CN110287888A - A kind of TV station symbol recognition method and system - Google Patents
A kind of TV station symbol recognition method and system Download PDFInfo
- Publication number
- CN110287888A CN110287888A CN201910562188.1A CN201910562188A CN110287888A CN 110287888 A CN110287888 A CN 110287888A CN 201910562188 A CN201910562188 A CN 201910562188A CN 110287888 A CN110287888 A CN 110287888A
- Authority
- CN
- China
- Prior art keywords
- logo
- picture
- symbol recognition
- station symbol
- trained
- 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
- 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
- G06V20/20—Scenes; Scene-specific elements in augmented reality scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/09—Recognition of logos
Abstract
The present embodiments relate to a kind of TV station symbol recognition methods, this method comprises: initial pictures of the acquisition with logo and monitor station target position, logo image to be trained is obtained according to the position of logo;Deep learning frame darknet is used to be trained to obtain TV station symbol recognition model the logo image to be trained;Image to be identified is identified using the TV station symbol recognition model to obtain the classification of logo;Probability corresponding to each logo classification is ranked up, using the highest classification of probability as the classification of images to be recognized.The invention also discloses a kind of TV station symbol recognition systems.The logo information in video or picture file can be fast and accurately identified in TV station symbol recognition method and system of the invention.
Description
Technical field
The invention belongs to the technical field of picture and text software more particularly to a kind of TV station symbol recognition method and system.
Background technique
For the picture or video file of magnanimity in network, it is often necessary to identify the logo information in picture or video file
(i.e. TV station and section purpose mark, it will usually be shown in the specific location of picture or video file).As can correctly identifying platform
Mark, it is all very great to program and TV station's meaning.
Station identification method for distinguishing in the prior art is: firstly, artificially setting the feature of logo picture;Again by logo figure
Piece is converted into a vector;Then, classifier is recycled to identify logo, the feature due to artificially setting picture is not unified
Method, therefore the logo picture inputted can only be converted into corresponding feature vector, and again without other use, and this
Mark recognition methods there is also the speed of identification logo slow, problem that accuracy is low.
Summary of the invention
In view of this, the embodiment of the present invention provides a kind of TV station symbol recognition method and system, so as to accurate, more efficient
Identify the logo information in video or picture file in ground.
In a first aspect, the embodiment of the invention provides a kind of TV station symbol recognition methods, this method comprises:
Initial pictures of the acquisition with logo and monitor station target position, logo to be trained is obtained according to the position of logo
Image;
Deep learning frame darknet is used to be trained to obtain TV station symbol recognition mould the logo image to be trained
Type;
Image to be identified is identified using the TV station symbol recognition model to obtain the classification of logo;
Probability corresponding to each logo classification is ranked up, using the highest classification of probability as images to be recognized
Classification;
Wherein, monitor station target position includes following sub-step:
Piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
Remove the sequence of pictures to be processed noise denoised after picture;
Binary conversion treatment is carried out to the picture after denoising;
Picture after checking binary conversion treatment using rectangle carries out expansive working;
The area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and by default
Threshold value determining table target position.
Second aspect, the embodiment of the invention provides a kind of TV station symbol recognition system, which includes acquisition module, detection mould
Block, TV station symbol recognition model obtain module, identification module and sorting module:
The acquisition module is used to acquire the initial pictures with logo, and the detection module is based on the logo that has
Initial pictures monitor station target position simultaneously obtains logo image to be trained according to the position of logo;
The TV station symbol recognition model obtains module using deep learning frame darknet to the logo image to be trained
It is trained to obtain TV station symbol recognition model;
The TV station symbol recognition model obtains module and identifies image to be identified to obtain platform using the TV station symbol recognition model
Target classification;
The sorting module is ranked up probability corresponding to each logo classification and makees the highest classification of probability
For the classification of images to be recognized;
Wherein, the detection module executes following operation:
Piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
Remove the sequence of pictures to be processed noise denoised after picture;
Binary conversion treatment is carried out to the picture after denoising;
Picture after checking binary conversion treatment using rectangle carries out expansive working;
The area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and by default
Threshold value determining table target position.
TV station symbol recognition method of the invention carries out the logo image to be trained using deep learning frame darknet
To obtain TV station symbol recognition model, the classification for identifying image to be identified using the TV station symbol recognition model to obtain logo is right for training
Probability corresponding to each logo classification is ranked up, using the highest classification of probability as the classification of images to be recognized, thus
The classification of logo and position in images to be recognized are obtained, so as to be fast and accurately identified in video or picture file
Logo information.
Detailed description of the invention
Fig. 1 is a kind of flow diagram of station identification method for distinguishing provided in an embodiment of the present invention;
Fig. 2 is a kind of structural schematic diagram of the system of TV station symbol recognition provided in an embodiment of the present invention.
Specific embodiment
The present invention is described in detail for each embodiment shown in reference to the accompanying drawing, but it should be stated that, these
Embodiment is not limitation of the present invention, those of ordinary skill in the art according to these embodiments made by function, method,
Or equivalent transformation or substitution in structure, all belong to the scope of protection of the present invention within.
Embodiment one
One, a kind of station identification method for distinguishing
Referring to Fig. 1, TV station symbol recognition method in the present embodiment the following steps are included:
S100: initial pictures of the acquisition with logo and monitor station target position are obtained according to the position of logo wait train
Logo image;
S200: deep learning frame darknet is used to be trained to obtain logo the logo image to be trained
Identification model;
S300: image to be identified is identified to obtain the classification of logo using the TV station symbol recognition model;
S400: probability corresponding to each logo classification is ranked up, using the highest classification of probability as to be identified
The classification of image;
Wherein, monitor station target position includes following sub-step:
Piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
Remove the sequence of pictures to be processed noise denoised after picture;
Binary conversion treatment is carried out to the picture after denoising;
Picture after checking binary conversion treatment using rectangle carries out expansive working;
The area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and by default
Threshold value determining table target position.
Two, a kind of specific work process of station identification method for distinguishing
The specific work process of one of the present embodiment station identification method for distinguishing will be described in detail below.
S100: initial pictures of the acquisition with logo and monitor station target position are obtained according to the position of logo wait train
Logo image;
In the present embodiment, the initial pictures with logo can be acquired by interception or from modes such as network downloadings, if should
The type of initial pictures is video, then is parsed to the video first to obtain the picture of key frame of video, then detect parsing
The position of logo in the picture of key frame of video afterwards;If the type of the initial pictures is picture, directly detect in the picture
The position of logo.
Specifically, monitor station target position includes following sub-step S110-S150;
S110: piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
The present embodiment is generally present in the priori knowledge in four corners of video using logo, in order to reduce monitor station target
Data volume;Piecemeal, splicing are carried out to the initial pictures with logo using the method for optimal region planning;
Wherein, the optimal region planning carries out piecemeal to the initial pictures with logo, splicing includes following sub-step:
Firstly, the initial pictures with logo are divided according to preset ratio both horizontally and vertically (for example,
The ratio of horizontal direction can be 3: 5: 3 ratio, and the ratio of vertical direction is also 3: 5: 3 ratio);
Then, then by four corners of the initial pictures with logo it is stitched together, to obtain picture to be processed
Sequence.
S120: the noise of the removal sequence of pictures to be processed denoised after picture;
Wherein, the noise of the removal sequence of pictures to be processed includes following sub-step:
It, can be first to sequence of pictures to be processed according to the generally changeless priori knowledge of logo in the present embodiment
It carries out pretreatment and obtains pretreatment picture;
Secondly, extracting the marginal value of picture after pretreatment;Preferably, it can use edge detection algorithm and extract pretreatment figure
The marginal value of piece;
Then, the picture after being weighted is weighted and averaged to the marginal value of pretreatment picture;
Finally, the gray value of the picture after weighting is restored with the picture after being denoised;Specifically, can will add
The gray value of picture after power is restored between 0-255, can remove the interference that noise detects logo to greatest extent in this way.
S130: binary conversion treatment is carried out to the picture after denoising;
S140: expansion behaviour is carried out to the picture after binary conversion treatment using rectangle core (for example, it may be rectangle core of 7*7)
Make;
S150: the area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and is led to
Preset threshold value determining table target position is crossed, to realize the detection of logo.
S200: deep learning frame darknet is used to be trained to obtain logo the logo image to be trained
Identification model;
The present embodiment, the TV station symbol recognition model trained using deep learning frame darknet can be identified accurately
The logo classification of images to be recognized, obtaining the TV station symbol recognition model includes sub-step S210 and S220, in which:
S210: sample image to be trained is constructed into training dataset according to VOC standard data set format, specific method is such as
Under:
It is labeled firstly, treating trained logo image using picture annotation tool (such as: LabelImg tool), and
Generate xml document;
Then, the xml document is converted into txt file, logo training dataset is constructed based on the txt file.
S220: the sample that the logo training data is concentrated is trained and obtains TV station symbol recognition model;Specific method is such as
Under:
Firstly, the category file of training sample of the modification comprising logo image and the configuration text of the whole figure training pattern of YOLO
Part;
Secondly, being trained using the sample that the whole figure training pattern of YOLO concentrates logo training data, station identification is obtained
Other model;
Preferably, it in the present embodiment, is trained using the sample that YOLO V2 network concentrates logo training data, it can
To further increase the accuracy and speed of the detection of TV station symbol recognition model, wherein the basic model knot of the YOLO V2 network
Structure is exactly Darknet-19, includes 19 convolutional layers, 5 maximum value pond layers (max pooling layers).
Wherein, YOLO neural network is based on individual end-to-end (end-to-end) neural network, completes from sample
The output for being input to object space and classification of image, network structure include 24 convolutional layers and 2 full articulamentums.Wherein,
Convolutional layer is used to extract characteristics of image, and full articulamentum is used to future position and class probability value.The mind that the present embodiment uses
GoogLeNet sorter network structure has been used for reference through network.Unlike, inception module is not used (in GoogLeNet
Wherein one layer), but simply substituted using+3 × 3 convolutional layer of 1 × 1 convolutional layer, wherein the presence of 1 × 1 convolutional layer herein is
In order to be integrated across channel information.Meanwhile being made using mean square error (average that each data deviate the square distance sum of true value)
Carry out Optimized model parameter for loss function (loss function), i.e. S × S × (B × 5+C) dimensional vector of neural network output and true
Correspondence S × S of image × (B × 5+C) dimensional vector side and error.
For configuring neural network model .cfg file, specifically, by there is 20 classification in this present embodiment, therefore will
The value of class is changed to 20;It is modified according to only 20 class to filters value in region area level, due to
Filters=(classes+coords+1) × NUM, wherein classes indicates categorical measure, is 20 in the present embodiment,
Coords indicates that 4 coordinates tx, ty, tw, the th of BoundingBox (bounding box), coords=4 in the present embodiment, NUM are indicated
BoundingBox number of each grid cell (grid cell) prediction, NUM=5 in the present embodiment;Therefore, the present embodiment
In, the value of filters=(20+4+1) × 5=125, i.e., modified filters are 125.Further, names is changed to
The list of all logos, then saving neural network model configuration file is voc_logo.cfg.
Then training script reads ImageSet/ according to configured voc_logo.cfg file generated network object
Trainval.txt and test.txt under Main file, it is all according to the name acquiring of the sample image saved in text document
The corresponding xml document of sample image reads the sample graph under JPEGImages file according to the information saved in xml document
Picture, at the same from xml document read sample image tab area;Sample image is divided into multiple regions, then to sample
Pixel value in image in all areas is trained study.After training, the parameter that script optimizes iterative process is protected
It deposits to weight model voc_logo_8000.weight.
In addition, YOLO (You Only Look Once), it can disposably predict multiple pre-selection frames position (Box) and class
Other convolutional neural networks can be realized Target detection and identification end to end, and maximum advantage is exactly that speed is fast, usually come
It says, the essence to target detection is exactly to return, and one is realized that the neural network for returning function does not need complicated design process,
Thus, YOLO does not select the mode training network of sliding window (Silding Window), but directly selects whole figure training
Model.
S300: image to be identified is identified to obtain the classification of logo using the TV station symbol recognition model;
Specifically, the classification of for example trained logo image is CCTV1, CCTV2 logo, then the identification of TV station symbol recognition model to
Whether the image of identification is CCTV1 and CCTV2 logo.
S400: probability corresponding to each logo classification is ranked up, using the highest classification of probability as to be identified
The classification of image;
Specifically, calculating CCTV1, CCTV2 if step S300 identifies that image to be identified is CCTV1, CCTV2 logo
And the probability of CCTV1 and CCTV2 logo is obtained, the probability of CCTV1 and CCTV2 logo is ranked up, by the highest class of probability
Not as the classification of images to be recognized, and recognition result is saved into database.
The logo in video or picture file can be fast and accurately identified in TV station symbol recognition method through this embodiment
Information.
Embodiment two
A kind of specific embodiment of TV station symbol recognition system provided in an embodiment of the present invention is described below, referring to fig. 2, this is
System includes acquisition module, detection module, TV station symbol recognition model acquisition module, identification module and sorting module:
The acquisition module is used to acquire the initial pictures with logo, and the detection module is based on the logo that has
Initial pictures monitor station target position simultaneously obtains logo image to be trained according to the position of logo;
The TV station symbol recognition model obtains module using deep learning frame darknet to the logo image to be trained
It is trained to obtain TV station symbol recognition model;
The TV station symbol recognition model obtains module and identifies image to be identified to obtain platform using the TV station symbol recognition model
Target classification;
The sorting module is ranked up probability corresponding to each logo classification and makees the highest classification of probability
For the classification of images to be recognized;
Wherein, the detection module executes following operation:
Piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
Remove the sequence of pictures to be processed noise denoised after picture;
Binary conversion treatment is carried out to the picture after denoising;
Picture after checking binary conversion treatment using rectangle carries out expansive working;
The area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and by default
Threshold value determining table target position.
Further, piecemeal, splicing are carried out to the initial pictures with logo using the method for optimal region planning.
Further, the picture after the noise that the detection module removes the sequence of pictures to be processed is denoised is held
The following operation of row:
Pretreatment is carried out to sequence of pictures to be processed and obtains pretreatment picture;
Extract the marginal value of picture after pre-processing;
Picture after being weighted is weighted and averaged to the marginal value of pretreatment picture;
The gray value of picture after weighting is restored with the picture after being denoised.
Further, it includes building module and training module that the TV station symbol recognition model, which obtains module:
Sample image of the building module for will be to be trained constructs training data according to VOC standard data set format
Collection,
The training module, which is used to be trained the sample that the logo training data is concentrated, obtains TV station symbol recognition model.
Further, the building module executes following operation:
Trained logo image is treated using picture annotation tool to be labeled, and generates xml document;
The xml document is converted into txt file, logo training dataset is constructed based on the txt file.
Further, the training module executes following operation:
The category file of training sample of the modification comprising logo image and the configuration file of the whole figure training pattern of YOLO;
It is trained using the sample that the whole figure training pattern of YOLO concentrates logo training data, obtains TV station symbol recognition mould
Type.
Further, the whole figure training pattern of the YOLO is YOLO V2 network.
The specific work process of TV station symbol recognition system and the course of work of a upper embodiment are almost the same in the present embodiment,
This is no longer specifically repeated.
Beneficial effects of the present invention:
TV station symbol recognition method of the invention carries out the logo image to be trained using deep learning frame darknet
To obtain TV station symbol recognition model, the classification for identifying image to be identified using the TV station symbol recognition model to obtain logo is right for training
Probability corresponding to each logo classification is ranked up, using the highest classification of probability as the classification of images to be recognized, thus
The classification of logo and position in images to be recognized are obtained, so as to be fast and accurately identified in video or picture file
Logo information.
Those of ordinary skill in the art may be aware that the embodiment in conjunction with disclosed in the embodiment of the present invention describe it is each
Exemplary unit and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
In embodiment provided herein, it should be understood that disclosed device and method can pass through others
Mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
A kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or components can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or communication link of device or unit
It connects, can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, server or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
And storage medium above-mentioned includes: that USB flash disk, mobile hard disk, ROM, RAM, magnetic or disk etc. are various can store program code
Medium.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be subject to the protection scope in claims.
Claims (10)
1. a kind of TV station symbol recognition method, which is characterized in that this method comprises:
Initial pictures of the acquisition with logo and monitor station target position, logo figure to be trained is obtained according to the position of logo
Picture;
Deep learning frame darknet is used to be trained to obtain TV station symbol recognition model the logo image to be trained;
Image to be identified is identified using the TV station symbol recognition model to obtain the classification of logo;
Probability corresponding to each logo classification is ranked up, using the highest classification of probability as the class of images to be recognized
Not;
Wherein, monitor station target position includes following sub-step:
Piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
Remove the sequence of pictures to be processed noise denoised after picture;
Binary conversion treatment is carried out to the picture after denoising;
Picture after checking binary conversion treatment using rectangle carries out expansive working;
The area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and passes through preset threshold
It is worth determining table target position.
2. the method according to claim 1, wherein using the method for optimal region planning to first with logo
Beginning image carries out piecemeal, splicing.
3. the method according to claim 1, wherein the noise of the removal sequence of pictures to be processed obtains
Picture after to denoising includes following sub-step:
Pretreatment is carried out to sequence of pictures to be processed and obtains pretreatment picture;
Extract the marginal value of picture after pre-processing;
Picture after being weighted is weighted and averaged to the marginal value of pretreatment picture;
The gray value of picture after weighting is restored with the picture after being denoised.
4. the method according to claim 1, wherein the acquisition TV station symbol recognition model includes following sub-step:
Sample image to be trained is constructed into training dataset according to VOC standard data set format,
The sample that the logo training data is concentrated is trained and obtains TV station symbol recognition model.
5. according to the method described in claim 4, it is characterized in that, the building training dataset includes following sub-step:
Trained logo image is treated using picture annotation tool to be labeled, and generates xml document;
The xml document is converted into txt file, logo training dataset is constructed based on the txt file.
6. according to the method described in claim 4, it is characterized in that, the sample concentrated to the logo training data carries out
It includes following sub-step that training, which obtains TV station symbol recognition model:
The category file of training sample of the modification comprising logo image and the configuration file of the whole figure training pattern of YOLO;
It is trained using the sample that the whole figure training pattern of YOLO concentrates logo training data, obtains TV station symbol recognition model.
7. according to the method described in claim 6, it is characterized in that, the whole figure training pattern of the YOLO is YOLO V2 network.
8. a kind of TV station symbol recognition system, which is characterized in that the system includes that acquisition module, detection module, TV station symbol recognition model obtain
Modulus block, identification module and sorting module:
The acquisition module is used to acquire the initial pictures with logo, and the detection module is based on described initial with logo
The position of image detection logo simultaneously obtains logo image to be trained according to the position of logo;
The TV station symbol recognition model is obtained module and is carried out using deep learning frame darknet to the logo image to be trained
Training is to obtain TV station symbol recognition model;
The TV station symbol recognition model obtains module and identifies image to be identified to obtain logo using the TV station symbol recognition model
Classification;
The sorting module probability corresponding to each logo classification is ranked up and using the highest classification of probability as to
Identify the classification of image;
Wherein, the detection module executes following operation:
Piecemeal, splicing are carried out to obtain sequence of pictures to be processed to the initial pictures with logo;
Remove the sequence of pictures to be processed noise denoised after picture;
Binary conversion treatment is carried out to the picture after denoising;
Picture after checking binary conversion treatment using rectangle carries out expansive working;
The area information in the profile and profile of target is obtained according to the picture obtained after expansive working, and passes through preset threshold
It is worth determining table target position.
9. system according to claim 8, which is characterized in that it includes building module that the TV station symbol recognition model, which obtains module,
And training module:
Sample image of the building module for will be to be trained constructs training dataset according to VOC standard data set format,
The training module, which is used to be trained the sample that the logo training data is concentrated, obtains TV station symbol recognition model.
10. system according to claim 9, which is characterized in that the training module executes following operation:
The category file of training sample of the modification comprising logo image and the configuration file of the whole figure training pattern of YOLO;
It is trained using the sample that the whole figure training pattern of YOLO concentrates logo training data, obtains TV station symbol recognition model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910562188.1A CN110287888A (en) | 2019-06-26 | 2019-06-26 | A kind of TV station symbol recognition method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910562188.1A CN110287888A (en) | 2019-06-26 | 2019-06-26 | A kind of TV station symbol recognition method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110287888A true CN110287888A (en) | 2019-09-27 |
Family
ID=68006210
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910562188.1A Pending CN110287888A (en) | 2019-06-26 | 2019-06-26 | A kind of TV station symbol recognition method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110287888A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111444915A (en) * | 2020-03-26 | 2020-07-24 | 山东云缦智能科技有限公司 | Television station logo detection method based on edge detection |
CN111860472A (en) * | 2020-09-24 | 2020-10-30 | 成都索贝数码科技股份有限公司 | Television station caption detection method, system, computer equipment and storage medium |
CN111950424A (en) * | 2020-08-06 | 2020-11-17 | 腾讯科技(深圳)有限公司 | Video data processing method and device, computer and readable storage medium |
CN113055708A (en) * | 2021-01-21 | 2021-06-29 | 北京市博汇科技股份有限公司 | Program copyright protection method and device based on station caption identification |
CN113435443A (en) * | 2021-06-28 | 2021-09-24 | 中国兵器装备集团自动化研究所有限公司 | Method for automatically identifying landmark from video |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102306377A (en) * | 2011-09-21 | 2012-01-04 | 深圳市理邦精密仪器股份有限公司 | Method and device for reducing noise in ultrasound image |
CN103020650A (en) * | 2012-11-23 | 2013-04-03 | Tcl集团股份有限公司 | Station caption identifying method and device |
CN103902987A (en) * | 2014-04-17 | 2014-07-02 | 福州大学 | Station caption identifying method based on convolutional network |
CN104639952A (en) * | 2015-01-23 | 2015-05-20 | 小米科技有限责任公司 | Method and device for identifying station logo |
CN104869433A (en) * | 2015-05-20 | 2015-08-26 | 小米科技有限责任公司 | TV channel determination method, apparatus and server |
CN105868680A (en) * | 2015-11-24 | 2016-08-17 | 乐视致新电子科技(天津)有限公司 | Channel logo classification method and apparatus |
CN106488313A (en) * | 2016-10-31 | 2017-03-08 | Tcl集团股份有限公司 | A kind of TV station symbol recognition method and system |
CN106507188A (en) * | 2016-11-25 | 2017-03-15 | 南京中密信息科技有限公司 | A kind of video TV station symbol recognition device and method of work based on convolutional neural networks |
CN106845442A (en) * | 2017-02-15 | 2017-06-13 | 杭州当虹科技有限公司 | A kind of station caption detection method based on deep learning |
CN107330027A (en) * | 2017-06-23 | 2017-11-07 | 中国科学院信息工程研究所 | A kind of Weakly supervised depth station caption detection method |
CN109003390A (en) * | 2018-06-29 | 2018-12-14 | 深圳和而泰数据资源与云技术有限公司 | A kind of commodity recognition method, self-service machine and computer readable storage medium |
-
2019
- 2019-06-26 CN CN201910562188.1A patent/CN110287888A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102306377A (en) * | 2011-09-21 | 2012-01-04 | 深圳市理邦精密仪器股份有限公司 | Method and device for reducing noise in ultrasound image |
CN103020650A (en) * | 2012-11-23 | 2013-04-03 | Tcl集团股份有限公司 | Station caption identifying method and device |
CN103902987A (en) * | 2014-04-17 | 2014-07-02 | 福州大学 | Station caption identifying method based on convolutional network |
CN104639952A (en) * | 2015-01-23 | 2015-05-20 | 小米科技有限责任公司 | Method and device for identifying station logo |
CN104869433A (en) * | 2015-05-20 | 2015-08-26 | 小米科技有限责任公司 | TV channel determination method, apparatus and server |
CN105868680A (en) * | 2015-11-24 | 2016-08-17 | 乐视致新电子科技(天津)有限公司 | Channel logo classification method and apparatus |
CN106488313A (en) * | 2016-10-31 | 2017-03-08 | Tcl集团股份有限公司 | A kind of TV station symbol recognition method and system |
CN106507188A (en) * | 2016-11-25 | 2017-03-15 | 南京中密信息科技有限公司 | A kind of video TV station symbol recognition device and method of work based on convolutional neural networks |
CN106845442A (en) * | 2017-02-15 | 2017-06-13 | 杭州当虹科技有限公司 | A kind of station caption detection method based on deep learning |
CN107330027A (en) * | 2017-06-23 | 2017-11-07 | 中国科学院信息工程研究所 | A kind of Weakly supervised depth station caption detection method |
CN109003390A (en) * | 2018-06-29 | 2018-12-14 | 深圳和而泰数据资源与云技术有限公司 | A kind of commodity recognition method, self-service machine and computer readable storage medium |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111444915A (en) * | 2020-03-26 | 2020-07-24 | 山东云缦智能科技有限公司 | Television station logo detection method based on edge detection |
CN111950424A (en) * | 2020-08-06 | 2020-11-17 | 腾讯科技(深圳)有限公司 | Video data processing method and device, computer and readable storage medium |
CN111950424B (en) * | 2020-08-06 | 2023-04-07 | 腾讯科技(深圳)有限公司 | Video data processing method and device, computer and readable storage medium |
CN111860472A (en) * | 2020-09-24 | 2020-10-30 | 成都索贝数码科技股份有限公司 | Television station caption detection method, system, computer equipment and storage medium |
CN113055708A (en) * | 2021-01-21 | 2021-06-29 | 北京市博汇科技股份有限公司 | Program copyright protection method and device based on station caption identification |
CN113055708B (en) * | 2021-01-21 | 2023-04-11 | 北京市博汇科技股份有限公司 | Program copyright protection method and device based on station caption identification |
CN113435443A (en) * | 2021-06-28 | 2021-09-24 | 中国兵器装备集团自动化研究所有限公司 | Method for automatically identifying landmark from video |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110287888A (en) | A kind of TV station symbol recognition method and system | |
CN113160192B (en) | Visual sense-based snow pressing vehicle appearance defect detection method and device under complex background | |
CN103049763B (en) | Context-constraint-based target identification method | |
WO2020139743A1 (en) | Computer-executed method and apparatus for assessing vehicle damage | |
CN108846401A (en) | Commodity detect terminal, method, system and computer equipment, readable medium | |
CN111813997B (en) | Intrusion analysis method, device, equipment and storage medium | |
CN109767422A (en) | Pipe detection recognition methods, storage medium and robot based on deep learning | |
CN109583483A (en) | A kind of object detection method and system based on convolutional neural networks | |
CN112016464A (en) | Method and device for detecting face shielding, electronic equipment and storage medium | |
CN110390308B (en) | Video behavior identification method based on space-time confrontation generation network | |
CN103295009B (en) | Based on the license plate character recognition method of Stroke decomposition | |
CN113657274B (en) | Table generation method and device, electronic equipment and storage medium | |
CN114758288A (en) | Power distribution network engineering safety control detection method and device | |
CN110910445B (en) | Object size detection method, device, detection equipment and storage medium | |
CN111382808A (en) | Vehicle detection processing method and device | |
CN110349167A (en) | A kind of image instance dividing method and device | |
CN112149533A (en) | Target detection method based on improved SSD model | |
CN112488229A (en) | Domain self-adaptive unsupervised target detection method based on feature separation and alignment | |
CN109740553A (en) | A kind of image, semantic segmentation data screening method and system based on identification | |
CN110516676A (en) | A kind of bank's card number identifying system based on image procossing | |
CN112861970A (en) | Fine-grained image classification method based on feature fusion | |
CN113901911B (en) | Image recognition method, image recognition device, model training method, model training device, electronic equipment and storage medium | |
CN116385430A (en) | Machine vision flaw detection method, device, medium and equipment | |
CN111275694B (en) | Attention mechanism guided progressive human body division analysis system and method | |
CN109740554A (en) | A kind of road edge line recognition methods and system |
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 |
Application publication date: 20190927 |
|
RJ01 | Rejection of invention patent application after publication |