WO2022242133A1 - 一种手势分类识别方法及其应用 - Google Patents

一种手势分类识别方法及其应用 Download PDF

Info

Publication number
WO2022242133A1
WO2022242133A1 PCT/CN2021/138067 CN2021138067W WO2022242133A1 WO 2022242133 A1 WO2022242133 A1 WO 2022242133A1 CN 2021138067 W CN2021138067 W CN 2021138067W WO 2022242133 A1 WO2022242133 A1 WO 2022242133A1
Authority
WO
WIPO (PCT)
Prior art keywords
gesture
classification
recognition method
gesture classification
signal
Prior art date
Application number
PCT/CN2021/138067
Other languages
English (en)
French (fr)
Inventor
郭伟钰
杨永魁
陈瑞
陈超
辛锦瀚
王峥
Original Assignee
中国科学院深圳先进技术研究院
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中国科学院深圳先进技术研究院 filed Critical 中国科学院深圳先进技术研究院
Publication of WO2022242133A1 publication Critical patent/WO2022242133A1/zh

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
    • G06V40/20—Movements or behaviour, e.g. gesture recognition
    • G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • 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
    • 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
    • G06N3/08—Learning methods

Definitions

  • the present application belongs to the technical field of data classification, and in particular relates to a gesture classification and recognition method and its application.
  • gesture is a very important perception channel in human-computer interaction.
  • identifying the types of gestures that can interact with computers is of great significance for the current research on the control of prosthetics.
  • sEMG surface electromyography
  • the sEMG signal is an important bioelectrical signal generated with the movement of human muscles, and changes in the signal on the surface of the muscle are recorded through electrodes.
  • Different gestures correspond to different muscle movements, and the sEMG signals generated are also different.
  • the sEMG signals have great application value in gesture classification.
  • Gesture classification and recognition based on sEMG signals has become one of the research hotspots in the field of prosthesis control and rehabilitation training, and future gesture recognition will also be widely used in the fields of sports medicine and clinical muscle diagnosis.
  • the sEMG signal is a time series signal.
  • the Recurrent Neural Network (RNN) is compared with the machine learning network and the convolutional neural network. It uses the recurrent neural network to memorize and learn the sequence data of the signal. , is a good tool for dealing with sequence problems.
  • the cycle of RNN is only a simple linear relationship, there is a long-term dependence problem. In the multiple iterations of the calculation process, the multiplication of coefficients will become smaller and smaller, which will indirectly lead to the loss of data information with long distances.
  • This application provides A gesture classification and recognition method and its application.
  • the present application provides a gesture classification and recognition method, the method comprising the following steps: acquiring surface electromyographic signals; performing feature extraction on the surface electromyographic signals to obtain gesture feature sequences and gesture types; The gesture feature sequence and the gesture type are input into a recurrent gate circuit neural network training to obtain a classification model, and the classification model is used to realize gesture classification and recognition.
  • obtaining surface electromyographic signals includes wiping the experimental equipment and the skin surface of the subject with alcohol, placing the electrodes for collecting electromyographic signals on the skin surface of the subject, and collecting the electromyographic signals of different arms in real time. Muscle signal changes brought about by hand gestures.
  • Another embodiment provided by the present application is: there are 8 electrodes, and the 8 electrodes are equidistantly distributed on the forearm.
  • the surface electromyography signal is transmitted to a smart terminal through a wireless device for data processing.
  • processing the surface electromyography signal includes converting the surface electromyography signal into a digital signal, and performing feature value extraction on the digital signal.
  • Another implementation manner provided by the present application is: the surface electromyography signal is input to the smart terminal by using a sliding windowing method.
  • the frequency of the collected surface electromyography signal is 2kHZ
  • the width of the sliding window is 100ms
  • the sliding step is 0.5ms.
  • Another implementation manner provided by the present application is: the feature value extraction is performed by calculating a root mean square value.
  • the recurrent gate circuit neural network includes a fully connected layer.
  • the present application also provides a gesture classification and recognition method, which is applied to a human-computer interaction system.
  • the gesture classification and recognition method provided in the present application is a gesture classification and recognition method of surface electromyography (sEMG) based on a GRU model.
  • sEMG surface electromyography
  • the gesture classification and recognition method provided in this application is to process the collected muscle electrode signals, and input the processed gesture feature sequence and gesture type into the recurrent gate circuit neural network (Gate Recurrent Unit, GRU) for training to obtain the classification Model, in order to realize gesture classification and recognition.
  • GRU recurrent gate circuit neural network
  • this application greatly avoids the problems of overfitting and underfitting in the gradient calculation process, improves the robustness, and greatly reduces the amount of calculation and waste of resources. .
  • the gesture classification and recognition system provided in this application is a gesture classification and recognition algorithm based on a recurrent gate circuit neural network (Gate Recurrent Unit, GRU), which improves the accuracy of prediction classification.
  • GRU recurrent gate circuit neural network
  • GRU is an improved neural network of the cyclic neural network, which effectively avoids the dependence of RNN on data processing, and it can also avoid the traditional machine learning training of long sequence signal data.
  • problems such as gradient disappearance and gradient explosion in the stage.
  • the structure of GRU allows the network to adaptively capture dependencies from a large number of data sequences without discarding information in the early part of the sequence.
  • GRU is similar to the long-term short-term memory neural network LSTM, but part of the state is removed, and the hidden state is directly used for information transmission.
  • the GRU only contains the reset gate and the update gate, which greatly simplifies the calculation amount.
  • the gesture classification and recognition method provided in this application has high accuracy and strong robustness.
  • Fig. 1 is the schematic diagram of the experimental equipment device of the present application.
  • Fig. 2 is a schematic flow chart of the gesture classification and recognition method of the present application
  • FIG. 3 is a schematic diagram of comparison of results of the gesture classification and recognition methods of the present application.
  • the present application provides a gesture classification and recognition method, the method comprising the following steps: acquiring surface electromyographic signals; performing feature extraction on the surface electromyographic signals to obtain gesture feature sequences and gesture types; The gesture feature sequence and the gesture type are input into the circular gate circuit neural network training to obtain a classification model, and the classification model is used to realize gesture classification recognition.
  • obtaining the surface electromyographic signal includes wiping the experimental equipment and the skin surface of the subject with alcohol, placing the electrode for collecting the electromyographic signal on the skin surface of the subject, and collecting the muscle signal changes brought about by gestures of different arms in real time .
  • the 8 electrodes there are 8 electrodes, and the 8 electrodes are equidistantly distributed on the forearm.
  • the surface electromyography signal is transmitted to the smart terminal through a wireless device for data processing.
  • the smart terminal here is a computer, a tablet computer, a mobile phone or other electronic devices capable of data processing.
  • processing the surface electromyography signal includes converting the surface electromyography signal into a digital signal, and performing feature value extraction on the digital signal.
  • the surface electromyographic signal is input to the smart terminal by using a sliding window method.
  • the frequency of the collected surface electromyography signal is 2kHZ
  • the width of the sliding window is 100ms
  • the sliding step is 0.5ms. It can greatly increase the amount of data, effectively avoid the over-fitting problem in the training process, and also improve the data processing speed.
  • the feature value extraction is performed by calculating the root mean square value.
  • the amplitude change of the EMG signal can be better described.
  • loop gate circuit neural network includes a fully connected layer.
  • the present application also provides a gesture classification and recognition method, which is applied to a human-computer interaction system.
  • the application provides a gesture classification method based on the GRU model of surface electromyography signal acquisition, and the specific implementation steps include the following steps:
  • Step 1 Collect the surface EMG signals of different gestures of the subjects
  • Step 2 Processing of Surface EMG Signals
  • this experiment adopts the sliding window method.
  • the window contains 100ms of data, and the window slides backward for 0.5ms each time.
  • this application scrambles the data to speed up the convergence of the model.
  • Step 3 Classification and recognition of gesture actions
  • the gesture signals and corresponding gesture categories after feature extraction are input into the designed GRU neural network, and the neural network model is obtained through training as a gesture classifier.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Psychiatry (AREA)
  • Social Psychology (AREA)
  • Human Computer Interaction (AREA)
  • Multimedia (AREA)
  • User Interface Of Digital Computer (AREA)

Abstract

一种手势分类识别方法，属于数据分类技术领域，所述方法包括：获取表面肌电信号；对所述表面肌电信号进行特征提取得到手势特征序列和手势类型；将所述手势特征序列和所述手势类型输入循环门电路神经网络训练获得分类模型，采用所述分类模型实现手势分类识别。该方法解决了目前关于sEMG信号的手势分类识别算法，识别准确度低，训练模型过程中存在过拟合与欠拟合、梯度消失、鲁棒性差，训练时间长的问题，提高了预测分类的准确率。

Description

一种手势分类识别方法及其应用 技术领域
本申请属于数据分类技术领域，特别是涉及一种手势分类识别方法及其应用。
 背景技术
随着科学技术的发展，新的人机交互方法越来越受到研究人员的关注。手势作为一种生活中自然直观的交互手段，是人机交互中非常重要的感知渠道。在人机交互中，识别可与计算机进行交互的手势类型，对于当前对假肢的控制研究具有十分重要的意义。
与通过计算机视觉实现人机交互的方式相比，当前的研究使用表面肌电信号（surface electromyography， sEMG）可以有效的避免因光照等物理因素带来的影响。sEMG信号是随着人的肌肉运动而产生的重要的生物电信号，通过电极记录在肌肉表面的信号的变化。不同的手势运动对应不同的肌肉运动，产生的sEMG信号也不同，sEMG信号在手势动作分类中具有很大的应用价值。基于sEMG信号的手势分类识别已经成为假体控制和康复训练领域的研究热点之一，且未来手势动作的识别也将被广泛应用于运动医学以及临床肌肉诊断等领域。
sEMG信号是一种时间序列信号，在处理序列信号的问题上，循环神经网络（RecurrentNeuralNetwork，RNN）相比于机器学习网络和卷积神经网络，它通过循环神经网络来记忆和学习信号的序列数据，是一种很好的处理序列问题的工具。但是，由于RNN的循环只是简单的线性关系，存在长期的依赖问题，在计算过程的多次迭代中，系数相乘会越来越小，会间接导致距离较远的数据信息的丢失。
但是当前的一些关于sEMG信号的手势分类识别算法，识别准确度低，训练模型过程中还会存在过拟合与欠拟合、梯度消失、鲁棒性差，训练时间长。
 技术问题
基于当前的一些关于sEMG信号的手势分类识别算法，识别准确度低，训练模型过程中还会存在过拟合与欠拟合、梯度消失、鲁棒性差，训练时间长的问题，本申请提供了一种手势分类识别方法及其应用。
 技术解决方案
为了达到上述的目的，本申请提供了一种手势分类识别方法，所述方法包括如下步骤：获取表面肌电信号；对所述表面肌电信号进行特征提取得到手势特征序列和手势类型；将所述手势特征序列和所述手势类型输入循环门电路神经网络训练获得分类模型，采用所述分类模型实现手势分类识别。
本申请提供的另一种实施方式为：获取表面肌电信号包括采用酒精擦拭实验设备和被试者皮肤表面，将采集肌电信号的电极放置在被试者的皮肤表面，实时采集不同手臂的手势动作带来的肌肉信号变化。
本申请提供的另一种实施方式为：所述电极为8个，所述8个电极等距分布在前臂上。
本申请提供的另一种实施方式为：所述表面肌电信号通过无线装置传输至智能终端进行数据处理。
本申请提供的另一种实施方式为：对所述表面肌电信号进行处理包括将所述表面肌电信号转换为数字信号，对所述数字信号进行特征值提取。
本申请提供的另一种实施方式为：所述表面肌电信号采用滑动加窗法输入至所述智能终端。
本申请提供的另一种实施方式为：所述采集表面肌电信号频率为2kHZ，所述滑动窗口宽度为100ms，滑动步长为0.5ms。
本申请提供的另一种实施方式为：所述特征值提取通过计算均方根值进行提取。
本申请提供的另一种实施方式为：所述循环门电路神经网络包括全连接层。
本申请还提供一种手势分类识别方法，将所述的手势分类识别方法应用于人机交互系统。
 有益效果
与现有技术相比，本申请提供的一种手势分类识别方法及其应用的有益效果在于：
本申请提供的手势分类识别方法，为一种基于GRU模型的表面肌电信号（surfaceelectromyography，sEMG）的手势分类识别方法。
本申请提供的手势分类识别方法，通过将采集到的肌肉电极信号进行数据处理，并将处理后的手势特征序列和手势类型输入循环门电路神经网络（Gate Recurrent Unit，GRU）训练，获得分类的模型，以此来实现手势分类识别。本申请对比于之前的机器学习分类算法，极大的避免了梯度计算过程中出现的过拟合和欠拟合的问题，提高鲁棒性同时，极大地减少了计算量，减少了资源的浪费。
本申请提供的手势分类识别系统，为基于循环门电路神经网络（Gate Recurrent Unit，GRU）的手势分类识别算法，提高了预测分类的准确率。
本申请提供的手势分类识别方法，GRU是循环神经网络的一种改良神经网络，有效的规避了RNN在数据处理上的依赖问题，它还可以避免以往长序列信号的数据在传统的机器学习训练阶段出现的梯度消失和梯度爆炸等问题。
本申请提供的手势分类识别方法，GRU的结构允许网络自适应地从大量数据序列中捕获依赖项，而不丢弃序列早期部分的信息。GRU和长短期记忆神经网络LSTM相似，但是去掉了部分状态，直接使用隐藏状态来进行信息的传递。相比于循环神经网络，GRU 只包含重置门和更新门，极大地简化了运算量。
本申请提供的手势分类识别方法，准确率高，鲁棒性强。
 附图说明
图1是本申请的实验设备装置示意图；
图2是本申请的手势分类识别方法流程示意图；
图3是本申请的手势分类识别方法结果对比示意图。
 本发明的实施方式
在下文中，将参考附图对本申请的具体实施例进行详细地描述，依照这些详细的描述，所属领域技术人员能够清楚地理解本申请，并能够实施本申请。在不违背本申请原理的情况下，各个不同的实施例中的特征可以进行组合以获得新的实施方式，或者替代某些实施例中的某些特征，获得其它优选的实施方式。
参见图1～3，本申请提供一种手势分类识别方法，所述方法包括如下步骤：获取表面肌电信号；对所述表面肌电信号进行特征提取得到手势特征序列和手势类型；将所述手势特征序列和所述手势类型输入循环门电路神经网络训练获得分类模型，采用所述分类模型实现手势分类识别。
进一步地，获取表面肌电信号包括采用酒精擦拭实验设备和被试者皮肤表面，将采集肌电信号的电极放置在被试者的皮肤表面，实时采集不同手臂的手势动作带来的肌肉信号变化。
进一步地，所述电极为8个，所述8个电极等距分布在前臂上。
进一步地，所述表面肌电信号通过无线装置传输至智能终端进行数据处理。这里的智能终端为计算机、平板电脑、手机或者其他可以实现数据处理的电子设备。
进一步地，对所述表面肌电信号进行处理包括将所述表面肌电信号转换为数字信号，对所述数字信号进行特征值提取。
进一步地，所述表面肌电信号采用滑动加窗法输入至所述智能终端。
进一步地，所述采集表面肌电信号频率为2kHZ，所述滑动窗口宽度为100ms，滑动步长为0.5ms。可以极大的提升数据量，有效避免训练过程中出现的过拟合问题，也会提升数据处理速度。
进一步地，所述特征值提取通过计算均方根值进行提取。可以更好的描述肌电信号的幅值变化。
进一步地，所述循环门电路神经网络包括全连接层。
本申请还提供一种手势分类识别方法，将所述的手势分类识别方法应用于人机交互系统。
实施例
本申请提供一种基于表面肌电信号采集的GRU模型的手势分类方法，具体实施步骤包括以下步骤：
步骤1：采集被试者不同手势的表面肌电信号
使用酒精擦拭实验设备和被试者的皮肤表面，以避免皮肤表皮和油脂带来的噪声信号的干扰，将采集肌电信号的电极放置在被试者的皮肤表面，8个电极等距分布在前臂上，实时采集不同手臂的手势动作带来的肌肉信号的变化，实验装置图如图1所示。在实验过程中，本申请采集多种手势。并且，为减少采集过程中，由于被试者的肌肉疲劳带来的误差，本申请各种手势重复6次，每次手势动作持续5s，每次间隔休息5s。采集到的sEMG信号通过无线装置传输到计算机。
步骤2：表面肌电信号的处理
（a）将传输的模拟信号转换为数字信号；
（b）为了获得更多的实验数据，本实验采取了滑动加窗的方法，窗口中包含100ms的数据，窗口每次向后滑动0.5ms.前后两个窗口中的数据有99.5ms的重叠。由于实验设备的采集信号的频率是2kHZ，本申请设置的滑动窗口的宽度是100ms，滑动的步长是0.5ms。同时在训练之前，本申请对数据进行打乱，加速模型收敛速度。
（c）通过计算均方根值RMS，来进行特征值的提取。
步骤3：手势动作的分类识别
将特征提取后的手势信号和相应的手势类别，输入至设计好的GRU神经网络，并通过训练获得神经网络模型作为手势分类器。
对比基于传统机器学习的识别分类方法，例如决策树（Decision Tree）、随机森林（Random Forest）、支持向量机（SVM）、K-近邻算法（KNN）、朴素贝叶斯（NB），本申请在进行手势分类时，对手臂不同姿势的鲁棒性更高，且手势分类准确度更高，可以服务于场景更为复杂的人机交互系统。
通过实验验证了本申请的可行性。本实验均采用pytorch框架实现，在上述所采集的8位受试者的肌电信号上验证了本方法的性能。P1、P2和P3分别代表手臂平放在桌上、手臂与桌面呈45度角放在桌上和手臂平行放置在空中。训练数据仅为P1手臂姿势下前部分数据，测试数据为P1、P2和P3手臂姿势下的后部分数据。实验结果如下图所示，本申请所提出的基于GRU的手势分类方法准确度高达0.96，比其他基于传统机器学习的方法都高。并且，对于P2与P3的手臂姿势，本申请的方法的手势分类准确度下降最小，即鲁棒性最强。
尽管在上文中参考特定的实施例对本申请进行了描述，但是所属领域技术人员应当理解，在本申请公开的原理和范围内，可以针对本申请公开的配置和细节做出许多修改。本申请的保护范围由所附的权利要求来确定，并且权利要求意在涵盖权利要求中技术特征的等同物文字意义或范围所包含的全部修改。

Claims (10)

  1. 一种手势分类识别方法，其特征在于：所述方法包括如下步骤：
    获取表面肌电信号；对所述表面肌电信号进行特征提取得到手势特征序列和手势类型；将所述手势特征序列和所述手势类型输入循环门电路神经网络训练获得分类模型，采用所述分类模型实现手势分类识别。
  2. 如权利要求1所述的手势分类识别方法，其特征在于：获取表面肌电信号包括采用酒精擦拭实验设备和被试者皮肤表面，将采集肌电信号的电极放置在被试者的皮肤表面，实时采集不同手臂的手势动作带来的肌肉信号变化。
  3. 如权利要求2所述的手势分类识别方法，其特征在于：所述电极为8个，所述8个电极等距分布在前臂上。
  4. 如权利要求2所述的手势分类识别方法，其特征在于：所述表面肌电信号通过无线装置传输至智能终端进行数据处理。
  5. 如权利要求1所述的手势分类识别方法，其特征在于：对所述表面肌电信号进行处理包括将所述表面肌电信号转换为数字信号，对所述数字信号进行特征值提取。
  6. 如权利要求4所述的手势分类识别方法，其特征在于：所述表面肌电信号采用滑动加窗法输入至所述智能终端。
  7. 如权利要求6所述的手势分类识别方法，其特征在于：所述采集表面肌电信号频率为2kHZ，所述滑动窗口宽度为100ms，滑动步长为0.5ms。
  8. 如权利要求5所述的手势分类识别方法，其特征在于：所述特征值提取通过计算均方根值进行提取。
  9. 如权利要求1～8中任一项所述的手势分类识别方法，其特征在于：所述循环门电路神经网络包括全连接层。
  10. 一种手势分类识别方法，其特征在于：将权利要求1～9中任一项所述的手势分类识别方法应用于人机交互系统。
PCT/CN2021/138067 2021-05-18 2021-12-14 一种手势分类识别方法及其应用 WO2022242133A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110542132.7A CN113312994A (zh) 2021-05-18 2021-05-18 一种手势分类识别方法及其应用
CN202110542132.7 2021-05-18

Publications (1)

Publication Number Publication Date
WO2022242133A1 true WO2022242133A1 (zh) 2022-11-24

Family

ID=77373401

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/138067 WO2022242133A1 (zh) 2021-05-18 2021-12-14 一种手势分类识别方法及其应用

Country Status (2)

Country Link
CN (1) CN113312994A (zh)
WO (1) WO2022242133A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113312994A (zh) * 2021-05-18 2021-08-27 中国科学院深圳先进技术研究院 一种手势分类识别方法及其应用
CN114159079B (zh) * 2021-11-18 2023-05-02 中国科学院合肥物质科学研究院 基于特征提取和gru深度学习模型的多类型肌肉疲劳检测方法

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160091965A1 (en) * 2014-09-30 2016-03-31 Microsoft Corporation Natural motion-based control via wearable and mobile devices
US20180144245A1 (en) * 2016-11-23 2018-05-24 Microsoft Technology Licensing, Llc Differential recurrent neural network
CN108388348A (zh) * 2018-03-19 2018-08-10 浙江大学 一种基于深度学习和注意力机制的肌电信号手势识别方法
CN110537922A (zh) * 2019-09-09 2019-12-06 北京航空航天大学 基于深度学习的人体行走过程下肢运动识别方法及系统
CN111209885A (zh) * 2020-01-13 2020-05-29 腾讯科技（深圳）有限公司 一种手势信息处理方法、装置、电子设备及存储介质
CN111898526A (zh) * 2020-07-29 2020-11-06 南京邮电大学 基于多流卷积神经网络的肌电手势识别方法
CN112783327A (zh) * 2021-01-29 2021-05-11 中国科学院计算技术研究所 基于表面肌电信号进行手势识别的方法及系统
CN113312994A (zh) * 2021-05-18 2021-08-27 中国科学院深圳先进技术研究院 一种手势分类识别方法及其应用

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10061389B2 (en) * 2014-06-03 2018-08-28 Beijing TransBorder Information Technology Co., Ltd. Gesture recognition system and gesture recognition method
US11493993B2 (en) * 2019-09-04 2022-11-08 Meta Platforms Technologies, Llc Systems, methods, and interfaces for performing inputs based on neuromuscular control
CN110598676B (zh) * 2019-09-25 2022-08-02 南京邮电大学 基于置信度得分模型的深度学习手势肌电信号识别方法
CN111553307B (zh) * 2020-05-08 2023-03-24 中国科学院合肥物质科学研究院 一种融合生物电阻抗信息和肌电信息的手势识别系统

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160091965A1 (en) * 2014-09-30 2016-03-31 Microsoft Corporation Natural motion-based control via wearable and mobile devices
US20180144245A1 (en) * 2016-11-23 2018-05-24 Microsoft Technology Licensing, Llc Differential recurrent neural network
CN108388348A (zh) * 2018-03-19 2018-08-10 浙江大学 一种基于深度学习和注意力机制的肌电信号手势识别方法
CN110537922A (zh) * 2019-09-09 2019-12-06 北京航空航天大学 基于深度学习的人体行走过程下肢运动识别方法及系统
CN111209885A (zh) * 2020-01-13 2020-05-29 腾讯科技（深圳）有限公司 一种手势信息处理方法、装置、电子设备及存储介质
CN111898526A (zh) * 2020-07-29 2020-11-06 南京邮电大学 基于多流卷积神经网络的肌电手势识别方法
CN112783327A (zh) * 2021-01-29 2021-05-11 中国科学院计算技术研究所 基于表面肌电信号进行手势识别的方法及系统
CN113312994A (zh) * 2021-05-18 2021-08-27 中国科学院深圳先进技术研究院 一种手势分类识别方法及其应用

Also Published As

Publication number Publication date
CN113312994A (zh) 2021-08-27

Similar Documents

Publication Publication Date Title
WO2021143353A1 (zh) 一种手势信息处理方法、装置、电子设备及存储介质
Bao et al. A CNN-LSTM hybrid model for wrist kinematics estimation using surface electromyography
Yang et al. Dynamic gesture recognition using surface EMG signals based on multi-stream residual network
Jiang et al. A method of recognizing finger motion using wavelet transform of surface EMG signal
WO2022242133A1 (zh) 一种手势分类识别方法及其应用
CN110555468A (zh) 一种联合递归图和cnn的脑电信号识别方法及系统
CN108681396A (zh) 基于脑-肌电双模态神经信号的人机交互系统及其方法
CN111870242A (zh) 一种基于肌电信号的智能手势动作生成方法
CN113111831A (zh) 一种基于多模态信息融合的手势识别技术
Li et al. EEG signal classification method based on feature priority analysis and CNN
CN113901891A (zh) 帕金森病握拳任务的评估方法及系统、存储介质及终端
CN106845348B (zh) 一种基于手臂表面肌电信号的手势识别方法
Wang et al. A deep learning approach using attention mechanism and transfer learning for electromyographic hand gesture estimation
Xiong et al. A Global and Local Feature fused CNN architecture for the sEMG-based hand gesture recognition
Fan et al. Robust neural decoding for dexterous control of robotic hand kinematics
CN109498362A (zh) 一种偏瘫患者手部运动功能康复训练装置及模型训练方法
AlOmari et al. Novel hybrid soft computing pattern recognition system SVM–GAPSO for classification of eight different hand motions
Massa et al. Graph neural networks for hd emg-based movement intention recognition: An initial investigation
Kumar et al. A critical review on hand gesture recognition using semg: Challenges, application, process and techniques
Chen et al. SEMG-based gesture recognition using GRU with strong robustness against forearm posture
CN108255303B (zh) 一种基于自制数据手套的手势识别方法
CN114098768B (zh) 基于动态阈值和EasyTL的跨个体表面肌电信号手势识别方法
Fu et al. Identification of finger movements from forearm surface EMG using an augmented probabilistic neural network
Fazil et al. CNN-based controller for multi-DoF prosthetic wrist using SEMG data during activities of daily living
Zhou et al. Time–frequency feature transform suite for deep learning-based gesture recognition using sEMG signals

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21940576

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21940576

Country of ref document: EP

Kind code of ref document: A1