//--------------------------------------------------------------
void ofApp::setup(){
ofSetFrameRate(60);
//Setup the FFT
FFT fft;
fft.init(FFT_WINDOW_SIZE,FFT_HOP_SIZE,1,FFT::RECTANGULAR_WINDOW,true,false,DATA_TYPE_MATRIX);
//Setup the classifier
RandomForests forest;
forest.setForestSize( 10 );
forest.setNumRandomSplits( 100 );
forest.setMaxDepth( 10 );
forest.setMinNumSamplesPerNode( 10 );
//Add the feature extraction and classifier to the pipeline
pipeline.addFeatureExtractionModule( fft );
pipeline.setClassifier( forest );
trainingClassLabel = 1;
record = false;
processAudio = true;
trainingData.setNumDimensions( 1 ); //We are only going to use the data from one microphone channel, so the dimensions are 1
trainingSample.resize( AUDIO_BUFFER_SIZE, 1 ); //We will set the training matrix to match the audio buffer size
//Setup the audio card
ofSoundStreamSetup(2, 1, this, AUDIO_SAMPLE_RATE, AUDIO_BUFFER_SIZE, 4);
}
bool RandomForests::deepCopyFrom(const Classifier *classifier){
if( classifier == NULL ) return false;
if( this->getClassifierType() == classifier->getClassifierType() ){
RandomForests *ptr = (RandomForests*)classifier;
//Clear this tree
this->clear();
if( ptr->getTrained() ){
//Deep copy the forest
for(UINT i=0; i<ptr->forest.size(); i++){
this->forest.push_back( ptr->forest[i]->deepCopyTree() );
}
}
this->forestSize = ptr->forestSize;
this->numRandomSplits = ptr->numRandomSplits;
this->minNumSamplesPerNode = ptr->minNumSamplesPerNode;
this->maxDepth = ptr->maxDepth;
//Copy the base classifier variables
return copyBaseVariables( classifier );
}
return false;
}
bool RandomForests::combineModels( const RandomForests &forest ){
if( !getTrained() ){
errorLog << "combineModels( const RandomForests &forest ) - This instance has not been trained!" << endl;
return false;
}
if( !forest.getTrained() ){
errorLog << "combineModels( const RandomForests &forest ) - This external forest instance has not been trained!" << endl;
return false;
}
if( this->getNumInputDimensions() != forest.getNumInputDimensions() ) {
errorLog << "combineModels( const RandomForests &forest ) - The number of input dimensions of the external forest (";
errorLog << forest.getNumInputDimensions() << ") does not match the number of input dimensions of this instance (";
errorLog << this->getNumInputDimensions() << ")!" << endl;
return false;
}
//Add the trees in the other forest to this model
DecisionTreeNode *node;
for(UINT i=0; i<forest.getForestSize(); i++){
node = forest.getTree(i);
if( node ){
this->forest.push_back( node->deepCopy() );
forestSize++;
}
}
return true;
}
bool RandomForests::deepCopyFrom(const Classifier *classifier){
if( classifier == NULL ) return false;
if( this->getClassifierType() == classifier->getClassifierType() ){
RandomForests *ptr = (RandomForests*)classifier;
//Clear this tree
this->clear();
if( copyBaseVariables( classifier ) ){
//Deep copy the main node
if( this->decisionTreeNode != NULL ){
delete decisionTreeNode;
decisionTreeNode = NULL;
}
this->decisionTreeNode = ptr->deepCopyDecisionTreeNode();
if( ptr->getTrained() ){
//Deep copy the forest
this->forest.reserve( ptr->forest.size() );
for(size_t i=0; i<ptr->forest.size(); i++){
this->forest.push_back( ptr->forest[i]->deepCopy() );
}
}
this->forestSize = ptr->forestSize;
this->numRandomSplits = ptr->numRandomSplits;
this->minNumSamplesPerNode = ptr->minNumSamplesPerNode;
this->maxDepth = ptr->maxDepth;
this->removeFeaturesAtEachSpilt = ptr->removeFeaturesAtEachSpilt;
this->bootstrappedDatasetWeight = ptr->bootstrappedDatasetWeight;
this->trainingMode = ptr->trainingMode;
return true;
}
errorLog << "deepCopyFrom(const Classifier *classifier) - Failed to copy base variables!" << endl;
}
return false;
}
bool computeFeatureWeights( CommandLineParser &parser ){
infoLog << "Computing feature weights..." << endl;
string resultsFilename = "";
string modelFilename = "";
bool combineWeights = false;
//Get the model filename
if( !parser.get("model-filename",modelFilename) ){
errorLog << "Failed to parse filename from command line! You can set the model filename using the --model." << endl;
printUsage();
return false;
}
//Get the results filename
if( !parser.get("filename",resultsFilename) ){
errorLog << "Failed to parse results filename from command line! You can set the results filename using the -f." << endl;
printUsage();
return false;
}
//Get the results filename
parser.get("combine-weights",combineWeights);
//Load the model
GestureRecognitionPipeline pipeline;
if( !pipeline.load( modelFilename ) ){
errorLog << "Failed to load model from file: " << modelFilename << endl;
printUsage();
return false;
}
//Make sure the pipeline contains a random forest model and that it is trained
RandomForests *forest = pipeline.getClassifier< RandomForests >();
if( !forest ){
errorLog << "Model loaded, but the pipeline does not contain a RandomForests classifier!" << endl;
printUsage();
return false;
}
if( !forest->getTrained() ){
errorLog << "Model loaded, but the RandomForests classifier is not trained!" << endl;
printUsage();
return false;
}
//Compute the feature weights
if( combineWeights ){
VectorFloat weights = forest->getFeatureWeights();
if( weights.getSize() == 0 ){
errorLog << "Failed to compute feature weights!" << endl;
printUsage();
return false;
}
//Save the results to a file
fstream file;
file.open( resultsFilename.c_str(), fstream::out );
const unsigned int N = weights.getSize();
for(unsigned int i=0; i<N; i++){
file << weights[i] << endl;
}
file.close();
}else{
double norm = 0.0;
const unsigned int K = forest->getForestSize();
const unsigned int N = forest->getNumInputDimensions();
VectorFloat tmp( N, 0.0 );
MatrixDouble weights(K,N);
for(unsigned int i=0; i<K; i++){
DecisionTreeNode *tree = forest->getTree(i);
tree->computeFeatureWeights( tmp );
norm = 1.0 / Util::sum( tmp );
for(unsigned int j=0; j<N; j++){
tmp[j] *= norm;
weights[i][j] = tmp[j];
tmp[j] = 0;
}
}
//Save the results to a file
weights.save( resultsFilename );
}
return true;
}
bool combineModels( CommandLineParser &parser ){
infoLog << "Combining models..." << endl;
string directoryPath = "";
string modelFilename = "";
if( !parser.get("data-dir",directoryPath) ){
errorLog << "Failed to parse data-directory from command line! You can set the data-directory using the --data-dir option." << endl;
printUsage();
return false;
}
//Get the filename
if( !parser.get("model-filename",modelFilename) ){
errorLog << "Failed to parse filename from command line! You can set the model filename using the --model." << endl;
printUsage();
return false;
}
Vector< string > files;
infoLog << "- Parsing data directory: " << directoryPath << endl;
//Parse the directory to get all the csv files
if( !Util::parseDirectory( directoryPath, ".grt", files ) ){
errorLog << "Failed to parse data directory!" << endl;
return false;
}
RandomForests forest; //Used to validate the random forest type
GestureRecognitionPipeline *mainPipeline = NULL; // Points to the first valid pipeline that all the models will be merged to
Vector< GestureRecognitionPipeline* > pipelineBuffer; //Stores the pipeline for each file that is loaded
unsigned int inputVectorSize = 0; //Set to zero to mark we haven't loaded any models yet
const unsigned int numFiles = files.getSize();
bool mainPipelineSet = false;
bool combineModelsSuccessful = false;
pipelineBuffer.reserve( numFiles );
//Loop over the files, load them, and add valid random forest pipelines to the pipelineBuffer so they can be combined with the mainPipeline
for(unsigned int i=0; i<numFiles; i++){
infoLog << "- Loading model " << files[i] << ". File " << i+1 << " of " << numFiles << endl;
GestureRecognitionPipeline *pipeline = new GestureRecognitionPipeline;
if( pipeline->load( files[i] ) ){
infoLog << "- Pipeline loaded. Number of input dimensions: " << pipeline->getInputVectorDimensionsSize() << endl;
if( pipelineBuffer.size() == 0 ){
inputVectorSize = pipeline->getInputVectorDimensionsSize();
}
if( pipeline->getInputVectorDimensionsSize() != inputVectorSize ){
warningLog << "- Pipeline " << i+1 << " input vector size does not match the size of the first pipeline!" << endl;
}else{
Classifier *classifier = pipeline->getClassifier();
if( classifier ){
if( classifier->getClassifierType() == forest.getClassifierType() ){ //Validate the classifier is a random forest
if( !mainPipelineSet ){
mainPipelineSet = true;
mainPipeline = pipeline;
}else pipelineBuffer.push_back( pipeline );
}else{
warningLog << "- Pipeline " << i+1 << " does not contain a random forest classifer! Classifier type: " << classifier->getClassifierType() << endl;
}
}
}
}else{
warningLog << "- WARNING: Failed to load model from file: " << files[i] << endl;
}
}
if( mainPipelineSet ){
//Combine the random forest models with the main pipeline model
const unsigned int numPipelines = pipelineBuffer.getSize();
RandomForests *mainForest = mainPipeline->getClassifier< RandomForests >();
for(unsigned int i=0; i<numPipelines; i++){
infoLog << "- Combing model " << i+1 << " of " << numPipelines << " with main model..." << endl;
RandomForests *f = pipelineBuffer[i]->getClassifier< RandomForests >();
if( !mainForest->combineModels( *f ) ){
warningLog << "- WARNING: Failed to combine model " << i+1 << " with the main model!" << endl;
}
}
if( mainPipeline->getTrained() ){
infoLog << "- Saving combined pipeline to file..." << endl;
combineModelsSuccessful = mainPipeline->save( modelFilename );
}
}else{
errorLog << "Failed to combined models, no models were loaded!" << endl;
}
//Cleanup the pipeline buffer
for(unsigned int i=0; i<pipelineBuffer.getSize(); i++){
delete pipelineBuffer[i];
pipelineBuffer[i] = NULL;
}
return combineModelsSuccessful;
}
bool train( CommandLineParser &parser ){
string trainDatasetFilename = "";
string modelFilename = "";
unsigned int forestSize = 0;
unsigned int maxDepth = 0;
unsigned int minNodeSize = 0;
unsigned int numSplits = 0;
bool removeFeatures = false;
double bootstrapWeight = 0.0;
//Get the filename
if( !parser.get("filename",trainDatasetFilename) ){
errorLog << "Failed to parse filename from command line! You can set the filename using the -f." << endl;
printUsage();
return false;
}
//Get the model filename
parser.get("model-filename",modelFilename);
//Get the forest size
parser.get("forest-size",forestSize);
//Get the max depth
parser.get("max-depth",maxDepth);
//Get the min node size
parser.get("min-node-size",minNodeSize);
//Get the number of random splits
parser.get("num-splits",numSplits);
//Get the remove features
parser.get("remove-features",removeFeatures);
//Get the bootstrap weight
parser.get("bootstrap-weight",bootstrapWeight);
//Load some training data to train the classifier
ClassificationData trainingData;
infoLog << "- Loading Training Data..." << endl;
if( !trainingData.load( trainDatasetFilename ) ){
errorLog << "Failed to load training data!\n";
return false;
}
const unsigned int N = trainingData.getNumDimensions();
Vector< ClassTracker > tracker = trainingData.getClassTracker();
infoLog << "- Num training samples: " << trainingData.getNumSamples() << endl;
infoLog << "- Num dimensions: " << N << endl;
infoLog << "- Num classes: " << trainingData.getNumClasses() << endl;
infoLog << "- Class stats: " << endl;
for(unsigned int i=0; i<tracker.getSize(); i++){
infoLog << "- class " << tracker[i].classLabel << " number of samples: " << tracker[i].counter << endl;
}
//Create a new RandomForests instance
RandomForests forest;
//Set the decision tree node that will be used for each tree in the forest
string nodeType = "cluster-node"; //TODO: make this a command line option in the future
if( nodeType == "cluster-node" ){
forest.setDecisionTreeNode( DecisionTreeClusterNode() );
}
if( nodeType == "threshold-node" ){
forest.setTrainingMode( Tree::BEST_RANDOM_SPLIT );
forest.setDecisionTreeNode( DecisionTreeThresholdNode() );
}
//Set the number of trees in the forest
forest.setForestSize( forestSize );
//Set the maximum depth of the tree
forest.setMaxDepth( maxDepth );
//Set the minimum number of samples allowed per node
forest.setMinNumSamplesPerNode( minNodeSize );
//Set the number of random splits used per node
forest.setNumRandomSplits( numSplits );
//Set if selected features should be removed at each node
forest.setRemoveFeaturesAtEachSplit( removeFeatures );
//Set the bootstrap weight
forest.setBootstrappedDatasetWeight( bootstrapWeight );
//Add the classifier to a pipeline
GestureRecognitionPipeline pipeline;
pipeline.setClassifier( forest );
infoLog << "- Training model..." << endl;
//Train the classifier
if( !pipeline.train( trainingData ) ){
errorLog << "Failed to train classifier!" << endl;
return false;
}
infoLog << "- Model trained!" << endl;
infoLog << "- Training time: " << (pipeline.getTrainingTime() * 0.001) / 60.0 << " (minutes)" << endl;
infoLog << "- Saving model to: " << modelFilename << endl;
//Save the pipeline
if( !pipeline.save( modelFilename ) ){
warningLog << "Failed to save model to file: " << modelFilename << endl;
}
return true;
}
int main(int argc, const char * argv[])
{
//Parse the data filename from the argument list
if( argc != 2 ){
cout << "Error: failed to parse data filename from command line. You should run this example with one argument pointing to the data filename!\n";
return EXIT_FAILURE;
}
const string filename = argv[1];
//Create a new RandomForests instance
RandomForests forest;
//Set the number of trees in the forest
forest.setForestSize( 10 );
//Set the number of random candidate splits that will be used to choose the best splitting values
//More steps will give you a better model, but will take longer to train
forest.setNumRandomSplits( 100 );
//Set the maximum depth of the tree
forest.setMaxDepth( 10 );
//Set the minimum number of samples allowed per node
forest.setMinNumSamplesPerNode( 10 );
//Load some training data to train the classifier
ClassificationData trainingData;
cout << "Loading Training Data\n";
if( !trainingData.load( filename ) ){
cout << "Failed to load training data: " << filename << endl;
return EXIT_FAILURE;
}
//Use 20% of the training dataset to create a test dataset
ClassificationData testData = trainingData.partition( 80 );
//Train the classifier
if( !forest.train( trainingData ) ){
cout << "Failed to train classifier!\n";
return EXIT_FAILURE;
}
//Print the forest
forest.print();
//Save the model to a file
if( !forest.save("RandomForestsModel.grt") ){
cout << "Failed to save the classifier model!\n";
return EXIT_FAILURE;
}
//Load the model from a file
if( !forest.load("RandomForestsModel.grt") ){
cout << "Failed to load the classifier model!\n";
return EXIT_FAILURE;
}
//Test the accuracy of the model on the test data
double accuracy = 0;
for(UINT i=0; i<testData.getNumSamples(); i++){
//Get the i'th test sample
UINT classLabel = testData[i].getClassLabel();
VectorDouble inputVector = testData[i].getSample();
//Perform a prediction using the classifier
bool predictSuccess = forest.predict( inputVector );
if( !predictSuccess ){
cout << "Failed to perform prediction for test sampel: " << i <<"\n";
return EXIT_FAILURE;
}
//Get the predicted class label
UINT predictedClassLabel = forest.getPredictedClassLabel();
VectorDouble classLikelihoods = forest.getClassLikelihoods();
VectorDouble classDistances = forest.getClassDistances();
//Update the accuracy
if( classLabel == predictedClassLabel ) accuracy++;
cout << "TestSample: " << i << " ClassLabel: " << classLabel << " PredictedClassLabel: " << predictedClassLabel << endl;
}
cout << "Test Accuracy: " << accuracy/double(testData.getNumSamples())*100.0 << "%" << endl;
return EXIT_SUCCESS;
}
