CN108388348A - A kind of electromyography signal gesture identification method based on deep learning and attention mechanism - Google Patents
A kind of electromyography signal gesture identification method based on deep learning and attention mechanism Download PDFInfo
- Publication number
- CN108388348A CN108388348A CN201810224699.8A CN201810224699A CN108388348A CN 108388348 A CN108388348 A CN 108388348A CN 201810224699 A CN201810224699 A CN 201810224699A CN 108388348 A CN108388348 A CN 108388348A
- Authority
- CN
- China
- Prior art keywords
- deep learning
- data
- attention mechanism
- training
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000002567 electromyography Methods 0.000 title claims abstract description 43
- 230000007246 mechanism Effects 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000013135 deep learning Methods 0.000 title claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 43
- 238000012360 testing method Methods 0.000 claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 26
- 230000000306 recurrent effect Effects 0.000 claims abstract description 23
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 18
- 238000001914 filtration Methods 0.000 claims abstract description 9
- 230000009467 reduction Effects 0.000 claims abstract description 3
- 238000000605 extraction Methods 0.000 claims description 12
- 238000013480 data collection Methods 0.000 claims description 8
- 239000012141 concentrate Substances 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000004880 explosion Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 3
- 230000009471 action Effects 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims description 2
- 238000003475 lamination Methods 0.000 claims description 2
- 210000004218 nerve net Anatomy 0.000 claims description 2
- 230000008707 rearrangement Effects 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 238000007476 Maximum Likelihood Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 239000002003 electrode paste Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000013016 learning Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2203/00—Indexing scheme relating to G06F3/00 - G06F3/048
- G06F2203/01—Indexing scheme relating to G06F3/01
- G06F2203/011—Emotion or mood input determined on the basis of sensed human body parameters such as pulse, heart rate or beat, temperature of skin, facial expressions, iris, voice pitch, brain activity patterns
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Neurosurgery (AREA)
- General Health & Medical Sciences (AREA)
- Neurology (AREA)
- Health & Medical Sciences (AREA)
- Dermatology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of electromyography signal gesture identification method based on deep learning and attention mechanism, steps are as follows:Noise reduction filtering is carried out to gesture electromyography signal;One classical feature set is extracted to each window data using sliding window, and builds the myoelectricity image of new feature based;A kind of deep learning frame based on convolutional neural networks, Recognition with Recurrent Neural Network and attention mechanism is designed, and its network architecture parameters is optimized;It trains to obtain sorter model using designed deep learning frame and training data;Test data is input in trained deep learning network model, according to the likelihood of last layer of output, the corresponding classification of maximum likelihood is the classification identified.The present invention is based on new feature image and myoelectricity hand signal is identified in the deep learning frame based on attention mechanism.A variety of different gestures of same subject can be accurately identified using the electromyography signal gesture identification method based on deep learning and attention mechanism.
Description
Technical field
The invention belongs to computers to be combined field with bio signal, more particularly to a kind of based on deep learning and attention
The electromyography signal gesture identification method of mechanism.
Background technology
Surface electromyogram signal (surface electromyography, sEMG) is a kind of electrode paste by non-intrusion type
The bio signal of muscle activity is recorded in skin surface.Can be auxiliary and rehabilitation by recording and analyzing surface electromyogram signal
Technology provides more effective information, has for Sports Scientific Research, human-computer interaction, medical science of recovery therapy clinic and basic research etc.
Important learning value and application value.In such applications, the Gesture Recognition based on electromyography signal takes on important angle
Color.One classical electromyography signal gesture identification flow is made of data prediction, feature space structure and classification.Data are located in advance
Reason part mainly carries out rectification and filtering to reduce noise to signal, and feature space structure part converts pretreated signal
To feature space so that there is the discrimination of bigger between class, model finally is trained for classifying with a machine learning method.
The structure part of feature space and the other identification division of gesture class are to improve highly important two of recognition accuracy
Part.Therefore many researchers are dedicated to proposing new feature by their domain knowledge, such as Phinyomark spies
Collection.On the other hand, in research at home and abroad, many Machine learning classifiers are used in electromyography signal gesture identification, than
Such as artificial neural network, k nearest neighbor, linear judgment analysis, support vector machines and hidden Markov model.Wherein support vector machines and
Linear judgment analysis is two kinds of most common graders.
In progress both domestic and external in recent years, deep learning method all obtains table best at present in many fields
It is existing.Wherein foremost convolutional neural networks have also been successfully applied in the gesture identification of electromyography signal, are obtained at present most
Good recognition effect.Attention mechanism is a kind of method of highly effective enhancing Recognition with Recurrent Neural Network modeling ability, at present
Preferable effect is achieved in fields such as machine translation.But currently without using Recognition with Recurrent Neural Network combination convolutional neural networks
Electromyography signal gesture is identified in method, and the present invention by attention mechanism be added in Recognition with Recurrent Neural Network to model into
Row enhancing.
Invention content
In view of the above-mentioned deficiencies in the prior art, it is an object of the present invention to provide a kind of based on deep learning and attention mechanism
Electromyography signal gesture identification method, by designing the model based on convolutional neural networks, Recognition with Recurrent Neural Network and attention mechanism
Structure improves the accuracy of gesture identification.
The purpose of the present invention is achieved through the following technical solutions:It is a kind of based on deep learning and attention mechanism
Electromyography signal gesture identification method, includes the following steps:
(1) myoelectricity data, data prediction, including following sub-step are obtained:
(1.1) from public data collection NinaProDB1, NinaProDB2, BioPatRec subset, CapgMyo subsets and
Gesture motion myoelectricity data are obtained in csl-hdemg;
(1.2) noise reduction is filtered using different preprocess methods to different data collection respectively;
(2) division of original signal training dataset and original signal test data set, including following sub-step:
(2.1) according to the electromyography signal label got, the data in each electromyography signal file are divided into several
Electromyography signal gesture section, each gesture section are repeated comprising one-off;
(2.2) according to different appraisal procedures, the multiple action of gesture is repeated original signal training number is respectively divided
It is concentrated according to collection and original signal test data, completes original training and the division of test data set;
(3) data segmentation and feature extraction, including following sub-step:
(3.1) each gesture section is divided into the signal segment of multiple regular lengths with sliding window;
(3.2) feature extraction is carried out to each channel of the fixed length signal segment in each window, extracts various features;
(4) new myoelectricity image, including following sub-step are built:
(4.1) by the feature vector rearrangement in each channel in window so that each two channel can be adjacent;
(4.2) build new myoelectricity image, the width of new myoelectricity image is 1, it is a height of rearrange after port number, Color Channel
Number is characterized vector dimension;
(5) the electromyography signal multiclass gesture identification based on deep learning and attention mechanism, includes the following steps:
(5.1) model structure of projected depth study and attention mechanism, model structure is by convolutional neural networks, cycle god
It is constituted through network and primary attention power mechanism;Convolutional neural networks carry out high-level characteristic extraction to the new myoelectricity image of input, follow
Relationship of the ring neural network between the every frame of new myoelectricity image sequence models, and primary attention power mechanism is to Recognition with Recurrent Neural Network
Output carry out importance weighting, t moment attention weight αtCalculation formula be:
Mt=tanh (Whht)
αt=softmax (wTMt)
Wherein, htIt is the output of Recognition with Recurrent Neural Network, WhAnd wTIt is weight matrix to be trained, when T is a gesture section
Between length, r is the output of primary attention power machined part;Softmax functions are normalization exponential functions;
(5.2) original signal training data concentrates each sample to carry out the structure of new myoelectricity image, obtains new myoelectricity image
Input of the training dataset as whole network carries out the network parameter of convolutional neural networks and Recognition with Recurrent Neural Network excellent one by one
Change, obtains optimal model parameters;
(5.3) optimal model parameters and the training of new myoelectricity image training dataset obtained by step (5.2) training obtain
Disaggregated model;
(5.4) it concentrates each sample to carry out the structure of new myoelectricity image test data, obtains new myoelectricity image measurement number
According to collection, the disaggregated model that input step (5.3) obtains, output category result.
Further, right to NinaProDB1 using low pass butterworth filtering in the step (1.2)
NinaProDB2 filters and is downsampled to 100Hz using low pass butterworth, and BioPatRec subsets and CapgMyo subsets are not
It is filtered, rectification and low pass butterworth filtering is carried out to csl-hdemg.
Further, in the step (2.1), the division of original signal training dataset and original signal test data set
Using being assessed in subject;Different data collection uses different division methods:NinaProDB1 by the 1,3,4,6th of each subject the,
8,9 and 10 repetitions are as training data, and the 2nd, 5,7 time as test data;NinaProDB2 makees the 1,3,4,6th repetition
For training data, the 2nd, 5 time as test data;BioPatRec subsets repeat first time to be used as training data, twice
It repeats to be used as test data;CapgMyo subsets using the repetition of half as training data, i.e. 5 repetitions, make by addition 5 repetitions
For test data;The data of single-subject are divided into 10 parts by csl-hdemg data sets, and carry out 10 folding cross validations.
Further, in the step (3.1), different data sets uses different sliding window length and sliding step
It is long;The sliding window length of NinaProDB1 is 150ms and 200ms, sliding step 10ms;The sliding window of NinaProDB2
Length is 200ms, sliding step 100ms;The sliding window length of BioPatRec subsets is 50ms and 150ms, sliding step
For 50ms;The sliding window length of CapgMyo subsets is 40ms and 150ms, sliding step 1ms;The sliding window of csl-hdemg
Mouth length is 150ms and 170ms, sliding step 0.5ms.
Further, in the step (3.2), to the electromyography signal in window be based on classics feature set Phinyomark into
Row characteristic vector pickup, including characteristic signal amplitude absolute mean MAV, waveform length WL, autoregressive coefficient AR, absolute mean are oblique
Energy and gross energy ratio PSR and Willison amplitude WAMP near rate MAVSLP, average frequency MNF, power spectrum maximum value;
CapgMyo subsets and csl-hdemg are high density electromyography signals, without feature extraction, the structure figures directly in original signal
Picture.
Further, in the step (5.1), Recognition with Recurrent Neural Network part, the long mnemon (LSTM) in short-term of selection is come
Gradient is solved to disappear and gradient explosion issues.
Further, in the step (5.2), convolutional neural networks include level 2 volume lamination in optimal models, are followed by 2 layers
Local articulamentum finally connects 3 layers of full articulamentum, the length mnemon in short-term that Recognition with Recurrent Neural Network layer is 512 by output size
(LSTM) it constitutes, last identification division is made of the full articulamentums of a G-way and softmax layers.
Further, in the step (5.3), the training process of disaggregated model is:New myoelectricity image training dataset with
The each sample of the data set corresponds to input of the gesture label collectively as model, and obtaining model parameter by training is stored.
Further, in the step (5.4), the output of disaggregated model is label, that is, corresponds to the label of test sample, is used
Recognition accuracy weighs recognition result, and recognition accuracy is to identify correct sample number divided by all test sample numbers.
The beneficial effects of the invention are as follows:The present invention proposes a kind of myoelectricity based on convolutional neural networks and Recognition with Recurrent Neural Network
Signal gesture identification method can be extracted and be modeled to the room and time feature of electromyography signal simultaneously, with existing invention
In the method for being based purely on convolutional neural networks compare, this method can effectively promote discrimination.Attention mechanism is added
Into the model structure based on convolutional neural networks and Recognition with Recurrent Neural Network, the performance of model structure can be enhanced.Extraction tradition
Electromyography signal builds input of the new myoelectricity image as model, can effectively promote the accuracy rate of electromyography signal gesture identification.
Description of the drawings
Fig. 1 is the method for the present invention flow chart;
Fig. 2 is inventive network structure chart.
Specific implementation mode
Invention is further described in detail with reference to the accompanying drawings and detailed description.
As shown in Figure 1, a kind of electromyography signal gesture identification based on deep learning and attention mechanism provided by the invention
Method, specific implementation step are as follows:
Step (1) from public data collection NinaProDB1, NinaProDB2, BioPatRec subset, CapgMyo subsets and
Gesture motion myoelectricity data are obtained in csl-hdemg;It is right to NinaProDB1 using low pass butterworth filtering
NinaProDB2 filters and is downsampled to 100Hz using low pass butterworth, and BioPatRec subsets and CapgMyo subsets are not
It is filtered, rectification and low pass butterworth filtering is carried out to csl-hdemg.
The division of step (2) original signal training dataset and original signal test data set, according to the myoelectricity got
Data in each electromyography signal file are divided into electromyography signal gesture section one by one by signal label, and each gesture section includes
One-off repeats;Our test uses different division methods using assessment in subject, different data collection:NinaProDB1
By the 1st, 3,4,6,8,9 and 10 repetition of each subject as training data, the 2nd, 5,7 time as test data;
NinaProDB2 is by the 1st, 3,4,6 repetition as training data, and the 2nd, 5 time as test data;BioPatRec subsets are by
It is primary to repeat to be used as training data, it repeats to be used as test data twice;CapgMyo subsets are using the repetition of half as training
Data, i.e. 5 repetitions, in addition 5 repetitions are as test data;The data of single-subject are divided by csl-hdemg data sets
10 parts, and carry out 10 folding cross validations.
Step (3) is split to data and feature extraction, different data sets using different sliding window length and
Sliding step.The sliding window length of NinaProDB1 is 150ms and 200ms, sliding step 10ms;The cunning of NinaProDB2
Dynamic length of window is 200ms, sliding step 100ms;The sliding window length of BioPatRec subsets is 50ms and 150ms, sliding
Dynamic step-length is 50ms;The sliding window length of CapgMyo subsets is 40ms and 150ms, sliding step 1ms;Csl-hdemg's
Sliding window length is 150ms and 170ms, sliding step 0.5ms.Classical feature set is based on to the electromyography signal in window
Phinyomark carries out characteristic vector pickup, including characteristic signal amplitude absolute mean (MAV), waveform length (WL), autoregression
Energy and gross energy ratio near coefficient (AR), absolute mean slope (MAVSLP), average frequency (MNF), power spectrum maximum value
(PSR) and Willison amplitudes (WAMP).
Step (5) projected depth learns and the model structure of attention mechanism, and model structure is by convolutional neural networks, cycle
Neural network and primary attention power mechanism are constituted.Convolutional neural networks carry out high-level characteristic extraction to the new myoelectricity image of input,
Relationship of the Recognition with Recurrent Neural Network between the every frame of new myoelectricity image sequence models, and primary attention power mechanism is to recycling nerve net
The output of network carries out importance weighting, to obtain final expression for electromyography signal gesture identification.To convolutional neural networks
It is optimized one by one with the network parameter of Recognition with Recurrent Neural Network, optimal network structure is as shown in the table:
Layer | Title | Parameter |
1 | Convolutional layer 1 | 64 cores, core size (3*3) |
2 | Convolutional layer 2 | 64 cores, core size (3*3) |
3 | Local articulamentum 1 | 64 cores |
4 | Local articulamentum 2 | 64 cores |
5 | Full articulamentum 1 | 512 dimension outputs |
6 | Full articulamentum 2 | 512 dimension outputs |
7 | Full articulamentum 3 | 128 dimension outputs |
8 | Recycle nervous layer | Long short-term memory (LSTM) 512 ties up hidden state output |
9 | Full articulamentum 4 and softmax layers |
Recognition with Recurrent Neural Network part, we select long mnemon (LSTM) in short-term and disappear and gradient explosion to solve gradient
Problem.Attention mechanism increases behind Recognition with Recurrent Neural Network, i.e. the output of Recognition with Recurrent Neural Network is attention machined part
Input, calculation formula is:
Mt=tanh (Whht)
αt=softmax (wTMt)
Wherein, htIt is the output of Recognition with Recurrent Neural Network, WhAnd wTIt is weight matrix to be trained, when T is a gesture section
Between length, r is the output of primary attention power machined part;Softmax functions are normalization exponential functions.Training process is:It is former
Beginning signal training data concentrates each sample to carry out the structure of new myoelectricity image, obtains new myoelectricity image training dataset, will be new
Myoelectricity image training dataset gesture label corresponding with each sample of the data set collectively as model input, by trained
To model parameter and stored.Test process is:It concentrates each sample to carry out the structure of new myoelectricity image test data, obtains
To new myoelectricity image test data set, loads by the new trained model of myoelectricity image training dataset, input new myoelectricity image
Test data set is exported as gesture class label, is weighed to recognition result with recognition accuracy, and recognition accuracy is identification
Correct sample number divided by all sample numbers.
To NinaProDB1, NinaProDB2, BioPatRec subset, CapgMyo subsets and csl-hdemg data sets
Gesture complete or collected works are identified.NinaProDB1 includes 52 gestures, and NinaProDB2 includes 50 gestures, and BioPatRec subsets include
26 gestures, CapgMyo subsets include 8 gestures, and csl-hdemg includes 27 gestures.Using the present invention is based on deep learnings and attention
The discrimination result of the electromyography signal gesture identification method of power mechanism is:
Claims (9)
1. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism, which is characterized in that including as follows
Step:
(1) myoelectricity data, data prediction, including following sub-step are obtained:
(1.1) from public data collection NinaProDB1, NinaProDB2, BioPatRec subset, CapgMyo subsets and csl-
Gesture motion myoelectricity data are obtained in hdemg;
(1.2) noise reduction is filtered using different preprocess methods to different data collection respectively;
(2) division of original signal training dataset and original signal test data set, including following sub-step:
(2.1) according to the electromyography signal label got, the data in each electromyography signal file are divided into several myoelectricities
Signal gesture section, each gesture section are repeated comprising one-off;
(2.2) according to different appraisal procedures, the multiple action of gesture is repeated original signal training dataset is respectively divided
It is concentrated with original signal test data, completes original training and the division of test data set;
(3) data segmentation and feature extraction, including following sub-step:
(3.1) each gesture section is divided into the signal segment of multiple regular lengths with sliding window;
(3.2) feature extraction is carried out to each channel of the fixed length signal segment in each window, extracts various features;
(4) new myoelectricity image, including following sub-step are built:
(4.1) by the feature vector rearrangement in each channel in window so that each two channel can be adjacent;
(4.2) build new myoelectricity image, the width of new myoelectricity image is 1, it is a height of rearrange after port number, Color Channel number is
Feature vector dimension;
(5) the electromyography signal multiclass gesture identification based on deep learning and attention mechanism, includes the following steps:
(5.1) model structure of projected depth study and attention mechanism, model structure is by convolutional neural networks, cycle nerve net
Network and primary attention power mechanism are constituted;Convolutional neural networks carry out high-level characteristic extraction, cycle god to the new myoelectricity image of input
Relationship through network between the every frame of new myoelectricity image sequence models, and primary attention power mechanism is to the defeated of Recognition with Recurrent Neural Network
Go out and carries out importance weighting, t moment attention weight αtCalculation formula be:
Mt=tanh (Whht)
αt=softmax (wTMt)
Wherein, htIt is the output of Recognition with Recurrent Neural Network, WhAnd wTIt is weight matrix to be trained, T is that the time of a gesture section is long
Degree, r is the output of primary attention power machined part;Softmax functions are normalization exponential functions;
(5.2) original signal training data concentrates each sample to carry out the structure of new myoelectricity image, obtains new myoelectricity image training
Input of the data set as whole network optimizes the network parameter of convolutional neural networks and Recognition with Recurrent Neural Network one by one,
Obtain optimal model parameters;
(5.3) optimal model parameters and the training of new myoelectricity image training dataset obtained by step (5.2) training are classified
Model;
(5.4) it concentrates each sample to carry out the structure of new myoelectricity image test data, obtains new myoelectricity image test data set,
The disaggregated model that input step (5.3) obtains, output category result.
2. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (1.2), to NinaProDB1 using low pass butterworth filtering, NinaProDB2 is used
Low pass butterworth filter and be downsampled to 100Hz, BioPatRec subsets and CapgMyo subsets without filtering, it is right
Csl-hdemg carries out rectification and low pass butterworth filtering.
3. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (2.1), the division of original signal training dataset and original signal test data set uses subject
Interior assessment;Different data collection uses different division methods:NinaProDB1 is by the 1,3,4,6,8,9th and 10 time of each subject
It repeats to be used as training data, the 2nd, 5,7 time as test data;NinaProDB2 is using the 1,3,4,6th repetition as training number
According to the 2nd, 5 time as test data;BioPatRec subsets repeat first time to be used as training data, twice repeatedly conduct
Test data;CapgMyo subsets are using the repetition of half as training data, i.e. 5 repetitions, and in addition 5 repetitions are as test number
According to;The data of single-subject are divided into 10 parts by csl-hdemg data sets, and carry out 10 folding cross validations.
4. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (3.1), different data sets uses different sliding window length and sliding step;
The sliding window length of NinaProDB1 is 150ms and 200ms, sliding step 10ms;The sliding window of NinaProDB2 is long
Degree is 200ms, sliding step 100ms;The sliding window length of BioPatRec subsets is 50ms and 150ms, and sliding step is
50ms;The sliding window length of CapgMyo subsets is 40ms and 150ms, sliding step 1ms;The sliding window of csl-hdemg
Length is 150ms and 170ms, sliding step 0.5ms.
5. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
Be characterized in that, in the step (3.2), to the electromyography signal in window be based on classical feature set Phinyomark carry out feature to
Amount extraction, including characteristic signal amplitude absolute mean (MAV), waveform length (WL), autoregressive coefficient (AR), absolute mean slope
(MAVSLP), energy and gross energy ratio (PSR) and Willison amplitudes near average frequency (MNF), power spectrum maximum value
(WAMP);CapgMyo subsets and csl-hdemg are high density electromyography signals, without feature extraction, directly in original signal
Build image.
6. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (5.1), Recognition with Recurrent Neural Network part selects long mnemon (LSTM) in short-term to solve gradient
It disappears and gradient explosion issues.
7. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (5.2), convolutional neural networks include level 2 volume lamination in optimal models, are followed by 2 layers of part connection
Layer finally connects 3 layers of full articulamentum, length mnemon (LSTM) structure in short-term that Recognition with Recurrent Neural Network layer is 512 by output size
At last identification division is made of the full articulamentums of a G-way and softmax layers.
8. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (5.3), the training process of disaggregated model is:New myoelectricity image training dataset and the data set
Each sample corresponds to input of the gesture label collectively as model, and obtaining model parameter by training is stored.
9. a kind of electromyography signal gesture identification method based on deep learning and attention mechanism according to claim 1,
It is characterized in that, in the step (5.4), the output of disaggregated model is label, that is, corresponds to the label of test sample, accurate with identification
Rate weighs recognition result, and recognition accuracy is to identify correct sample number divided by all test sample numbers.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810224699.8A CN108388348B (en) | 2018-03-19 | 2018-03-19 | Myoelectric signal gesture recognition method based on deep learning and attention mechanism |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810224699.8A CN108388348B (en) | 2018-03-19 | 2018-03-19 | Myoelectric signal gesture recognition method based on deep learning and attention mechanism |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108388348A true CN108388348A (en) | 2018-08-10 |
CN108388348B CN108388348B (en) | 2020-11-24 |
Family
ID=63066958
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810224699.8A Expired - Fee Related CN108388348B (en) | 2018-03-19 | 2018-03-19 | Myoelectric signal gesture recognition method based on deep learning and attention mechanism |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108388348B (en) |
Cited By (45)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109124625A (en) * | 2018-09-04 | 2019-01-04 | 大连理工大学 | A kind of driver fatigue state horizontal mipmap method |
CN109190578A (en) * | 2018-09-13 | 2019-01-11 | 合肥工业大学 | The sign language video interpretation method merged based on convolution network with Recognition with Recurrent Neural Network |
CN109257622A (en) * | 2018-11-01 | 2019-01-22 | 广州市百果园信息技术有限公司 | A kind of audio/video processing method, device, equipment and medium |
CN109498362A (en) * | 2018-09-10 | 2019-03-22 | 南京航空航天大学 | A kind of hemiplegic patient's hand movement function device for healing and training and model training method |
CN109567789A (en) * | 2018-12-03 | 2019-04-05 | 东软集团股份有限公司 | Division processing method, device and the readable storage medium storing program for executing of ECG data |
CN109886978A (en) * | 2019-02-20 | 2019-06-14 | 贵州电网有限责任公司 | A kind of end-to-end warning information recognition methods based on deep learning |
CN109924977A (en) * | 2019-03-21 | 2019-06-25 | 西安交通大学 | A kind of surface electromyogram signal classification method based on CNN and LSTM |
CN110033126A (en) * | 2019-03-14 | 2019-07-19 | 贵州大学 | Shot and long term memory network prediction technique based on attention mechanism and logistic regression |
CN110262511A (en) * | 2019-07-12 | 2019-09-20 | 同济人工智能研究院(苏州)有限公司 | Biped robot's adaptivity ambulation control method based on deeply study |
CN110333783A (en) * | 2019-07-10 | 2019-10-15 | 中国科学技术大学 | A kind of unrelated gesture processing method and system for robust myoelectric control |
CN110399846A (en) * | 2019-07-03 | 2019-11-01 | 北京航空航天大学 | A kind of gesture identification method based on multichannel electromyography signal correlation |
CN110443309A (en) * | 2019-08-07 | 2019-11-12 | 浙江大学 | A kind of electromyography signal gesture identification method of combination cross-module state association relation model |
CN110537922A (en) * | 2019-09-09 | 2019-12-06 | 北京航空航天大学 | Human body walking process lower limb movement identification method and system based on deep learning |
CN110598628A (en) * | 2019-09-11 | 2019-12-20 | 南京邮电大学 | Electromyographic signal hand motion recognition method based on integrated deep learning |
CN110598676A (en) * | 2019-09-25 | 2019-12-20 | 南京邮电大学 | Deep learning gesture electromyographic signal identification method based on confidence score model |
CN110610172A (en) * | 2019-09-25 | 2019-12-24 | 南京邮电大学 | Myoelectric gesture recognition method based on RNN-CNN architecture |
CN110618754A (en) * | 2019-08-30 | 2019-12-27 | 电子科技大学 | Surface electromyogram signal-based gesture recognition method and gesture recognition armband |
CN110658915A (en) * | 2019-07-24 | 2020-01-07 | 浙江工业大学 | Electromyographic signal gesture recognition method based on double-current network |
CN110929243A (en) * | 2019-11-22 | 2020-03-27 | 武汉大学 | Pedestrian identity recognition method based on mobile phone inertial sensor |
CN111046731A (en) * | 2019-11-11 | 2020-04-21 | 中国科学院计算技术研究所 | Transfer learning method and recognition method for gesture recognition based on surface electromyogram signals |
CN111053549A (en) * | 2019-12-23 | 2020-04-24 | 威海北洋电气集团股份有限公司 | Intelligent biological signal abnormality detection method and system |
CN111103976A (en) * | 2019-12-05 | 2020-05-05 | 深圳职业技术学院 | Gesture recognition method and device and electronic equipment |
CN111144269A (en) * | 2019-12-23 | 2020-05-12 | 威海北洋电气集团股份有限公司 | Signal-related behavior identification method and system based on deep learning |
CN111184512A (en) * | 2019-12-30 | 2020-05-22 | 电子科技大学 | Method for recognizing rehabilitation training actions of upper limbs and hands of stroke patient |
CN111616706A (en) * | 2020-05-20 | 2020-09-04 | 山东中科先进技术研究院有限公司 | Surface electromyogram signal classification method and system based on convolutional neural network |
CN111920405A (en) * | 2020-09-15 | 2020-11-13 | 齐鲁工业大学 | Atrial fibrillation signal identification system and method |
CN111973388A (en) * | 2019-05-22 | 2020-11-24 | 中国科学院沈阳自动化研究所 | Hand rehabilitation robot control method based on sEMG |
WO2020258611A1 (en) * | 2019-06-28 | 2020-12-30 | 山东科技大学 | Lymph node ct detection system employing recurrent spatio-temporal attention mechanism |
CN112783327A (en) * | 2021-01-29 | 2021-05-11 | 中国科学院计算技术研究所 | Method and system for gesture recognition based on surface electromyogram signals |
CN112816122A (en) * | 2020-12-31 | 2021-05-18 | 武汉地震工程研究院有限公司 | Bolt tightness degree monitoring method based on deep learning and piezoelectric active sensing |
CN112861604A (en) * | 2020-12-25 | 2021-05-28 | 中国科学技术大学 | Myoelectric action recognition and control method irrelevant to user |
CN112906673A (en) * | 2021-04-09 | 2021-06-04 | 河北工业大学 | Lower limb movement intention prediction method based on attention mechanism |
CN113143261A (en) * | 2021-04-30 | 2021-07-23 | 中国科学院自动化研究所 | Electromyographic signal-based identity recognition system, method and equipment |
CN113205074A (en) * | 2021-05-29 | 2021-08-03 | 浙江大学 | Gesture recognition method fusing multi-mode signals of myoelectricity and micro-inertia measurement unit |
CN113312994A (en) * | 2021-05-18 | 2021-08-27 | 中国科学院深圳先进技术研究院 | Gesture classification recognition method and application thereof |
CN113609923A (en) * | 2021-07-13 | 2021-11-05 | 中国矿业大学 | Attention-based continuous sign language sentence recognition method |
CN113627401A (en) * | 2021-10-12 | 2021-11-09 | 四川大学 | Myoelectric gesture recognition method of feature pyramid network fused with double-attention machine system |
CN113729738A (en) * | 2021-09-13 | 2021-12-03 | 武汉科技大学 | Construction method of multi-channel electromyographic feature image |
CN113934302A (en) * | 2021-10-21 | 2022-01-14 | 燕山大学 | Myoelectric gesture recognition method based on SeNet and gating time sequence convolution network |
CN114343679A (en) * | 2021-12-24 | 2022-04-15 | 杭州电子科技大学 | Surface electromyogram signal upper limb action recognition method and system based on transfer learning |
CN114504333A (en) * | 2022-01-30 | 2022-05-17 | 天津大学 | Wearable vestibule monitoring system based on myoelectricity and application |
CN114569143A (en) * | 2022-03-03 | 2022-06-03 | 上海交通大学宁波人工智能研究院 | Myoelectric gesture recognition method based on attention mechanism and multi-feature fusion |
CN114931389A (en) * | 2022-04-27 | 2022-08-23 | 福州大学 | Electromyographic signal identification method based on residual error network and graph convolution network |
CN116738295A (en) * | 2023-08-10 | 2023-09-12 | 齐鲁工业大学(山东省科学院) | sEMG signal classification method, system, electronic device and storage medium |
CN117281528A (en) * | 2023-11-27 | 2023-12-26 | 山东锋士信息技术有限公司 | Multi-lead pulse signal intelligent identification method and system based on deep learning |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105446484A (en) * | 2015-11-19 | 2016-03-30 | 浙江大学 | Electromyographic signal gesture recognition method based on hidden markov model |
CN105654037A (en) * | 2015-12-21 | 2016-06-08 | 浙江大学 | Myoelectric signal gesture recognition method based on depth learning and feature images |
CN106980367A (en) * | 2017-02-27 | 2017-07-25 | 浙江工业大学 | A kind of gesture identification method based on myoelectricity topographic map |
US20170262995A1 (en) * | 2016-03-11 | 2017-09-14 | Qualcomm Incorporated | Video analysis with convolutional attention recurrent neural networks |
US20170293804A1 (en) * | 2016-04-06 | 2017-10-12 | Nec Laboratories America, Inc. | Deep 3d attention long short-term memory for video-based action recognition |
-
2018
- 2018-03-19 CN CN201810224699.8A patent/CN108388348B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105446484A (en) * | 2015-11-19 | 2016-03-30 | 浙江大学 | Electromyographic signal gesture recognition method based on hidden markov model |
CN105654037A (en) * | 2015-12-21 | 2016-06-08 | 浙江大学 | Myoelectric signal gesture recognition method based on depth learning and feature images |
US20170262995A1 (en) * | 2016-03-11 | 2017-09-14 | Qualcomm Incorporated | Video analysis with convolutional attention recurrent neural networks |
US20170293804A1 (en) * | 2016-04-06 | 2017-10-12 | Nec Laboratories America, Inc. | Deep 3d attention long short-term memory for video-based action recognition |
CN106980367A (en) * | 2017-02-27 | 2017-07-25 | 浙江工业大学 | A kind of gesture identification method based on myoelectricity topographic map |
Non-Patent Citations (4)
Title |
---|
BAZIOTIS C ET AL.: "Deep LSTM with Attention for Message-level and Topic-based Sentiment Analysis", 《SEMECAL-2017》 * |
ORDONEZ FJ ET AL.: "Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition", 《SENSORS》 * |
WANG P ET AL.: "Sequentially Supervised Long Short-Term Memory for Gesture Recognition", 《COGNITIVE COMPUTATION》 * |
刘全 等: "深度强化学习综述", 《计算机学报》 * |
Cited By (64)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109124625A (en) * | 2018-09-04 | 2019-01-04 | 大连理工大学 | A kind of driver fatigue state horizontal mipmap method |
CN109124625B (en) * | 2018-09-04 | 2021-07-20 | 大连理工大学 | Driver fatigue state level grading method |
CN109498362A (en) * | 2018-09-10 | 2019-03-22 | 南京航空航天大学 | A kind of hemiplegic patient's hand movement function device for healing and training and model training method |
CN109190578A (en) * | 2018-09-13 | 2019-01-11 | 合肥工业大学 | The sign language video interpretation method merged based on convolution network with Recognition with Recurrent Neural Network |
CN109257622A (en) * | 2018-11-01 | 2019-01-22 | 广州市百果园信息技术有限公司 | A kind of audio/video processing method, device, equipment and medium |
CN109567789A (en) * | 2018-12-03 | 2019-04-05 | 东软集团股份有限公司 | Division processing method, device and the readable storage medium storing program for executing of ECG data |
CN109886978A (en) * | 2019-02-20 | 2019-06-14 | 贵州电网有限责任公司 | A kind of end-to-end warning information recognition methods based on deep learning |
CN110033126A (en) * | 2019-03-14 | 2019-07-19 | 贵州大学 | Shot and long term memory network prediction technique based on attention mechanism and logistic regression |
CN109924977A (en) * | 2019-03-21 | 2019-06-25 | 西安交通大学 | A kind of surface electromyogram signal classification method based on CNN and LSTM |
CN111973388A (en) * | 2019-05-22 | 2020-11-24 | 中国科学院沈阳自动化研究所 | Hand rehabilitation robot control method based on sEMG |
CN111973388B (en) * | 2019-05-22 | 2021-08-31 | 中国科学院沈阳自动化研究所 | Hand rehabilitation robot control method based on sEMG |
WO2020258611A1 (en) * | 2019-06-28 | 2020-12-30 | 山东科技大学 | Lymph node ct detection system employing recurrent spatio-temporal attention mechanism |
CN110399846A (en) * | 2019-07-03 | 2019-11-01 | 北京航空航天大学 | A kind of gesture identification method based on multichannel electromyography signal correlation |
CN110333783A (en) * | 2019-07-10 | 2019-10-15 | 中国科学技术大学 | A kind of unrelated gesture processing method and system for robust myoelectric control |
CN110262511B (en) * | 2019-07-12 | 2022-08-09 | 同济人工智能研究院(苏州)有限公司 | Biped robot adaptive walking control method based on deep reinforcement learning |
CN110262511A (en) * | 2019-07-12 | 2019-09-20 | 同济人工智能研究院(苏州)有限公司 | Biped robot's adaptivity ambulation control method based on deeply study |
CN110658915A (en) * | 2019-07-24 | 2020-01-07 | 浙江工业大学 | Electromyographic signal gesture recognition method based on double-current network |
CN110443309A (en) * | 2019-08-07 | 2019-11-12 | 浙江大学 | A kind of electromyography signal gesture identification method of combination cross-module state association relation model |
CN110618754A (en) * | 2019-08-30 | 2019-12-27 | 电子科技大学 | Surface electromyogram signal-based gesture recognition method and gesture recognition armband |
CN110537922A (en) * | 2019-09-09 | 2019-12-06 | 北京航空航天大学 | Human body walking process lower limb movement identification method and system based on deep learning |
CN110537922B (en) * | 2019-09-09 | 2020-09-04 | 北京航空航天大学 | Human body walking process lower limb movement identification method and system based on deep learning |
CN110598628A (en) * | 2019-09-11 | 2019-12-20 | 南京邮电大学 | Electromyographic signal hand motion recognition method based on integrated deep learning |
CN110598628B (en) * | 2019-09-11 | 2022-08-02 | 南京邮电大学 | Electromyographic signal hand motion recognition method based on integrated deep learning |
CN110598676B (en) * | 2019-09-25 | 2022-08-02 | 南京邮电大学 | Deep learning gesture electromyographic signal identification method based on confidence score model |
CN110610172B (en) * | 2019-09-25 | 2022-08-12 | 南京邮电大学 | Myoelectric gesture recognition method based on RNN-CNN architecture |
CN110598676A (en) * | 2019-09-25 | 2019-12-20 | 南京邮电大学 | Deep learning gesture electromyographic signal identification method based on confidence score model |
CN110610172A (en) * | 2019-09-25 | 2019-12-24 | 南京邮电大学 | Myoelectric gesture recognition method based on RNN-CNN architecture |
CN111046731A (en) * | 2019-11-11 | 2020-04-21 | 中国科学院计算技术研究所 | Transfer learning method and recognition method for gesture recognition based on surface electromyogram signals |
CN110929243A (en) * | 2019-11-22 | 2020-03-27 | 武汉大学 | Pedestrian identity recognition method based on mobile phone inertial sensor |
CN110929243B (en) * | 2019-11-22 | 2022-07-22 | 武汉大学 | Pedestrian identity recognition method based on mobile phone inertial sensor |
CN111103976A (en) * | 2019-12-05 | 2020-05-05 | 深圳职业技术学院 | Gesture recognition method and device and electronic equipment |
CN111103976B (en) * | 2019-12-05 | 2023-05-02 | 深圳职业技术学院 | Gesture recognition method and device and electronic equipment |
CN111144269A (en) * | 2019-12-23 | 2020-05-12 | 威海北洋电气集团股份有限公司 | Signal-related behavior identification method and system based on deep learning |
CN111053549A (en) * | 2019-12-23 | 2020-04-24 | 威海北洋电气集团股份有限公司 | Intelligent biological signal abnormality detection method and system |
CN111144269B (en) * | 2019-12-23 | 2023-11-24 | 威海北洋电气集团股份有限公司 | Signal correlation behavior recognition method and system based on deep learning |
CN111184512A (en) * | 2019-12-30 | 2020-05-22 | 电子科技大学 | Method for recognizing rehabilitation training actions of upper limbs and hands of stroke patient |
CN111184512B (en) * | 2019-12-30 | 2021-06-01 | 电子科技大学 | Method for recognizing rehabilitation training actions of upper limbs and hands of stroke patient |
CN111616706A (en) * | 2020-05-20 | 2020-09-04 | 山东中科先进技术研究院有限公司 | Surface electromyogram signal classification method and system based on convolutional neural network |
CN111616706B (en) * | 2020-05-20 | 2022-07-22 | 山东中科先进技术有限公司 | Surface electromyogram signal classification method and system based on convolutional neural network |
CN111920405A (en) * | 2020-09-15 | 2020-11-13 | 齐鲁工业大学 | Atrial fibrillation signal identification system and method |
CN112861604B (en) * | 2020-12-25 | 2022-09-06 | 中国科学技术大学 | Myoelectric action recognition and control method irrelevant to user |
CN112861604A (en) * | 2020-12-25 | 2021-05-28 | 中国科学技术大学 | Myoelectric action recognition and control method irrelevant to user |
CN112816122A (en) * | 2020-12-31 | 2021-05-18 | 武汉地震工程研究院有限公司 | Bolt tightness degree monitoring method based on deep learning and piezoelectric active sensing |
CN112783327A (en) * | 2021-01-29 | 2021-05-11 | 中国科学院计算技术研究所 | Method and system for gesture recognition based on surface electromyogram signals |
CN112906673A (en) * | 2021-04-09 | 2021-06-04 | 河北工业大学 | Lower limb movement intention prediction method based on attention mechanism |
CN113143261A (en) * | 2021-04-30 | 2021-07-23 | 中国科学院自动化研究所 | Electromyographic signal-based identity recognition system, method and equipment |
WO2022242133A1 (en) * | 2021-05-18 | 2022-11-24 | 中国科学院深圳先进技术研究院 | Gesture classification and recognition method and application thereof |
CN113312994A (en) * | 2021-05-18 | 2021-08-27 | 中国科学院深圳先进技术研究院 | Gesture classification recognition method and application thereof |
CN113205074A (en) * | 2021-05-29 | 2021-08-03 | 浙江大学 | Gesture recognition method fusing multi-mode signals of myoelectricity and micro-inertia measurement unit |
CN113609923A (en) * | 2021-07-13 | 2021-11-05 | 中国矿业大学 | Attention-based continuous sign language sentence recognition method |
CN113729738A (en) * | 2021-09-13 | 2021-12-03 | 武汉科技大学 | Construction method of multi-channel electromyographic feature image |
CN113729738B (en) * | 2021-09-13 | 2024-04-12 | 武汉科技大学 | Construction method of multichannel myoelectricity characteristic image |
CN113627401A (en) * | 2021-10-12 | 2021-11-09 | 四川大学 | Myoelectric gesture recognition method of feature pyramid network fused with double-attention machine system |
CN113934302B (en) * | 2021-10-21 | 2024-02-06 | 燕山大学 | Myoelectric gesture recognition method based on SeNet and gating time sequence convolution network |
CN113934302A (en) * | 2021-10-21 | 2022-01-14 | 燕山大学 | Myoelectric gesture recognition method based on SeNet and gating time sequence convolution network |
CN114343679A (en) * | 2021-12-24 | 2022-04-15 | 杭州电子科技大学 | Surface electromyogram signal upper limb action recognition method and system based on transfer learning |
CN114504333B (en) * | 2022-01-30 | 2023-10-27 | 天津大学 | Wearable vestibule monitoring system based on myoelectricity and application |
CN114504333A (en) * | 2022-01-30 | 2022-05-17 | 天津大学 | Wearable vestibule monitoring system based on myoelectricity and application |
CN114569143A (en) * | 2022-03-03 | 2022-06-03 | 上海交通大学宁波人工智能研究院 | Myoelectric gesture recognition method based on attention mechanism and multi-feature fusion |
CN114931389A (en) * | 2022-04-27 | 2022-08-23 | 福州大学 | Electromyographic signal identification method based on residual error network and graph convolution network |
CN114931389B (en) * | 2022-04-27 | 2024-07-05 | 福州大学 | Myoelectric signal identification method based on residual error network and graph convolution network |
CN116738295A (en) * | 2023-08-10 | 2023-09-12 | 齐鲁工业大学(山东省科学院) | sEMG signal classification method, system, electronic device and storage medium |
CN116738295B (en) * | 2023-08-10 | 2024-04-16 | 齐鲁工业大学(山东省科学院) | sEMG signal classification method, system, electronic device and storage medium |
CN117281528A (en) * | 2023-11-27 | 2023-12-26 | 山东锋士信息技术有限公司 | Multi-lead pulse signal intelligent identification method and system based on deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN108388348B (en) | 2020-11-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108388348A (en) | A kind of electromyography signal gesture identification method based on deep learning and attention mechanism | |
CN105446484B (en) | A kind of electromyography signal gesture identification method based on Hidden Markov Model | |
Zhao et al. | Noise rejection for wearable ECGs using modified frequency slice wavelet transform and convolutional neural networks | |
CN108491077B (en) | Surface electromyographic signal gesture recognition method based on multi-stream divide-and-conquer convolutional neural network | |
CN105654037B (en) | A kind of electromyography signal gesture identification method based on deep learning and characteristic image | |
CN104166548B (en) | Deep learning method based on Mental imagery eeg data | |
CN105426842B (en) | Multiclass hand motion recognition method based on support vector machines and surface electromyogram signal | |
CN104523266B (en) | A kind of electrocardiosignal automatic classification method | |
CN107203692A (en) | The implementation method of atrial fibrillation detection based on depth convolutional neural networks | |
CN108714026A (en) | The fine granularity electrocardiosignal sorting technique merged based on depth convolutional neural networks and on-line decision | |
CN108304917A (en) | A kind of P300 signal detecting methods based on LSTM networks | |
CN105956624B (en) | Mental imagery brain electricity classification method based on empty time-frequency optimization feature rarefaction representation | |
CN108899050A (en) | Speech signal analysis subsystem based on multi-modal Emotion identification system | |
CN107822623A (en) | A kind of driver fatigue and Expression and Action method based on multi-source physiologic information | |
CN106909784A (en) | Epileptic electroencephalogram (eeg) recognition methods based on two-dimentional time-frequency image depth convolutional neural networks | |
CN105488466B (en) | A kind of deep-neural-network and Acoustic Object vocal print feature extracting method | |
CN107736894A (en) | A kind of electrocardiosignal Emotion identification method based on deep learning | |
CN113205074B (en) | Gesture recognition method fusing multi-mode signals of myoelectricity and micro-inertia measurement unit | |
CN110432870A (en) | A kind of sleep signal based on 1D CNN-LSTM method by stages automatically | |
CN110399846A (en) | A kind of gesture identification method based on multichannel electromyography signal correlation | |
CN104799852A (en) | Method for extracting movement imagination electroencephalogram characteristics based on ultralimit learning machine self encoding | |
CN104887263A (en) | Identity recognition algorithm based on heart sound multi-dimension feature extraction and system thereof | |
CN111202517A (en) | Sleep automatic staging method, system, medium and electronic equipment | |
CN108567418A (en) | A kind of pulse signal inferior health detection method and detecting system based on PCANet | |
CN110443309A (en) | A kind of electromyography signal gesture identification method of combination cross-module state association relation model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201124 |
|
CF01 | Termination of patent right due to non-payment of annual fee |