def add_Training(clumpsFile, input_training, config): print('input_training', input_training) classesDict = dict() classesDict['landslide'] = [1, input_training] classesDict['no_landslide'] = [ 0, config['k_means_segmentation']['input']['training_labels'] ] tmpPath = config['k_means_segmentation']['temps']['shapes'] trainCol = 'class' trainColName = 'class_name' ratutils.populateClumpsWithClassTraining(clumpsFile, classesDict, tmpPath, trainCol, trainColName) return (clumpsFile)
# imageutils.popImageStats(outputMeanImg,True,0.,True) # populate clumps with training data print('Populating clumps with stats...') classesDict = dict() classesDict['Water'] = [1, waterMask] classesDict['Other'] = [2, otherMask] classesDict['VegWater'] = [3, vegwaterMask] tmpPath = './temp' classesIntCol = 'ClassInt' classesNameCol = 'ClassStr' ratutils.populateClumpsWithClassTraining(outputClumps, classesDict, tmpPath, classesIntCol, classesNameCol) # balance the training data rsgislib.classification.classratutils.balanceSampleTrainingRandom( outputClumps, classesIntCol, 'classesIntColBal', 50, 5000) classesIntCol = 'classesIntColBal' # classify the image --------------------------- # define the classifier classifier = ExtraTreesClassifier(n_estimators=500, n_jobs=-1) # classifier = RandomForestClassifier(n_estimators=100, max_features=3, oob_score=True, n_jobs=-1, verbose=0) # classifier = svm.SVC(kernel='rbf') # define the output colours