Gesture Recognition Toolkit (GRT) is a cross-platform, open-source, C++ machine learning library designed for real-time gesture recognition.

Overview

Gesture Recognition Toolkit (GRT)

The Gesture Recognition Toolkit (GRT) is a cross-platform, open-source, C++ machine learning library designed for real-time gesture recognition.

Build Status:

  • Master branch:
    • Master Build Status
  • Dev branch:
    • Dev Build Status

Current version: 0.2.5

Key things to know about the GRT:

  • The toolkit consists of two parts: a comprehensive C++ API and a front-end graphical user interface (GUI). You can access the source code for both the C++ API and GUI in this repository, a precompiled version of the GUI can be downloaded here
  • Both the C++ API and GUI are designed to work with real-time sensor data, but they can also be used for more conventional offline machine-learning tasks
  • The input to the GRT can be any N-dimensional floating-point vector - this means you can use the GRT with Cameras, Kinect, Leap Motion, accelerometers, or any other custom sensor you might have built
  • The toolkit defines a generic Float type, this defaults to double precision float, but can easily be changed to single precision via the main GRT Typedefs header
  • The precision of the GRT VectorFloat and MatrixFloat classes is automatically updated based on the main Float precision
  • The toolkit reserves the class label value of zero as a special null gesture class label for automatic gesture spotting, so if you want to use gesture spotting avoid labeling any of your gestures with the class label of zero
  • Training data and models are saved as custom .grt files. These consist of a simple header followed by the main dataset. In addition to the grt files, you can also import/export data via CSV files by using the .csv file extension when saving/loading files
  • Almost all the GRT classes support the following functions:
    • predict( ... ): uses the input data (...) and a pre-trained model to perform a prediction, such as classification or regression
    • train( ... ): uses the input data (...) to train a new model that can then be used for real-time prediction
    • save( ... ): saves a model or dataset to a file. The file format can be a custom GRT file (.grt) or a CSV file (.csv)
    • load( ... ): loads a pre-trained model or dataset from a file. The file format can be a custom GRT file (.grt) or a CSV file (.csv)
    • reset(): resets a module, for example resetting a filter module would clear the values in its history buffer and sets them to zero
    • clear(): clears a module, removing all pre-trained models, weights, etc.. For example, clearing a filter module would delete the filter coefficients, history buffer, etc.
  • Functions with an underscore, such as train_( ... ), pass the input arguments as references and are therefore more efficient to use with very large datasets

Core Resources

Core Algorithms

The GRT supports a wide number of supervised and unsupervised machine learning algorithms for classification, regression, and clustering, including:

In addition to the machine learning algorithms above, the toolkit also includes a large number of algorithms for preprocessing, feature extraction, and post processing.

See the wiki for more details.

GRT Extensions

There are now several extensions and third-party applications that use the GRT as the backend machine learning system, these include:

  • ofGrt: an extension of the GRT for openFrameworks
  • ml-lib, by Ali Momeni and Jamie Bullock: ml-lib is a library of machine learning externals for Max and Pure Data, designed to work on a variety of platforms including OS X, Windows, Linux, on Intel and ARM architectures.
  • ESP, by David A. Mellis and Ben Zhang: An interactive application that aims to help novices make sophisticated use of sensors in interactive projects through the application of machine learning. The system is built using openFrameworks and has several interesting examples built for Arduino sensor modules and more generic input data streams (e.g., network data).
  • Android Port: you can find a specific Android port of the GRT here.

GRT Architecture

To support flexibility while maintaining consistency, the GRT uses an object-oriented modular architecture. This architecture is built around a set of core modules and a central gesture recognition pipeline.

The input to both the modules and pipeline consists of a N-dimensional floating-point vector, making the toolkit flexible to the type of input signal. The algorithms in each module can be used as standalone classes; alternatively, a pipeline can be used to chain modules together to create a more sophisticated gesture-recognition system. The GRT includes modules for preprocessing, feature extraction, clustering, classification, regression and post processing.

The toolkit's source code is structured as follows:

  • ClassificationModules: Contains all the GRT classification algorithms, such as AdaBoost, Naive Bayes, K-Nearest Neighbor, Support Vector Machines, and more.
  • ClusteringModules: Contains all the GRT clustering algorithms, including K-Means, Gaussian Mixture Models, and Self-Organizing Maps.
  • ContextModules: Contains all the GRT context modules, these are modules that can be connected to a gesture recognition pipeline to input additional context to a real-time classification system.
  • CoreAlgorithms: Contains a number of algorithms that are used across the GRT, such as Particle Filters, Principal Component Analysis, and Restricted Boltzmann Machines.
  • CoreModules: Contains all the GRT base classes, such as MLBase, Classifier, FeatureExtraction, etc..
  • DataStructures: Contains all the GRT classes for recording, saving and loading datasets.
  • FeatureExtractionModules: Contains all the GRT feature extraction modules. These include FFT, Quantizers, and TimeDomainFeatures.
  • PostProcessingModules: Contains all the GRT post-processing modules, including ClassLabelFilter and ClassLabelTimeoutFilter.
  • PreProcessingModules: Contains all the GRT pre-processing modules, including LowPassFilter, HighPassFilter, DeadZone, and much more.
  • RegressionModules: Contains all the GRT regression modules, such as MLP Neural Networks, Linear Regression, and Logistic Regression.
  • Util: Contains a wide range of supporting classes, such as Logging, Util, TimeStamp, Random, and Matrix.

Getting Started Example

This example demonstrates a few key components of the GRT, such as:

  • how to load a dataset from a file (e.g., a CSV file)
  • how to split a dataset into a training and test dataset
  • how to set up a new Gesture Recognition Pipeline and add a classification algorithm to the pipeline
  • how to use a training dataset to train a new classification model
  • how to save/load a trained pipeline to/from a file
  • how to use an automatically test dataset to test the accuracy of a classification model
  • how to use a manually test dataset to test the accuracy of a classification model
  • how to print detailed test results, such as precision, recall, and the confusion matrix

You can find this source code and a large number of other examples and tutorials in the GRT examples folder.

You should run this example with one argument, pointing to the file you want to load, for example:

 ./example my_data.csv

You can find several examples CSV files and other datasets in the main GRT data directory.

classLabels = pipeline.getClassLabels(); //Print out the precision cout << "Precision: "; for (UINT k=0; k
//Include the main GRT header
#include <GRT/GRT.h>
using namespace GRT;
using namespace std;

int main (int argc, const char * argv[]) {
  //Parse the training data filename from the command line
  if (argc != 2) {
    cout << "Error: failed to parse data filename from command line. ";
    cout << "You should run this example with one argument pointing to a data file\n";
    return EXIT_FAILURE;
  }
  const string filename = argv[1];

  //Load some training data from a file
  ClassificationData trainingData;

  cout << "Loading dataset..." << endl;
  if (!trainingData.load(filename)) {
    cout << "ERROR: Failed to load training data from file\n";
    return EXIT_FAILURE;
  }

  cout << "Data Loaded" << endl;

  //Print out some stats about the training data
  trainingData.printStats();

  //Partition the training data into a training dataset and a test dataset. 80 means that 80%
  //of the data will be used for the training data and 20% will be returned as the test dataset
  cout << "Splitting data into training/test split..." << endl;
  ClassificationData testData = trainingData.split(80);

  //Create a new Gesture Recognition Pipeline
  GestureRecognitionPipeline pipeline;

  //Add a KNN classifier to the pipeline with a K value of 10
  pipeline << KNN(10);

  //Train the pipeline using the training data
  cout << "Training model..." << endl;
  if (!pipeline.train(trainingData)) {
    cout << "ERROR: Failed to train the pipeline!\n";
    return EXIT_FAILURE;
  }

  //Save the pipeline to a file
  if (!pipeline.save("HelloWorldPipeline.grt")) {
    cout << "ERROR: Failed to save the pipeline!\n";
    return EXIT_FAILURE;
  }

  //Load the pipeline from a file
  if (!pipeline.load("HelloWorldPipeline.grt")) {
    cout << "ERROR: Failed to load the pipeline!\n";
    return EXIT_FAILURE;
  }

  //Test the pipeline using the test data
  cout << "Testing model..." << endl;
  if (!pipeline.test(testData)) {
    cout << "ERROR: Failed to test the pipeline!\n";
    return EXIT_FAILURE;
  }

  //Print some stats about the testing
  cout << "Pipeline Test Accuracy: " << pipeline.getTestAccuracy() << endl;

  //Manually project the test dataset through the pipeline
  Float testAccuracy = 0.0;
  for (UINT i=0; i
    getNumSamples(); i++) {
    pipeline.
    predict(testData[i].
    getSample());

    
    if (testData[i].
    getClassLabel() == pipeline.
    getPredictedClassLabel()) {
      testAccuracy++;
    }
  }
  cout << 
    "Manual test accuracy: " << testAccuracy / testData.
    getNumSamples() * 
    100.0 << endl;
   
  
    //Get the vector of class labels from the pipeline
  Vector< UINT > classLabels = pipeline.
    getClassLabels();

  
    //Print out the precision
  cout << 
    "Precision: ";
  
    for (UINT k=
    0; k
    
     getNumClassesInModel(); k++) {
    cout << 
     "\t" << pipeline.
     getTestPrecision(classLabels[k]);
  }cout << endl;

  
     //Print out the recall
  cout << 
     "Recall: ";
  
     for (UINT k=
     0; k
     
      getNumClassesInModel(); k++) {
    cout << 
      "\t" << pipeline.
      getTestRecall(classLabels[k]);
  }cout << endl;

  
      //Print out the f-measure
  cout << 
      "FMeasure: ";
  
      for (UINT k=
      0; k
      
       getNumClassesInModel(); k++) {
    cout << 
       "\t" << pipeline.
       getTestFMeasure(classLabels[k]);
  }cout << endl;

  
       //Print out the confusion matrix
  MatrixFloat confusionMatrix = pipeline.
       getTestConfusionMatrix();
  cout << 
       "ConfusionMatrix: \n";
  
       for (UINT i=
       0; i
       
        getNumRows(); i++) {
    
        for (UINT j=
        0; j
        
         getNumCols(); j++) { cout << confusionMatrix[i][j] << 
         "\t"; }cout << endl; } 
         return EXIT_SUCCESS; }
        
       
      
     
    
   

Tutorials and Examples

You can find a large number of tutorials and examples in the examples folder. You can also find a wide range of examples and references on the main GRT wiki:

http://www.nickgillian.com/wiki/pmwiki.php?n=GRT.GestureRecognitionToolkit

If you build the GRT using CMake, an examples folder will automatically be generated in the build directory after you successfully build the main GRT library. Example applications can then be directly run from this example directory. To run any of the examples, open terminal in the grt/build/examples directory and run:

./ExampleName

where ExampleName is the name of the example application you want to run.

Forum

Note, at the moment the forum server is currently broken, we are working to resolve this. In the meantime, use GitHub issues and pull requests.

You can find the link for the old forum at: http://www.nickgillian.com/forum/

Bugs

Please submit bugs to the github bug tracker.

Contributions

All contributions are welcome, there are several ways in which users can contribute to the toolkit:

  • improving the doxygen generated API (by improving coverage and quality of the current documents in the existing code)
  • improving the higher-level documentation that can be found in the wiki
  • adding new examples or tutorials, or improving the existing ones
  • adding new unit tests, to help ensure the quality of the current functions and catch potential bugs in the future

Please submit pull requests for any contribution.

GRT Floating Point Precision

The GRT defaults to double precision floating point values. The precision of the toolkit is defined by the following Float typedef:

typedef double Float; ///
   

This can easily be changed to single precision accuracy if needed by modifying the main GRT Float typedef value, defined in GRT/Util/GRTTypedefs.h header.

VectorFloat and MatrixFloat Data Structures

The GRT uses two main data structures throughout the toolkit: Vector and Matrix. These are templates and can, therefore, generalize to any C++ class. The main things to know about these data types are:

//Create an integer vector with a size of 3 elements
Vector< int > vec1(3);

//Create a string vector with a size of 2 elements
Vector< string > vec2(2);

//Create a Foo vector with a size of 5 elements
Vector< Foo > vec3(5);
  • Matrix: this provides the base class for storing two dimensional arrays:
//Create an integer matrix with a size of 3x2
Matrix< int > mat1(3,2);

//Create a string matrix with a size of 2x2
Matrix< string > mat2(2,2);

//Create a Foo matrix with a size of 5x3
Matrix< Foo > mat3(5,3);
  • VectorFloat: this provides the main data structure for storing floating point vector data. The precision of VectorFloat will automatically match that of GRT Float.
//Create a new vector with 10 elements
VectorFloat vector( 10 );
for(UINT i=0; i
   1.0; 
}
  
  • MatrixFloat: this provides the main data structure for storing floating point matrix data. The precision of MatrixFloat will automatically match that of GRT Float.
//Create a [5x2] floating point matrix
MatrixFloat matrix(5,2);

//Loop over the data and set the values to a basic incrementing value
UINT counter = 0;
for(UINT i=0; i
   for(UINT j=
   0; j
   
    getNumCols(); j++){
        matrix[i][j] = counter++;
    }
}
   
  

Building the GRT

You can find a CMakeLists file in the build folder that you can use to auto generate a makefile for your machine.

Read the readme file in the build folder to see how to build the GRT as a static library for Linux, OS X, or Windows.

Installing and using the GRT in your C++ projects

See the build directory for details on how to build, install, and use the GRT in your C++ projects.

License

The Gesture Recognition Toolkit is available under an MIT license.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Issues
  • Issue with the GUI not saving training data

    Issue with the GUI not saving training data

    Hi Nick,

    Whenever I run the Processing Classification Example / GRT-GUI following the steps in the tutorial, I an error of "WARNING: Failed to save training data saved to file" appears in the Info Text box in the GUI.

    Additionally, when I click on "Save" in the Data Manager tab, the dialog allows me to click on the "save" button, but when I click on "Load", no file is shown.

    I'm running OS X 10.11.1. Thank you!

    opened by narner 5
  • Failed to resize matrix, rows and cols == zero!

    Failed to resize matrix, rows and cols == zero!

    Hello, sorry to bother. Im getting that error when I try to train an HMM continous model. Any Idea of what might be causing it? I could get to train it with a smaller data set. but with the original size one it throws that error. the data set its 36 dimensions and has 842 training examples and 8 clases, its 15Mbytes.... A DTW model have been train with no errors with this data set

    opened by JairoJs 4
  • CSV File format

    CSV File format

    Hello I am trying to load a data set that is in a csv file, but "No header found" error keeps poping up. where can i get some example of the format a csv file must have to load succesfully in a classification data object

    opened by JairoJs 4
  • PyGRT: Python bindings for GRT via SWIG

    PyGRT: Python bindings for GRT via SWIG

    This implements a Python API for GRT.

    Enables writing Python code like this:

        print("Loading dataset...")
        training_data = GRT.ClassificationData()
        training_data.load(filename)
        print("Data Loaded")
        
        # Print out some stats about the training data
        training_data.printStats()
        
        print("Splitting data into training/test split...")
        test_data = training_data.split(80)
        
        # Create a new Gesture Recognition Pipeline
        pipeline = GRT.GestureRecognitionPipeline()
        
        # Add a KNN classifier to the pipeline with a K value of 10
        knn = GRT.KNN(10)
        pipeline.setClassifier(knn)
    
        print("Training model...")
        pipeline.train(training_data)
    

    I started with the examples to get a good API coverage.

    The Travis build fails because the MLPRegressor unit test fail, i.e. the problem is not with this code, but GRT itself.

    opened by royshil 3
  • Crash without any information

    Crash without any information

    Hi, my program ends crashing when i train the pipeline with this trainingData and ANBC https://transfer.sh/gaR2j/gestures.grt However, with a similar file but using one class it works...

    I'm using Windows 10 with visual studio 2015

    opened by mikaoP 3
  • Adding unit tests for more classifiers

    Adding unit tests for more classifiers

    Hey Nick,

    I started adding in some more unit tests for the Classifier modules. Right now, the test files are only testing the TrainBasicDataset methods, but I'm going to look at other tests as well. Hope this is a helpful start to getting in more tests!

    opened by narner 3
  • Polymorphism with classifiers

    Polymorphism with classifiers

    Hello Nick,

    Since the forum is down, I thought it best to message you here.

    I was planning on using several different classifiers at the same time. For this purposed I defined an array of Classifiers and assigned different subtypes to each index. However, when trying to load the stored models I realised that the function called resides in MLBase and not in each of the appropriate ML Classifiers. Upon inspection I noticed that this function is not virtual and it just returns false. I guess it wasn't intended to be used in that way. Would you mind letting me know how I could achieve polymorphism? Could it be that I need to do an array of pipelines? I just want to use the library the way it was intended to be used. Don't want to go against its design principles.

    Thanks a lot. M

    opened by lumyx2 3
  • OSC in GUI bug

    OSC in GUI bug

    When using just classification mode, the Processing example works fine (keys 'r', '[', ']'). But when switching to timeseries mode, the OSC messages are received, but GUI is not updated and new samples are not saved, therefore it's impossible to perform training on dataset in timeseries mode.

    opened by cyberluke 3
  • What are the

    What are the "Dimensions" at the code?

    I see at the examples that Dimensions always set to 1. Even in getting started page,

    but I use 101 samples (Accelerometer x axis 101 samples) to train the pipeline, trainingData.addSample((UINT) 1 , DataVecot) results with : [WARNING ClassificationData] addSample(const UINT classLabel, VectorFloat &sample) - the size of the new sample (101) does not match the number of dimensions of the dataset (1), setting dimensionality to: 1 [ERROR] process(const VectorFloat &inputVector) - The size of the inputVector (101) does not match that of the filter (1)!

    Help?

    Thanks

    opened by hadarszostak 2
  • Face exception when running Synapse.exe in Windows 8.

    Face exception when running Synapse.exe in Windows 8.

    Hi Nick and others,

    I try to use grt with kinect. And I follow the written instructions to install Synapse and Open Framework.

    However, I face exception when running the Synapse.exe file. (See the attachment below).

    untitled

    I believe my open ni and nite path is correct and I did run the file with admin priviledge but it doesn't help either.

    For your information, I am using Windows 8. I wonder is it better to use Windows 7 instead?

    I understand Mac is a more suitable OS to use Synapse. But, I hope I can run it in Windows.

    Regards, Jimmy

    opened by jingyi2811 2
  • what's the differenece between continuousHMM and discreateHMM

    what's the differenece between continuousHMM and discreateHMM

    Hello, Nickgillian!

    I'm studying the HMM, and read the code of continuousHiddenmarkovmodel.cpp. In the train_(...) function, i didn't find the code to implement Baum-Welch algorithm, but the algorithm is implemented in DiscreteHiddenMarkovModel.cpp. So, i wanna to know what difference is between this two kind of algorithm, and why continuousHMM need not to implement Baum-Welch algorithm? Thanks!

    opened by daniellocust 2
  • Windows crash when clearing model with SVM

    Windows crash when clearing model with SVM

    This bug is specific to SVM and only occurs for me on Windows 10.

    To reproduce:

    • Load a saved SVM model
    • Load another saved SVM model or clear the loaded model

    Behaviour:

    • GRT crashes
    opened by jamiebullock 1
  • GRT GUI Download Link broken

    GRT GUI Download Link broken

    hi, as mentioned in the read me, the GRT GUI Download: http://www.nickgillian.com/wiki/pmwiki.php/GRT/Download

    This above link is broken, could you please help with the binaries link for the same?

    Thanks

    opened by bx5974 0
  • Problems to write and train on Max

    Problems to write and train on Max

    Dear, I hope everybody is safe. It's an urgent situation. I'm facing problem to train using ml.ann object on Max v. 8.1.5 (64 bit - OS High Sierra 10.13.6) and problem to write using svm and ml.ann on Max v. 8.1.8. (OS Catalina 10.15.7)

    Any help would be great. Thanks Sol

    opened by gustavosol 0
  • MLBase::getMinChange() const’ member function declared in class ‘GRT::MLBase’

    MLBase::getMinChange() const’ member function declared in class ‘GRT::MLBase’

    /grt/GRT/CoreModules/MLBase.cpp:262:30: error: no ‘GRT::Float GRT::MLBase::getMinChange() const’ member function declared in class ‘GRT::MLBase’ Float MLBase::getMinChange() const{ ^~~~~ /home/ldd/project_computervision/grt/GRT/CoreModules/MLBase.cpp:316:42: error: no ‘bool GRT::MLBase::getRandomiseTrainingOrder() const’ member function declared in class ‘GRT::MLBase’ bool MLBase::getRandomiseTrainingOrder() const { return randomiseTrainingOrder; } ^~~~~ CMakeFiles/grt.dir/build.make:2150: recipe for target 'CMakeFiles/grt.dir/home/ldd/project_computervision/grt/GRT/CoreModules/MLBase.cpp.o' failed make[2]: *** [CMakeFiles/grt.dir/home/ldd/project_computervision/grt/GRT/CoreModules/MLBase.cpp.o] Error 1 CMakeFiles/Makefile2:1597: recipe for target 'CMakeFiles/grt.dir/all' failed make[1]: *** [CMakeFiles/grt.dir/all] Error 2 Makefile:140: recipe for target 'all' failed make: *** [all] Error 2

    have sovlved Float MLBase::getMinChange() const; bool MLBase::getRandomiseTrainingOrder() const; these two function was annotated

    opened by liudongdong1 4
  • Build issues on OS X 10.15

    Build issues on OS X 10.15

    error: out-of-line definition of 'getMinChange' does not match any declaration in 'GRT::MLBase' Float MLBase::getMinChange() const{ error: out-of-line definition of 'getRandomiseTrainingOrder' does not match any declaration in 'GRT::MLBase' bool MLBase::getRandomiseTrainingOrder() const { return randomiseTrainin...

    In a recent merge, bool getUseValidationSet() const; and bool getRandomiseTrainingOrder() const; were commented out inside GRT/CoreModules/MLBase.h.

    Uncommenting them removed the make errors.

    Was getting errors from fstream inside examples, so had to remove the whole directory for now(since I have to get this working for a project asap). That helped with the make continuing further.

    Then finally got errors from all the cpp files inside tools for the #include <GRT/GRT.h>. Replaced all with #include "../GRT/GRT.h" and that fixed the issue. But I think there was a warning above those includes for this anyway.

    opened by anadi198 1
  • Allow a VectorFloat to be initialised from a vector<Float>

    Allow a VectorFloat to be initialised from a vector

    This allows for more convenient inline adding of feature vectors, for example:

    ClassificationData data;
    
    data.addSample(1, vector<Float>{1, 2, 3, 4, 5});
    
    opened by jamiebullock 0
Releases(v0.2.4)
  • v0.2.4(Dec 5, 2016)

    This version is a snapshot of the Gesture Recognition Toolkit prior to introducing some major updates as the toolkit transitions to version v0.2.4.

    The API for this version can be found at: nickgillian.com/grt/api/0.2.4

    Source code(tar.gz)
    Source code(zip)
  • v0.2.0(Aug 8, 2016)

    This version is a snapshot of the Gesture Recognition Toolkit prior to introducing some major updates as the toolkit transitions to version v0.2.0.

    The API for this version can be found at: nickgillian.com/grt/api/0.2.0

    Source code(tar.gz)
    Source code(zip)
  • v0.1.0(Feb 21, 2016)

    This version is a snapshot of the Gesture Recognition Toolkit prior to introducing some major changes as the toolkit transitions to version v0.1.0.

    The API for this version can be found at: nickgillian.com/grt/api/0.1.0

    Source code(tar.gz)
    Source code(zip)
Owner
Nicholas Gillian
Google and GRT
Nicholas Gillian
In this tutorial, we will use machine learning to build a gesture recognition system that runs on a tiny microcontroller, the RP2040.

Pico-Motion-Recognition This Repository has the code used on the 2 parts tutorial TinyML - Motion Recognition Using Raspberry Pi Pico The first part i

Marcelo Rovai 16 Jun 18, 2022
UVIC ECE 499 Real-Time Gesture Recognition Project

Welcome to GitHub Pages You can use the editor on GitHub to maintain and preview the content for your website in Markdown files. Whenever you commit t

Lyndon Bauto 3 Sep 21, 2019
Insight Toolkit (ITK) is an open-source, cross-platform toolkit for N-dimensional scientific image processing, segmentation, and registration

ITK: The Insight Toolkit C++ Python Linux macOS Windows Linux (Code coverage) Links Homepage Download Discussion Software Guide Help Examples Issue tr

Insight Software Consortium 1k Jun 26, 2022
Vowpal Wabbit is a machine learning system which pushes the frontier of machine learning with techniques such as online, hashing, allreduce, reductions, learning2search, active, and interactive learning.

This is the Vowpal Wabbit fast online learning code. Why Vowpal Wabbit? Vowpal Wabbit is a machine learning system which pushes the frontier of machin

Vowpal Wabbit 8k Jun 19, 2022
Microsoft Cognitive Toolkit (CNTK), an open source deep-learning toolkit

CNTK Chat Windows build status Linux build status The Microsoft Cognitive Toolkit (https://cntk.ai) is a unified deep learning toolkit that describes

Microsoft 17.2k Jun 24, 2022
Microsoft Cognitive Toolkit (CNTK), an open source deep-learning toolkit

The Microsoft Cognitive Toolkit is a unified deep learning toolkit that describes neural networks as a series of computational steps via a directed graph.

Microsoft 17.2k Jun 26, 2022
🐸 Coqui STT is an open source Speech-to-Text toolkit which can run in real time on devices ranging from a Raspberry Pi 4 to high power GPU servers

Coqui STT ( ?? STT) is an open-source deep-learning toolkit for training and deploying speech-to-text models. ?? STT is battle tested in both producti

Coqui.ai 1.4k Jun 27, 2022
An Open-Source Analytical Placer for Large Scale Heterogeneous FPGAs using Deep-Learning Toolkit

DREAMPlaceFPGA An Open-Source Analytical Placer for Large Scale Heterogeneous FPGAs using Deep-Learning Toolkit. This work leverages the open-source A

Rachel Selina Rajarathnam 14 May 30, 2022
An open source machine learning library for performing regression tasks using RVM technique.

Introduction neonrvm is an open source machine learning library for performing regression tasks using RVM technique. It is written in C programming la

Siavash Eliasi 33 May 31, 2022
An open-source, low-code machine learning library in Python

An open-source, low-code machine learning library in Python ?? Version 2.3.6 out now! Check out the release notes here. Official • Docs • Install • Tu

PyCaret 5.9k Jun 27, 2022
Real time monaural source separation base on fully convolutional neural network operates on Time-frequency domain.

Real time monaural source separation base on fully convolutional neural network operates on Time-frequency domain.

James Fung 94 Jun 20, 2022
An Open Source Machine Learning Framework for Everyone

Documentation TensorFlow is an end-to-end open source platform for machine learning. It has a comprehensive, flexible ecosystem of tools, libraries, a

null 165.8k Jun 20, 2022
Deep Scalable Sparse Tensor Network Engine (DSSTNE) is an Amazon developed library for building Deep Learning (DL) machine learning (ML) models

Amazon DSSTNE: Deep Scalable Sparse Tensor Network Engine DSSTNE (pronounced "Destiny") is an open source software library for training and deploying

Amazon Archives 4.4k Jun 26, 2022
A library for creating Artificial Neural Networks, for use in Machine Learning and Deep Learning algorithms.

iNeural A library for creating Artificial Neural Networks, for use in Machine Learning and Deep Learning algorithms. What is a Neural Network? Work on

Fatih Küçükkarakurt 5 Apr 5, 2022
DeepSpeech is an open source embedded (offline, on-device) speech-to-text engine which can run in real time on devices ranging from a Raspberry Pi 4 to high power GPU servers.

Project DeepSpeech DeepSpeech is an open-source Speech-To-Text engine, using a model trained by machine learning techniques based on Baidu's Deep Spee

Mozilla 19.8k Jun 28, 2022
ClanLib is a cross platform C++ toolkit library.

ClanLib ClanLib is a cross platform toolkit library with a primary focus on game creation. The library is Open Source and free for commercial use, und

Kenneth Gangstø 299 Jun 21, 2022
OpenSpeaker is a completely independent and open source speaker recognition project.

OpenSpeaker is a completely independent and open source speaker recognition project. It provides the entire process of speaker recognition including multi-platform deployment and model optimization.

ZY 26 Jun 14, 2022
DeepRTS is a high-performance Real-TIme strategy game for Reinforcement Learning research written in C++

DeepRTS is a high-performance Real-TIme strategy game for Reinforcement Learning research. It is written in C++ for performance, but provides an python interface to better interface with machine-learning toolkits. Deep RTS can process the game with over 6 000 000 steps per second and 2 000 000 steps when rendering graphics. In comparison to other solutions, such as StarCraft, this is over 15 000% faster simulation time running on Intel i7-8700k with Nvidia RTX 2080 TI.

Centre for Artificial Intelligence Research (CAIR) 142 Jun 5, 2022
Lite.AI.ToolKit 🚀🚀🌟: A lite C++ toolkit of awesome AI models such as RobustVideoMatting🔥, YOLOX🔥, YOLOP🔥 etc.

Lite.AI.ToolKit ?? ?? ?? : A lite C++ toolkit of awesome AI models which contains 70+ models now. It's a collection of personal interests. Such as RVM, YOLOX, YOLOP, YOLOR, YoloV5, DeepLabV3, ArcFace, etc.

DefTruth 1.7k Jun 27, 2022