CN107480600A - A kind of gesture identification method based on depth convolutional neural networks - Google Patents
A kind of gesture identification method based on depth convolutional neural networks Download PDFInfo
- Publication number
- CN107480600A CN107480600A CN201710597440.3A CN201710597440A CN107480600A CN 107480600 A CN107480600 A CN 107480600A CN 201710597440 A CN201710597440 A CN 201710597440A CN 107480600 A CN107480600 A CN 107480600A
- Authority
- CN
- China
- Prior art keywords
- mrow
- convolutional neural
- neural networks
- hand
- training
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/107—Static hand or arm
- G06V40/113—Recognition of static hand signs
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of gesture identification method based on depth convolutional neural networks, including：(1) division of edge detection process and sample set is carried out to the sample image of training set；（2）Build depth convolutional neural networks；（3）Determine activation primitive and loss function；（4）Train deep neural network；（5）Gesture identification is realized according to the depth convolutional neural networks after training：Its step includes：A) hand-type image is extracted from gesture data to be identified；B) hand-type image is subjected to rim detection and size normalized；C) normalized hand-type image is input in depth convolutional neural networks, the output valve according to output layer judges the ownership class of current gesture.The present invention uses multiple down-sampling technique construction depth convolutional neural networks, and carries out the training of neutral net as activation primitive using hyperbolic tangent function, can not only improve the efficiency of gesture identification, and can improve the accuracy rate of gesture identification.
Description
Technical field
The present invention relates to living things feature recognition field, more particularly to a kind of gesture identification based on depth convolutional neural networks
Method.
Background technology
Living things feature recognition is one of key technologies in field such as video monitoring, safety certification.Biological characteristic can be divided into
Physiological characteristic and behavioural characteristic.Physiological characteristic mainly includes face, fingerprint and iris etc., and behavioural characteristic then includes gait, gesture
Deng.Typically the recognition methods based on physiological characteristic has fingerprint recognition, palm shape and outline identification, recognition of face, and iris is known
Not etc..Fingerprint recognition is one of current most widely used personal identification method based on biological characteristic.Fingerprint recognition has skill
The advantages that art is ripe, and cost is cheap.Its shortcoming is contact, has the property invaded, and the problem of health aspect, while fingerprint be present
It is and easy to wear.Face recognition technology is a very active in recent years research field, have intuitive it is good, conveniently, friend
Well, the advantages of being easily accepted.Recognition of face is contactless, passive discerning, it is not necessary to which people's cooperates with one's own initiative；But shortcoming
The influence of illumination, visual angle, shelter, environment, expression etc. is susceptible to, causes identification difficult.The safety of iris feature identification
Degree and precision are very high, but collection apparatus process is extremely difficult.The identity recognizing technology of Behavior-based control feature, common are
Gait Recognition and gesture identification.The input of Gait Recognition is the sequence of video images of one section of walking, and data volume is very big, causes to calculate
Complexity is high, deals with relatively difficult.And important component of the gesture identification as non-contact type human-machine interaction, it is big at present
In terms of most researchers are primarily focused on the final identification of gesture, it will usually gesture background is simplified, and in the single back of the body
Gesture is split using the algorithm studied under scape, then passed through the implication that gesture is expressed using conventional recognition methods
Network analysis comes out.
The content of the invention
In view of the above-mentioned deficiencies in the prior art, it is an object of the present invention to provide a kind of gesture based on depth convolutional neural networks
Recognition methods.
The purpose of the present invention is achieved through the following technical solutions：A kind of gesture identification based on deep neural network
Method, comprise the following steps：
(1) division of edge detection process and sample set is carried out to the sample image of training set：First to the hand of training
Power set image carries out hand-type detection and edge detection process, and by the hand-type Image Adjusting of extraction to unified size；It will locate in advance
Data after reason are divided into training sample set and checking sample set；
(2) depth convolutional neural networks are built：If I and O are respectively the input layer and output layer of depth convolutional neural networks,
Hidden layer between I and O is H1, H2 ..., Hn.Wherein, the hand-type image that input layer obtains for step (1), output layer one
The gesture feature vector that individual length is N, hidden layer use multiple down-sampling technology, it is allowed to have between the piecemeal of down-sampling overlapping；
(3) activation primitive and loss function are determined：The non-linear hyperbolic tan shown in formula (1) is selected as nerve
The activation primitive of member；Select the loss function shown in formula (2).Wherein, n is the number of sample in training set, and x is hand-type figure
Picture, y are the gesture feature vector corresponding to x, and θ is parameter vector；
(4) deep neural network is trained：Concentrated from training sample and choose m training sample, using steepest descent method
Calculate gradient；Then, verified using checking sample set, when accuracy exceedes predetermined threshold value 99.5%, terminate training, from
And obtain with the deep neural network for determining weight w and bias term b；
(5) gesture identification is realized according to the depth convolutional neural networks after training：Its step includes：A) from gesture to be identified
Extracting data hand-type image；B) hand-type image is subjected to rim detection and size normalized；C) hand-type image is inputted
Into depth convolutional neural networks, the output valve according to output layer judges the ownership class of current gesture.
The beneficial effects of the invention are as follows：The gesture identification method based on depth convolutional neural networks of the present invention, using more
Weight down-sampling technique construction depth convolutional neural networks, and neutral net is carried out as activation primitive using hyperbolic tangent function
Training, can not only improve the efficiency of gesture identification, and can improve the accuracy rate of gesture identification.
Brief description of the drawings
Fig. 1 is the implementing procedure of this method；
Fig. 2 hand-type image zooming-out schematic diagrames；
The multiple down-sampling schematic diagrames of Fig. 3；
Fig. 4 hyperbolic tangent function curves；
The error rate that Fig. 5 discriminations compare.
Embodiment
The present invention is described in detail below in conjunction with the accompanying drawings.
As shown in figure 1, the gesture identification method of the invention based on depth convolutional neural networks is divided into training stage and identification
Stage.
In the training stage, depth convolutional neural networks are built, and its weights and offset parameter are determined using training set data
Value.Specifically include following sub-step：
1st, hand-type image is extracted from training data：For the frame in gesture interaction, as shown in Fig. 2 (a), the present invention is first
Rough hand region image is first extracted according to features of skin colors, as shown in Fig. 2 (b)；Secondly, it is swollen using filtering and quick ecological
Swollen, erosion algorithm is modified to hand area image, further hand-type image is obtained, as shown in Fig. 2 (c)；
2nd, rim detection and size normalized are carried out to hand-type image：The hand-type image obtained to step 1 carries out side
Edge detects and binary conversion treatment, preliminary hand-type contour images is obtained, as shown in Fig. 2 (d)；Then, hand-type contour curve is entered
Row trimming and perfect, and be adjusted to unified size obtain corresponding to hand-type image；
3rd, depth convolutional neural networks are built：The hand-type image obtained for step 2, the present invention construct depth convolution god
It is used for gesture identification through network.If I and O are respectively the input layer and output layer of depth convolutional neural networks, hidden between I and O
Tibetan layer be H1, H2 ..., Hn.Wherein, input layer is the hand-type image that step 2 obtains, and output layer is the gesture that a length is N
Characteristic vector, hidden layer use multiple down-sampling technology, it is allowed to have between the piecemeal of down-sampling it is overlapping, as shown in Figure 3；
4th, activation primitive and loss function are determined：As shown in figure 4, selection non-linear hyperbolic tan is as neuron
Activation primitive；The squared error function shown in formula (2) is selected as loss function.
5th, depth convolutional neural networks are trained：
Concentrated from training sample and choose m training sample, gradient is calculated using steepest descent method, and according to loss letter
Several parameters to each hidden layer are iterated optimization；Then, verified using checking sample set, when accuracy exceedes default threshold
During value 99.5%, terminate training, so as to obtain with the deep neural network for determining weight w and bias term b.
In cognitive phase, first from extracting data hand-type image to be identified, and rim detection and size normalization are carried out
Processing；Then, the depth convolutional neural networks that train are entered into judge the ownership class of current gesture.Finally, it incite somebody to action this
Method compares with the method using data glove and sequence similarity detection (SSDA) method, and Fig. 5 gives mistake
Rate experimental data.
Claims (1)
1. a kind of gesture identification method based on depth convolutional neural networks, comprises the following steps：
(1) division of edge detection process and sample set is carried out to the sample image of training set：First to the gesture collection of training
Image carries out hand-type detection and edge detection process, and by the hand-type Image Adjusting of extraction to unified size；After pre-processing
Data be divided into training sample set and checking sample set；
(2) depth convolutional neural networks are built：If I and O are respectively the input layer and output layer of depth convolutional neural networks, I and O
Between hidden layer for H1, H2 ..., Hn.Wherein, input layer is the hand-type image that step (1) obtains, and output layer is a length
For N gesture feature vector, hidden layer uses multiple down-sampling technology, it is allowed to has between the piecemeal of down-sampling overlapping；
(3) activation primitive and loss function are determined：The non-linear hyperbolic tan shown in formula (1) is selected as neuron
Activation primitive；Select the loss function shown in formula (2).Wherein, n is the number of sample in training set, and x is in hand-type image
Point, y are the output valve corresponding to x, and θ is parameter vector；
<mrow>
<mi>tanh</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mn>1</mn>
<mo>-</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mn>2</mn>
<mi>x</mi>
</mrow>
</msup>
</mrow>
<mrow>
<mn>1</mn>
<mo>+</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mn>2</mn>
<mi>x</mi>
</mrow>
</msup>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>L</mi>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>n</mi>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mi>x</mi>
</munder>
<mo>|</mo>
<mo>|</mo>
<mi>y</mi>
<mo>-</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
(4) deep neural network is trained：Concentrated from training sample and choose m training sample, calculated using steepest descent method
Gradient；Then, verified using checking sample set, when accuracy exceedes predetermined threshold value, terminate training, so as to be had
Determine weight w and bias term b deep neural network；
(5) gesture identification is realized according to the depth convolutional neural networks after training：Its step includes：A) from gesture data to be identified
Middle extraction hand-type image；B) hand-type image is subjected to rim detection and size normalized；C) hand-type image is input to depth
Spend in convolutional neural networks, the output valve according to output layer judges the ownership class of current gesture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710597440.3A CN107480600A (en) | 2017-07-20 | 2017-07-20 | A kind of gesture identification method based on depth convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710597440.3A CN107480600A (en) | 2017-07-20 | 2017-07-20 | A kind of gesture identification method based on depth convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107480600A true CN107480600A (en) | 2017-12-15 |
Family
ID=60595146
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710597440.3A Pending CN107480600A (en) | 2017-07-20 | 2017-07-20 | A kind of gesture identification method based on depth convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107480600A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334814A (en) * | 2018-01-11 | 2018-07-27 | 浙江工业大学 | A kind of AR system gesture identification methods based on convolutional neural networks combination user's habituation behavioural analysis |
| CN109117742A (en) * | 2018-07-20 | 2019-01-01 | 百度在线网络技术（北京）有限公司 | Gestures detection model treatment method, apparatus, equipment and storage medium |
| CN109890573A (en) * | 2019-01-04 | 2019-06-14 | 珊口（上海）智能科技有限公司 | Control method, device, mobile robot and the storage medium of mobile robot |
| CN111338470A (en) * | 2020-02-10 | 2020-06-26 | 烟台持久钟表有限公司 | Method for controlling big clock through gestures |
| CN112889075A (en) * | 2018-10-29 | 2021-06-01 | Sk电信有限公司 | Improving prediction performance using asymmetric hyperbolic tangent activation function |
| CN113703581A (en) * | 2021-09-03 | 2021-11-26 | 广州朗国电子科技股份有限公司 | Window adjusting method based on gesture switching, electronic whiteboard and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529470A (en) * | 2016-11-09 | 2017-03-22 | 济南大学 | Gesture recognition method based on multistage depth convolution neural network |
| US20170161607A1 (en) * | 2015-12-04 | 2017-06-08 | Pilot Ai Labs, Inc. | System and method for improved gesture recognition using neural networks |
-
2017
- 2017-07-20 CN CN201710597440.3A patent/CN107480600A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170161607A1 (en) * | 2015-12-04 | 2017-06-08 | Pilot Ai Labs, Inc. | System and method for improved gesture recognition using neural networks |
| CN106529470A (en) * | 2016-11-09 | 2017-03-22 | 济南大学 | Gesture recognition method based on multistage depth convolution neural network |
Non-Patent Citations (2)
| Title |
|---|
| LEE的白板报: "卷积神经网络初探", 《斯坦福深度学习HTTPS://MY.OSCHINA.NET/FINDBILL/BLOG/550565》 * |
| 蔡娟: "基于卷积神经网络的手势识别", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334814A (en) * | 2018-01-11 | 2018-07-27 | 浙江工业大学 | A kind of AR system gesture identification methods based on convolutional neural networks combination user's habituation behavioural analysis |
| CN108334814B (en) * | 2018-01-11 | 2020-10-30 | 浙江工业大学 | Gesture recognition method of AR system |
| CN109117742A (en) * | 2018-07-20 | 2019-01-01 | 百度在线网络技术（北京）有限公司 | Gestures detection model treatment method, apparatus, equipment and storage medium |
| CN109117742B (en) * | 2018-07-20 | 2022-12-27 | 百度在线网络技术（北京）有限公司 | Gesture detection model processing method, device, equipment and storage medium |
| CN112889075A (en) * | 2018-10-29 | 2021-06-01 | Sk电信有限公司 | Improving prediction performance using asymmetric hyperbolic tangent activation function |
| CN112889075B (en) * | 2018-10-29 | 2024-01-26 | Sk电信有限公司 | Improved predictive performance using asymmetric hyperbolic tangent activation function |
| CN109890573A (en) * | 2019-01-04 | 2019-06-14 | 珊口（上海）智能科技有限公司 | Control method, device, mobile robot and the storage medium of mobile robot |
| CN109890573B (en) * | 2019-01-04 | 2022-05-03 | 上海阿科伯特机器人有限公司 | Control method and device for mobile robot, mobile robot and storage medium |
| CN111338470A (en) * | 2020-02-10 | 2020-06-26 | 烟台持久钟表有限公司 | Method for controlling big clock through gestures |
| CN111338470B (en) * | 2020-02-10 | 2022-10-21 | 烟台持久钟表有限公司 | Method for controlling big clock through gestures |
| CN113703581A (en) * | 2021-09-03 | 2021-11-26 | 广州朗国电子科技股份有限公司 | Window adjusting method based on gesture switching, electronic whiteboard and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107480600A (en) | A kind of gesture identification method based on depth convolutional neural networks | |
| CN106529468B (en) | A kind of finger vein identification method and system based on convolutional neural networks | |
| CN100492400C (en) | Matching identification method by extracting characters of vein from finger | |
| CN100395770C (en) | Hand-characteristic mix-together identifying method based on characteristic relation measure | |
| CN101777117B (en) | Method for extracting finger vein feature for matching identification | |
| CN108460318B (en) | Identity authentication/identification method and equipment based on electrocardiosignals | |
| CN104463100A (en) | Intelligent wheelchair man-machine interaction system and method based on facial expression recognition mode | |
| CN101281598A (en) | Method for recognizing human face based on amalgamation of multicomponent and multiple characteristics | |
| CN104346607A (en) | Face recognition method based on convolutional neural network | |
| CN104239769A (en) | Identity recognition method and system based on finger vein characteristics | |
| CN106529504B (en) | A kind of bimodal video feeling recognition methods of compound space-time characteristic | |
| CN102542281A (en) | Non-contact biometric feature identification method and system | |
| CN107392114A (en) | A kind of finger vein identification method and system based on neural network model | |
| CN101789075A (en) | Finger vein identifying method based on characteristic value normalization and bidirectional weighting | |
| CN110555380A (en) | Finger vein identification method based on Center Loss function | |
| Podder et al. | An efficient iris segmentation model based on eyelids and eyelashes detection in iris recognition system | |
| CN106778498A (en) | A kind of method for strengthening Fingerprint recognition | |
| Wang et al. | Damaged fingerprint classification by Deep Learning with fuzzy feature points | |
| Yeh et al. | ResNeXt with cutout for finger vein analysis | |
| Chen et al. | A finger vein recognition algorithm based on deep learning | |
| CN107977935A (en) | A kind of vena metacarpea image texture filtering reinforcement method | |
| CN103886291B (en) | Palmprint image wrinkle detection method based on dynamic template | |
| Gaxiola et al. | Modular neural networks for person recognition using segmentation and the iris biometric measurement with image pre-processing | |
| Thiyaneswaran et al. | Iris Recognition using Left and Right Iris Feature of the Human Eye for Biometric Security System | |
| CN114973307A (en) | Finger vein identification method and system for generating countermeasure and cosine ternary loss function |
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: 20171215 |