CN105678322A - Sample labeling method and apparatus - Google Patents
Sample labeling method and apparatus Download PDFInfo
- Publication number
- CN105678322A CN105678322A CN201511025974.6A CN201511025974A CN105678322A CN 105678322 A CN105678322 A CN 105678322A CN 201511025974 A CN201511025974 A CN 201511025974A CN 105678322 A CN105678322 A CN 105678322A
- Authority
- CN
- China
- Prior art keywords
- target
- image
- marked
- detector model
- sample
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a sample labeling method and apparatus. The sample labeling method includes the steps: detecting an image to be labeled by means of a detector model, and detecting and labeling a target in the image to be labeled; and determining the category of the target by means of a recognizer, and determining the category belonging to the target. The sample labeling method and apparatus have no more need of labeling through manual work, thus significantly improving the operating speed for sample labeling.
Description
Technical field
The present embodiments relate to computer vision technique, particularly relate to a kind of sample mask method and device.
Background technology
Target detection and identification is the direction, forward position received much concern in technical field of computer vision in recent years, and it detects from the image sequence comprising target and identifies target. Therefore, needing substantial amounts of sample when the target in image is detected and identified, these samples come from the image marking target and classification.
In prior art, when the target in image and classification are labeled, completing generally by artificial mode, namely by drawing rectangle frame around the artificial target comprised in the picture, then the mode providing this target generic completes mark.
Prior art is primarily present problems with: first, it is determined that target and provide classification and all completed by same people, it is easy to causes that two kinds of operations are obscured, affects operating speed; Second, when number of samples increases, operating speed is relatively low. Therefore, prior art has a drawback in that the mark bothersome effort of work, and operating speed is relatively low.
Summary of the invention
In view of this, the embodiment of the present invention provides a kind of sample mask method and device, to improve the operating speed of sample mark.
First aspect, embodiments provides a kind of sample mask method, and described method includes:
Utilize detector model that image to be marked is detected, detect and mark out the target in described image to be marked;
Utilize detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
Second aspect, the embodiment of the present invention additionally provides a kind of sample annotation equipment, and described device includes:
Target labeling module, is used for utilizing detector model that image to be marked is detected, detects and mark out the target in described image to be marked;
Kind judging module, is used for utilizing detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
The technical scheme of the embodiment of the present invention, by utilizing detector model that image to be marked is detected, and mark out the target in described image to be marked, utilize detector model that described target is carried out kind judging, determine the classification belonging to described target, it is no longer necessary to manually be labeled, hence it is evident that improve the operating speed of sample mark.
Accompanying drawing explanation
Fig. 1 is the flow chart of a kind of sample mask method that the embodiment of the present invention one provides;
Fig. 2 is the flow chart of a kind of sample mask method that the embodiment of the present invention two provides;
Fig. 3 is the flow chart of a kind of sample mask method that the embodiment of the present invention three provides;
Fig. 4 is the flow chart of a kind of sample mask method that the embodiment of the present invention four provides;
Fig. 5 is the structural representation of a kind of sample annotation equipment that the embodiment of the present invention five provides.
Detailed description of the invention
Below in conjunction with drawings and Examples, the present invention is described in further detail. It is understood that specific embodiment described herein is used only for explaining the present invention, but not limitation of the invention. It also should be noted that, for the ease of describing, accompanying drawing illustrate only part related to the present invention but not full content.
Embodiment one
Fig. 1 is the flow chart of a kind of sample mask method that the embodiment of the present invention one provides, and the present embodiment is applicable to the situation that the sample in object detection and recognition is labeled, and the method can be performed by computer, specifically includes as follows:
S110, utilizes detector model that image to be marked is detected, and detects and mark out the target in described image to be marked.
The present embodiment, when image to be marked is labeled, depends on the trained detector model completed. Wherein, detector model is the image that a pair comprises target, and target can set according to demand.
Detector model and image to be marked is utilized to carry out matching detection, it is determined that whether image to be marked to comprise target, when image to be marked comprises target, marks out the target in image to be marked by the form of rectangle frame.
Wherein, utilize detector model that image to be marked is detected, detect and mark out the target in described image to be marked and preferably include:
According to detector model, it is determined that scan box;
Utilize described scan box that described image to be marked is scanned, the imagery exploitation detector model scanned is detected, it is determined that whether the image scanned is target;
Will determine as the scanogram of target to mark out.
Size according to detector model, determine the size of scan box, utilize described scan box that described image to be marked is scanned, and the image scanned is mated with detector model, when the image scanned and detector Model Matching success, determining in image to be marked and comprise target, the image scanned is target, marks out by the form of the image rectangle frame scanned. Image to be marked is scanned, it is possible to determine in image to be marked whether comprise target accurately, it is to avoid omit the target in image to be marked by the scan box with detector model size.
S120, utilizes detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
The present embodiment, when target is carried out kind judging, depends on the trained detector model completed.
Detector model includes the multiple classifications belonging to identical target (if detector model can be people, and detector model can for everyone concrete name), the target in the detector model image to be marked to marking out is utilized to carry out kind judging, the target being about in the image to be marked marked out is mated with detector model, when the match is successful, it is determined that the classification belonging to described detector model is the specific category belonging to described target.
The technical scheme of the present embodiment, by utilizing detector model that image to be marked is detected, and mark out the target in described image to be marked, utilize detector model that described target is carried out kind judging, determine the classification belonging to described target, it is no longer necessary to manually be labeled, hence it is evident that improve the operating speed of sample mark.
On the basis of technique scheme, before utilizing detector model that image to be marked is detected, it is also preferred that including:
The first sample marking out target is trained and learns, obtains detector model.
The second sample providing classification is trained and learns, is identified device model.
Wherein, target can set according to demand, and the first sample is the substantial amounts of image marking out target, and the second sample is the substantial amounts of image providing classification described in target. Utilize Adaboost, SVM (SupportVectorMachine, support vector machine), RCNN (RegionConvolutionalNeuralNetworks, convolutional neural networks based on region) etc. Many Detection the first sample marking out target is trained and learns, obtain detector model; Utilize the multiple recognizers such as Adaboost, SVM, CNN (ConvolutionalNeuralNetworks, convolutional neural networks) that the second sample having been given by classification is trained and is learnt, be identified device model. Detection model and detector model by being trained obtaining to substantial amounts of sample are more accurate.
Wherein, Adaboost is a kind of iterative algorithm, and its core concept is the Weak Classifier different for the training of same training set, then these weak classifier set is got up, constitutes a higher final grader. SVM method is by a nonlinear mapping, sample space is mapped in a higher-dimension or even infinite dimensional feature space so that the problem of the linear separability that the problem of Nonlinear separability is converted in feature space in original sample space. CNN is the one of artificial neural network, its weights share network structure so as to be more closely similar to biological neural network, reduce the complexity of network model, decrease the quantity of weights, what this advantage showed when the input of network is multidimensional image becomes apparent from, make the image can directly as the input of network, convolutional neural networks is a multilayer perceptron for identifying two-dimensional shapes and particular design, and the deformation of translation, proportional zoom, inclination or his form common is had height invariance by this network structure. With tradition CNN the difference is that, multiple regions are also fed into network while picture is sent into network by RCNN.
Embodiment two
Fig. 2 is the flow chart of a kind of sample mask method that the embodiment of the present invention two provides, the present embodiment in embodiment on the basis of embodiment one, described target being stored in the first data base for after mark personnel are labeled, add and extract target and store the content of target, specifically include as follows:
S210, utilizes detector model that image to be marked is detected, and detects and mark out the target in described image to be marked.
S220, extracts described target, and described target is stored in the first data base and supplies mark personnel to be labeled.
According to the image to be marked marking out target, determine described target pixel coordinate in described image to be marked, according to described target pixel coordinate in described image to be marked, described image to be marked extracts described target (by the little image cropping of target in described image to be marked out), and described target is stored in the first data base, it is easy to mark personnel described target is labeled, namely determined that whether the target marked out in image to be marked is correct target by mark personnel, namely be conducive to mark personnel that the target marked out in image to be marked is corrected.
S230, utilizes detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
The technical scheme of the present embodiment, on the basis of embodiment one, after the target in marking out described image to be marked, extracts described target, and described target is stored in the first data base, be conducive to mark personnel that described target is corrected.
On the basis of technique scheme, described target being stored in the first data base for after mark personnel are labeled, it is also preferred that including:
Receive the mark personnel mark to described target, the annotation results of mark personnel is stored in the first Sample Storehouse.
The target being stored in the first data base can be labeled by mark personnel, determine that whether the target marked out in image to be marked is correct target, after receiving the mark personnel mark to described target, the annotation results (being labeled as the target of correct target by mark personnel) of mark personnel is stored in the first Sample Storehouse as the first sample, be conducive to follow-up being trained according to the first sample in the first Sample Storehouse, obtain detector model, target after correcting by mark personnel is as sample, training obtains detector model, thus the detector model obtained is more stable, accurately.
Embodiment three
Fig. 3 is the flow chart of a kind of sample mask method that the embodiment of the present invention three provides, and the present embodiment in embodiment, after determining the classification belonging to described target, adds the content of storage target and generic, specifically includes as follows on the basis of embodiment one:
S310, utilizes detector model that image to be marked is detected, and detects and mark out the target in described image to be marked.
S320, utilizes detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
S330, is stored in described target and generic in the second data base and carries out classification setting for mark personnel.
The target marked out in image to be marked and the described target generic determined are stored in the second data base, in order to the mark personnel target to determining and generic carry out classification setting, correct incorrect target generic.
The technical scheme of the present embodiment, on the basis of embodiment one, after determining the classification belonging to described target, described target and described classification are stored in the second data base and carry out classification setting for mark personnel, it is simple to described target generic is corrected by staff.
On the basis of technique scheme, described target and generic being stored in the second data base for after mark personnel carry out classification setting, it is also preferred that including:
Receive the mark personnel setting to described target generic, and be stored in arranging result in the second Sample Storehouse.
Show the target in described second data base, and the interface of setting is provided, receive the mark personnel setting to described target generic, the classification of mark personnel is arranged result and is stored in the second Sample Storehouse as the second sample, being conducive to follow-up being trained according to the second sample in the second Sample Storehouse, obtain detector model, the classification after correcting by mark personnel arranges result as sample, training is identified device model, thus the detector model obtained is more stable, accurate.
Embodiment four
Fig. 4 is the flow chart of a kind of sample mask method that the embodiment of the present invention four provides, detector model is updated as the first sample with the image to be marked marking out target by the present embodiment based on positive feedback theory, as the second sample, detector model is updated by the target arranging out classification, specifically includes as follows:
S410, is trained the first sample marking out target and learns, obtaining detector model.
S420, utilizes detector model that image to be marked is detected, and detects and mark out the target in described image to be marked.
S430, extracts described target, and described target is stored in the first data base and supplies mark personnel to be labeled.
S440, receives the mark personnel mark to described target, is stored in the first Sample Storehouse by the annotation results of mark personnel, performs S410.
The annotation results of mark personnel is stored in the first Sample Storehouse as the first sample, it is easy to be trained obtaining detector model according to the first sample successfully marking out target, the detector model updated by such iteration carrys out label target, can significantly reduce and can filter the image to be marked without target, accurate target position can also be obtained, thus significantly reducing the workload of target mark (namely rectangle frame is drawn). Moreover, after utilizing the corrected new sample learning of mark personnel to obtain new detector model, more stable, detector model accurately will be obtained. This positive feedback theory can help optimized detector model and sample initialization result, significantly reduces artificial and consuming time, promotes the operating speed of sample mark.
S450, is trained the second sample providing classification and learns, being identified device model.
S460, utilizes detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
S470, is stored in described target and generic in the second data base and carries out classification setting for mark personnel.
S480, receives the mark personnel setting to described target generic, and is stored in the second Sample Storehouse by arranging result, performs S450.
The classification of mark personnel is arranged result and is stored in the second Sample Storehouse as the second sample, it is easy to the second sample according to correctly arranging classification be trained, it is identified device model, the detector model updated by such iteration carries out kind judging, classification correct greatly can be provided, thus significantly reducing the workload setting classification. Moreover, after utilizing the corrected classification of mark personnel to obtain new detector model, more stable, classification accurately will be obtained. This positive feedback theory can help Statistical error device model and sample to initialize category result, significantly reduces artificial and consuming time.
The technical scheme of the present embodiment, it is possible to significantly reduce the operating speed for sample mark required in Target detection and identification, save man power and material. Through practice, adopt the technical scheme of the present embodiment, it is possible in tradition 500 operating speeds for each person every day are risen to 2000 for each person every day.
Embodiment five
Fig. 5 is the structural representation of a kind of sample annotation equipment that the embodiment of the present invention five provides, as it is shown in figure 5, the sample annotation equipment described in the present embodiment includes: target labeling module 510 and kind judging module 520.
Wherein, target labeling module 510 is used for utilizing detector model that image to be marked is detected, and detects and mark out the target in described image to be marked;
Kind judging module 520 is used for utilizing detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
Preferably, this sample annotation equipment also includes:
Detector training module, for, before utilizing detector model that image to be marked is detected, being trained the first sample marking out target and learn, obtaining detector model.
Evaluator training module, for the second sample providing classification is trained and is learnt, is identified device model.
Preferably, this sample annotation equipment also includes:
Target memory module, after the target in marking out described image to be marked, extracts described target, and described target is stored in the first data base and supplies mark personnel to be labeled.
Preferably, this sample annotation equipment also includes:
Mark receiver module, for described target being stored in the first data base for, after mark personnel are labeled, receiving the mark personnel mark to described target, being stored in the annotation results of mark personnel in the first Sample Storehouse.
Preferably, this sample annotation equipment also includes:
Target classification memory module, for, after determining the classification belonging to described target, being stored in described target and generic in the second data base and carry out classification setting for mark personnel.
Preferably, this sample annotation equipment also includes:
Classification receiver module, for being stored in the second data base for, after mark personnel carry out classification setting, receiving the mark personnel setting to described target generic, and be stored in arranging result in the second Sample Storehouse by described target and generic.
Preferably, described target labeling module includes:
Scan box determines unit, for according to detector model, it is determined that scan box;
Target determination unit, is used for utilizing described scan box that described image to be marked is scanned, is detected by the imagery exploitation detector model scanned, it is determined that whether the image scanned is target;
Target mark unit, the scanogram for will determine as target marks out.
The said goods can perform the method that any embodiment of the present invention provides, and possesses the corresponding functional module of execution method and beneficial effect.
Note, above are only presently preferred embodiments of the present invention and institute's application technology principle. It will be appreciated by those skilled in the art that and the invention is not restricted to specific embodiment described here, various obvious change can be carried out for a person skilled in the art, readjust and substitute without departing from protection scope of the present invention. Therefore, although the present invention being described in further detail by above example, but the present invention is not limited only to above example, when without departing from present inventive concept, other Equivalent embodiments more can also be included, and the scope of the present invention is determined by appended right.
Claims (6)
1. a sample mask method, it is characterised in that described method includes:
Utilize detector model that image to be marked is detected, detect and mark out the target in described image to be marked;
Utilize detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
2. method according to claim 1, it is characterised in that before utilizing detector model that image to be marked is detected, also include:
The first sample marking out target is trained and learns, obtains detector model.
The second sample providing classification is trained and learns, is identified device model.
3. method according to claim 1 and 2, it is characterised in that utilize detector model that image to be marked is detected, detects and marks out the target in described image to be marked and include:
According to detector model, it is determined that scan box;
Utilize described scan box that described image to be marked is scanned, the imagery exploitation detector model scanned is detected, it is determined that whether the image scanned is target;
Will determine as the scanogram of target to mark out.
4. a sample annotation equipment, it is characterised in that described device includes:
Target labeling module, is used for utilizing detector model that image to be marked is detected, detects and mark out the target in described image to be marked;
Kind judging module, is used for utilizing detector model that described target is carried out kind judging, it is determined that the classification belonging to described target.
5. device according to claim 4, it is characterised in that also include:
Detector training module, for, before utilizing detector model that image to be marked is detected, being trained the first sample marking out target and learn, obtaining detector model.
Evaluator training module, for the second sample providing classification is trained and is learnt, is identified device model.
6. the device according to claim 4 or 5, it is characterised in that described target labeling module includes:
Scan box determines unit, for according to detector model, it is determined that scan box;
Target determination unit, is used for utilizing described scan box that described image to be marked is scanned, is detected by the imagery exploitation detector model scanned, it is determined that whether the image scanned is target;
Target mark unit, the scanogram for will determine as target marks out.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511025974.6A CN105678322A (en) | 2015-12-31 | 2015-12-31 | Sample labeling method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511025974.6A CN105678322A (en) | 2015-12-31 | 2015-12-31 | Sample labeling method and apparatus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105678322A true CN105678322A (en) | 2016-06-15 |
Family
ID=56298183
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201511025974.6A Pending CN105678322A (en) | 2015-12-31 | 2015-12-31 | Sample labeling method and apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105678322A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106845440A (en) * | 2017-02-13 | 2017-06-13 | 山东万腾电子科技有限公司 | A kind of augmented reality image processing method and system |
| CN107341436A (en) * | 2016-08-19 | 2017-11-10 | 北京市商汤科技开发有限公司 | Gestures detection network training, gestures detection and control method, system and terminal |
| CN107481327A (en) * | 2017-09-08 | 2017-12-15 | 腾讯科技（深圳）有限公司 | On the processing method of augmented reality scene, device, terminal device and system |
| CN107562838A (en) * | 2017-08-24 | 2018-01-09 | 百度在线网络技术（北京）有限公司 | A kind of method and apparatus for image information collecting |
| CN107886105A (en) * | 2016-09-30 | 2018-04-06 | 法乐第(北京)网络科技有限公司 | A kind of annotation equipment of image |
| CN107886104A (en) * | 2016-09-30 | 2018-04-06 | 法乐第(北京)网络科技有限公司 | A kind of mask method of image |
| CN108491930A (en) * | 2018-03-23 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of processing method and data processing equipment of sample data |
| CN108717789A (en) * | 2018-06-28 | 2018-10-30 | 深圳市金溢科技股份有限公司 | A kind of the acquisition mask method and device of vehicle sample |
| CN108898188A (en) * | 2018-07-06 | 2018-11-27 | 四川奇迹云科技有限公司 | A kind of image data set aid mark system and method |
| CN109190676A (en) * | 2018-08-06 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | model training method, device, equipment and storage medium |
| CN109857878A (en) * | 2018-12-27 | 2019-06-07 | 深兰科技（上海）有限公司 | Article mask method and device, electronic equipment and storage medium |
| CN109903308A (en) * | 2017-12-08 | 2019-06-18 | 百度在线网络技术（北京）有限公司 | For obtaining the method and device of information |
| CN110058756A (en) * | 2019-04-19 | 2019-07-26 | 北京朗镜科技有限责任公司 | A kind of mask method and device of image pattern |
| CN110135407A (en) * | 2018-02-09 | 2019-08-16 | 北京世纪好未来教育科技有限公司 | Sample mask method and computer storage medium |
| CN110267040A (en) * | 2019-06-27 | 2019-09-20 | 国网山东省电力公司建设公司 | A kind of method for compressing image based on video flow detection |
| CN110334633A (en) * | 2019-06-27 | 2019-10-15 | 北京御航智能科技有限公司 | Identification inspection data and the method, apparatus and storage medium for updating identification model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101200252A (en) * | 2007-12-10 | 2008-06-18 | 中国科学院计算技术研究所 | Target place optimized dispatching method and system |
| CN102289686A (en) * | 2011-08-09 | 2011-12-21 | 北京航空航天大学 | Method for identifying classes of moving targets based on transfer learning |
| CN104217225A (en) * | 2014-09-02 | 2014-12-17 | 中国科学院自动化研究所 | A visual target detection and labeling method |
| CN104463249A (en) * | 2014-12-09 | 2015-03-25 | 西北工业大学 | Remote sensing image airport detection method based on weak supervised learning frame |
| CN104700099A (en) * | 2015-03-31 | 2015-06-10 | 百度在线网络技术（北京）有限公司 | Method and device for recognizing traffic signs |
-
2015
- 2015-12-31 CN CN201511025974.6A patent/CN105678322A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101200252A (en) * | 2007-12-10 | 2008-06-18 | 中国科学院计算技术研究所 | Target place optimized dispatching method and system |
| CN102289686A (en) * | 2011-08-09 | 2011-12-21 | 北京航空航天大学 | Method for identifying classes of moving targets based on transfer learning |
| CN104217225A (en) * | 2014-09-02 | 2014-12-17 | 中国科学院自动化研究所 | A visual target detection and labeling method |
| CN104463249A (en) * | 2014-12-09 | 2015-03-25 | 西北工业大学 | Remote sensing image airport detection method based on weak supervised learning frame |
| CN104700099A (en) * | 2015-03-31 | 2015-06-10 | 百度在线网络技术（北京）有限公司 | Method and device for recognizing traffic signs |
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107341436A (en) * | 2016-08-19 | 2017-11-10 | 北京市商汤科技开发有限公司 | Gestures detection network training, gestures detection and control method, system and terminal |
| CN107341436B (en) * | 2016-08-19 | 2019-02-22 | 北京市商汤科技开发有限公司 | Gestures detection network training, gestures detection and control method, system and terminal |
| CN107886105A (en) * | 2016-09-30 | 2018-04-06 | 法乐第(北京)网络科技有限公司 | A kind of annotation equipment of image |
| CN107886104A (en) * | 2016-09-30 | 2018-04-06 | 法乐第(北京)网络科技有限公司 | A kind of mask method of image |
| CN106845440A (en) * | 2017-02-13 | 2017-06-13 | 山东万腾电子科技有限公司 | A kind of augmented reality image processing method and system |
| CN106845440B (en) * | 2017-02-13 | 2020-04-10 | 山东万腾电子科技有限公司 | Augmented reality image processing method and system |
| CN107562838A (en) * | 2017-08-24 | 2018-01-09 | 百度在线网络技术（北京）有限公司 | A kind of method and apparatus for image information collecting |
| US11875467B2 (en) | 2017-09-08 | 2024-01-16 | Tencent Technology (Shenzhen) Company Ltd | Processing method for combining a real-world environment with virtual information according to a video frame difference value to provide an augmented reality scene, terminal device, system, and computer storage medium |
| US11410415B2 (en) | 2017-09-08 | 2022-08-09 | Tencent Technology (Shenzhen) Company Ltd | Processing method for augmented reality scene, terminal device, system, and computer storage medium |
| WO2019047789A1 (en) * | 2017-09-08 | 2019-03-14 | 腾讯科技（深圳）有限公司 | Augmented reality scene related processing method, terminal device and system and computer storage medium |
| CN107481327B (en) * | 2017-09-08 | 2019-03-15 | 腾讯科技（深圳）有限公司 | About the processing method of augmented reality scene, device, terminal device and system |
| CN107481327A (en) * | 2017-09-08 | 2017-12-15 | 腾讯科技（深圳）有限公司 | On the processing method of augmented reality scene, device, terminal device and system |
| CN109903308B (en) * | 2017-12-08 | 2021-02-26 | 百度在线网络技术（北京）有限公司 | Method and device for acquiring information |
| CN109903308A (en) * | 2017-12-08 | 2019-06-18 | 百度在线网络技术（北京）有限公司 | For obtaining the method and device of information |
| CN110135407A (en) * | 2018-02-09 | 2019-08-16 | 北京世纪好未来教育科技有限公司 | Sample mask method and computer storage medium |
| CN108491930A (en) * | 2018-03-23 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of processing method and data processing equipment of sample data |
| CN108491930B (en) * | 2018-03-23 | 2022-04-15 | 腾讯科技（深圳）有限公司 | Sample data processing method and data processing device |
| CN108717789A (en) * | 2018-06-28 | 2018-10-30 | 深圳市金溢科技股份有限公司 | A kind of the acquisition mask method and device of vehicle sample |
| CN108898188A (en) * | 2018-07-06 | 2018-11-27 | 四川奇迹云科技有限公司 | A kind of image data set aid mark system and method |
| CN109190676A (en) * | 2018-08-06 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | model training method, device, equipment and storage medium |
| CN109190676B (en) * | 2018-08-06 | 2022-11-08 | 百度在线网络技术（北京）有限公司 | Model training method, device, equipment and storage medium for image recognition |
| CN109857878A (en) * | 2018-12-27 | 2019-06-07 | 深兰科技（上海）有限公司 | Article mask method and device, electronic equipment and storage medium |
| CN109857878B (en) * | 2018-12-27 | 2021-08-20 | 深兰科技（上海）有限公司 | Article labeling method and device, electronic equipment and storage medium |
| CN110058756A (en) * | 2019-04-19 | 2019-07-26 | 北京朗镜科技有限责任公司 | A kind of mask method and device of image pattern |
| CN110058756B (en) * | 2019-04-19 | 2021-03-02 | 北京朗镜科技有限责任公司 | Image sample labeling method and device |
| CN110267040A (en) * | 2019-06-27 | 2019-09-20 | 国网山东省电力公司建设公司 | A kind of method for compressing image based on video flow detection |
| CN110334633A (en) * | 2019-06-27 | 2019-10-15 | 北京御航智能科技有限公司 | Identification inspection data and the method, apparatus and storage medium for updating identification model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105678322A (en) | Sample labeling method and apparatus | |
| Yuliang et al. | Detecting curve text in the wild: New dataset and new solution | |
| CN110070536B (en) | Deep learning-based PCB component detection method | |
| CN107016387B (en) | Method and device for identifying label | |
| US20200134377A1 (en) | Logo detection | |
| CN110647829A (en) | Bill text recognition method and system | |
| CN107403128B (en) | Article identification method and device | |
| CN114155527A (en) | Scene text recognition method and device | |
| CN103903013A (en) | Optimization algorithm of unmarked flat object recognition | |
| CN111444850B (en) | Picture detection method and related device | |
| CN107403179B (en) | Registration method and device for article packaging information | |
| CN115116074A (en) | Handwritten character recognition and model training method and device | |
| CN114648771A (en) | Character recognition method, electronic device and computer readable storage medium | |
| CN113947714A (en) | Multi-mode collaborative optimization method and system for video monitoring and remote sensing | |
| CN115810197A (en) | Multi-mode electric power form recognition method and device | |
| CN111177308B (en) | Emotion recognition method for text content | |
| CN113628113A (en) | Image splicing method and related equipment thereof | |
| US20230110558A1 (en) | Systems and methods for detecting objects | |
| CN111753618A (en) | Image recognition method and device, computer equipment and computer readable storage medium | |
| CN114049648B (en) | Engineering drawing text detection and recognition method, device and system | |
| CN112232272B (en) | Pedestrian recognition method by fusing laser and visual image sensor | |
| CN113822275B (en) | Image language identification method and related equipment thereof | |
| CN106648171B (en) | A kind of interactive system and method based on lettering pen | |
| Wang et al. | Recognition and distance estimation of an irregular object in package sorting line based on monocular vision | |
| CN114022684A (en) | Human body posture estimation method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160615 |
|
| RJ01 | Rejection of invention patent application after publication |