1 (en) * | 2020-09-17 | 2022-03-24 | 上海连尚网络科技有限公司 | Video definition identification method, electronic device, and storage medium |
WO2022116811A1 (en) * | 2020-12-04 | 2022-06-09 | 武汉绿色网络信息服务有限责任公司 | Method and device for predicting definition of video having encrypted traffic |
CN117041625A (en) * | 2023-08-02 | 2023-11-10 | 成都梵辰科技有限公司 | Method and system for constructing ultra-high definition video image quality detection network |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105809704A (en) * | 2016-03-30 | 2016-07-27 | 北京小米移动软件有限公司 | Method and device for identifying image definition |
-
2017
- 2017-11-03 CN CN201711071515.0A patent/CN107833214A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105809704A (en) * | 2016-03-30 | 2016-07-27 | 北京小米移动软件有限公司 | Method and device for identifying image definition |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110113669A (en) * | 2019-06-14 | 2019-08-09 | 北京达佳互联信息技术有限公司 | Obtain method, apparatus, electronic equipment and the storage medium of video data |
CN110278485A (en) * | 2019-07-29 | 2019-09-24 | 北京华雨天成文化传播有限公司 | A kind of method and device for assessing video quality |
WO2021129435A1 (en) * | 2019-12-27 | 2021-07-01 | 百果园技术(新加坡)有限公司 | Method for training video definition evaluation model, video recommendation method, and related device |
CN111836073A (en) * | 2020-07-10 | 2020-10-27 | 腾讯科技(深圳)有限公司 | Method, device and equipment for determining video definition and storage medium |
WO2022057789A1 (en) * | 2020-09-17 | 2022-03-24 | 上海连尚网络科技有限公司 | Video definition identification method, electronic device, and storage medium |
WO2022116811A1 (en) * | 2020-12-04 | 2022-06-09 | 武汉绿色网络信息服务有限责任公司 | Method and device for predicting definition of video having encrypted traffic |
CN113628286A (en) * | 2021-08-09 | 2021-11-09 | 咪咕视讯科技有限公司 | Video color gamut detection method and device, computing equipment and computer storage medium |
CN113628286B (en) * | 2021-08-09 | 2024-03-22 | 咪咕视讯科技有限公司 | Video color gamut detection method, device, computing equipment and computer storage medium |
CN117041625A (en) * | 2023-08-02 | 2023-11-10 | 成都梵辰科技有限公司 | Method and system for constructing ultra-high definition video image quality detection network |
CN117041625B (en) * | 2023-08-02 | 2024-04-19 | 成都梵辰科技有限公司 | Method and system for constructing ultra-high definition video image quality detection network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107833214A (en) | Video definition detection method, device, computing device and computer-readable storage medium | |
CN107895172A (en) | Utilize the method, apparatus and computing device of image information detection anomalous video file | |
CN107801090A (en) | Utilize the method, apparatus and computing device of audio-frequency information detection anomalous video file | |
CN109922032B (en) | Method, device, equipment and storage medium for determining risk of logging in account | |
CN111695719A (en) | User value prediction method and system | |
WO2019111118A1 (en) | Robust gradient weight compression schemes for deep learning applications | |
CN106530305A (en) | Semantic segmentation model training and image segmentation method and device, and calculating equipment | |
CN107748898A (en) | File classifying method, device, computing device and computer-readable storage medium | |
CN107423613A (en) | The method, apparatus and server of device-fingerprint are determined according to similarity | |
CN107729560A (en) | User's portrait building method, device and computing device based on big data | |
CN107977748B (en) | Multivariable distorted time sequence prediction method | |
CN108109066A (en) | A kind of credit scoring model update method and system | |
CN113570007B (en) | Method, device and equipment for optimizing construction of part defect identification model and storage medium | |
US11574096B1 (en) | Systems and methods for time series simulation | |
CN107729873A (en) | Special object detection method, device and computing device in video file | |
CN110321430A (en) | Domain name identification and domain name identification model generation method, device and storage medium | |
CN111159241B (en) | Click conversion estimation method and device | |
CN111738474A (en) | Traffic state prediction method and device | |
CN109242165A (en) | A kind of model training and prediction technique and device based on model training | |
CN113065570B (en) | Image audit detection method and device and electronic equipment | |
Ackermann et al. | Black-box learning of parametric dependencies for performance models | |
CN113569162A (en) | Data processing method, device, equipment and storage medium | |
CN110765352B (en) | User interest identification method and device | |
CN110544166A (en) | Sample generation method, device and storage medium | |
EP4080789A1 (en) | Enhanced uncertainty management for optical communication systems |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180323 |