def visualizeAllAcc(self, correlated=False):
'''
Visualizes all datastreams of the Accelerometer
Arguments:
correlated: set correlated to True if you want to autocorrelate the data before graphing.
'''
time = np.linspace(
0, self.data.shape[0], self.data.shape[0]
) # from 0 to 10 seconds with [amount of datapoints] steps
f, (ax4, ax5, ax6) = plt.subplots(3, sharex=True, sharey=True)
ax4.plot(time, self.data.accX)
ax4.set_title("accX", y=0.65, size='smaller')
ax5.plot(time, self.data.accY)
ax5.set_title("accY", y=0.65, size='smaller')
ax6.plot(time, self.data.accZ)
ax6.set_title("accZ", y=0.65, size='smaller')
if correlated:
analyzer = DataAnalyzer.DataAnalyzer(self.data.accX)
ax4.plot(time, analyzer.getAutocorrelation())
analyzer = DataAnalyzer.DataAnalyzer(self.data.accY)
ax5.plot(time, analyzer.getAutocorrelation())
analyzer = DataAnalyzer.DataAnalyzer(self.data.accZ)
ax6.plot(time, analyzer.getAutocorrelation())
# Fine-tune figure; make subplots close to each other and hide x ticks for
# all but bottom plot.
f.subplots_adjust(hspace=0)
plt.setp([a.get_xticklabels() for a in f.axes[:-1]], visible=False)
plt.show()
def classify(classiFile):
global previousFile
if (previousFile == "none"):
previousFile = classiFile
FolderWatch.FolderWatch(CLASSIFY_FOLDER, classify)
else:
#print("classifying: " + previousFile)
dataFile = pd.read_csv(previousFile, header=0)
try:
print("trying...")
dataFile = analyzer.normalize(dataFile)
dataFile = analyzer.autoCorrelate(dataFile)
#is already being autocorrelated by getBPM
autoanalyzer = DataAnalyzer.AutoAnalyzer(dataFile)
output = autoanalyzer.getLastPeakTime()
detectedBPM = output['bpm']
time = output['time']
row = []
for secondData in training_data:
row.append(analyzer.DTWSimilarity(dataFile, secondData))
print("Classify: \t", str(model.predict([row])), ", BPM: ",
str(detectedBPM), ", time: ", str(time))
except:
print('raising exception')
pass
os.remove(previousFile)
previousFile = classiFile
FolderWatch.FolderWatch(CLASSIFY_FOLDER, classify)
def classify(classiFile):
global previousFile
if (previousFile == "none"):
previousFile = classiFile
FolderWatch.FolderWatch(CLASSIFY_FOLDER, classify)
else:
#print("classifying: " + previousFile)
try:
dataFile = pd.read_csv(previousFile, header=0)
except:
print("[EXCEPTION] cant read previous file")
pass
try:
print("Classificating...")
dataFile = analyzer.normalize(dataFile)
dataFile = analyzer.autoCorrelate(dataFile)
#is already being autocorrelated by getBPM
autoanalyzer = DataAnalyzer.AutoAnalyzer(dataFile)
output = autoanalyzer.getLastPeakTime() #calculates where the first occuring peak is in the file
detectedBPM = output['bpm']
time = output['time']
peakIndex = output['index'] #index number of the peak in the file
endPeakIndex = output['endPeriodIndex']
periodData = autoanalyzer.getPeriodsFromDataIndex(1, peakIndex)['data'] #get one period of data starting from peak index.
row = []
for secondData in training_data: # calculate the DTW Similarity between the sample and all items in the training dataset.
row.append(analyzer.DTWSimilarity(periodData, secondData))
gesture = model.predict([row])[0] #predict the gesture with KNN using the calculated distance matrix
# if BPM is larger than 200, you know for sure it isn't a movement. Temporary 'knutseloplossing'
if detectedBPM > 200:
gesture = "rest"
print("Classify: \t", str(gesture), ", BPM: ", str(detectedBPM), ", time: ", str(time))
socket.sendClassify(str(gesture), detectedBPM, time) #send back the classified gesture through the socket.
dataFile[peakIndex:endPeakIndex].to_csv(CLASSIFY_SAVE_FOLDER + str(gesture) + "-" + str(detectedBPM) + "-" + str(time) + ".csv")
except:
print('[EXCEPTION] during classification')
pass
try:
os.remove(previousFile)
except:
print("[EXCEPTION] cant delete previous file")
pass
previousFile = classiFile
FolderWatch.FolderWatch(CLASSIFY_FOLDER, classify)
def classify(classiFile):
global previousFile
if (previousFile == "none"):
previousFile = classiFile
FolderWatch.FolderWatch(CLASSIFY_FOLDER, classify)
else:
#print("classifying: " + previousFile)
try:
dataFile = pd.read_csv(previousFile, header=0)
except:
print("[EXCEPTION] cant read previous file")
pass
try:
print("Classificating...")
dataFile = analyzer.normalize(dataFile)
dataFile = analyzer.autoCorrelate(dataFile)
#is already being autocorrelated by getBPM
autoanalyzer = DataAnalyzer.AutoAnalyzer(dataFile)
output = autoanalyzer.getLastPeakTime()
detectedBPM = output['bpm']
time = output['time']
periodData = autoanalyzer.getPeriods(1, startIndexPeriod=1)['data']
row = []
for secondData in training_data:
row.append(analyzer.DTWSimilarity(periodData, secondData))
gesture = model.predict([row])[0]
#knutseloplossing
if detectedBPM > 200:
gesture = "rest"
print("Classify: \t", str(gesture), ", BPM: ", str(detectedBPM),
", time: ", str(time))
socket.sendClassify(str(gesture), detectedBPM, time)
print("tried to send...")
except:
print('[EXCEPTION] during classification')
pass
try:
os.remove(previousFile)
except:
print("[EXCEPTION] cant delete previous file")
pass
previousFile = classiFile
FolderWatch.FolderWatch(CLASSIFY_FOLDER, classify)
def visualizeStream(self,
dataStream,
correlated=False,
title='Data',
vLine=None):
'''Visualizes a particular datastream
Arguments
data: single datastream
title: [optional]
'''
time = np.linspace(0, dataStream.shape[0], dataStream.shape[0])
f, ax1 = plt.subplots(1)
ax1.plot(time, dataStream)
ax1.set_title(title)
if correlated:
analyzer = DataAnalyzer.DataAnalyzer(dataStream)
ax1.plot(time, analyzer.getAutocorrelation())
if (vLine):
plt.axvline(vLine)
plt.show()
# Helper functions:
# get all the data files from the directory
def getDataFileNames(dataType, movement="", dataFolder=DATA_FOLDER):
files = os.listdir(dataFolder)
output = []
for file in files:
if dataType in file and movement in file:
output.append(file)
return output
# ------------------- MAIN ------------------------------------
analyzer = DataAnalyzer.DataAnalyzer()
print("Doing " + str(ITERATIONS) + " iterations with a " +
str(TEST_SIZE_PERCENT) + " test ratio")
# -- training --
data_data = []
data_labels = []
data_data_length = []
files = getDataFileNames("training")
for trainingFile in files:
dataFile = pd.read_csv(DATA_FOLDER + trainingFile, header=0)
#data = [dataFile['alpha'], dataFile['beta'], dataFile['gamma'], dataFile['accX'], dataFile['accY'], dataFile['accZ']]
dataFile = analyzer.normalize(dataFile)
# get all the data files from the directory
def getDataFileNames(dataType, movement = "", dataFolder = DATA_FOLDER):
files = os.listdir(dataFolder);
output = []
for file in files:
if dataType in file and movement in file:
output.append(file)
return output
# ------------------- MAIN ------------------------------------
analyzer = DataAnalyzer.DataAnalyzer()
print("Connecting to socket..")
socket = Socket.Connection()
# -- training --
print("[TRAINING]")
training_data = []
training_labels = []
training_data_length = []
files = getDataFileNames("", dataFolder=SEQUENCE_FOLDER)
for trainingFile in files:
dataFile = pd.read_csv(SEQUENCE_FOLDER + trainingFile, header = 0)
from helpers import DataAnalyzer
import matplotlib.pyplot as plt
import numpy as np
from sklearn.preprocessing import normalize as skNorm
#dataFile = pd.read_csv("../../data/testing-leftright-qsMbpdsd6zQTqlrKAADi-1-72BPM.csv", header=0)
#dataFile = pd.read_csv("../../data/training-rotateclockwise-avkfxrmpauHdDpeaAAAa-6.csv", header=0)
#dataFile = pd.read_csv("../../data/training-updown-avkfxrmpauHdDpeaAAAa-1.csv", header=0)
#normal data
dataFile = pd.read_csv(
"../data/FLAWED/lessthan1period/updown-35-49-R53P95G3cV93UMlKAAB0-3_1.csv",
header=0)
da = DataAnalyzer.DataAnalyzer()
#Normalize each stream individually:
'''
dataNorm = dataFile.copy()
dataNorm['accX'] = skNorm(dataNorm['accX'])[0]
dataNorm['accY'] = skNorm(dataNorm['accY'])[0]
dataNorm['accZ'] = skNorm(dataNorm['accZ'])[0]
'''
#dataFile = da.normalize(dataFile)
#dataFile = da.autoCorrelate(dataFile)
daa = DataAnalyzer.AutoAnalyzer(dataFile)
visualizer = Visualizer.Visualizer(dataFile)
import pandas as pd
from helpers import Visualizer
from helpers import DataAnalyzer
#dataFile = pd.read_csv("../../data/testing-leftright-qsMbpdsd6zQTqlrKAADi-1-72BPM.csv", header=0)
#dataFile = pd.read_csv("../../data/training-rotateclockwise-avkfxrmpauHdDpeaAAAa-6.csv", header=0)
#dataFile = pd.read_csv("../../data/training-updown-avkfxrmpauHdDpeaAAAa-1.csv", header=0)
#normal data
dataFile = pd.read_csv(
"../data/forvisualize/training-w-WvUVD6yGptP0OmsZAABi-3.csv", header=0)
da = DataAnalyzer.DataAnalyzer()
#dataFile = da.normalize(dataFile)
#dataFile = da.autoCorrelate(dataFile)
das = DataAnalyzer.StreamDataAnalyzer(
dataFile['accZ']) #<---- PLAY WITH THIS PARAMETER Z, X, Y
#output = das.getFFTData()
daa = DataAnalyzer.AutoAnalyzer(dataFile)
'''
output = daa.getLastPeakTime(visualize=True, periods=4)
peakTime = output['index']
graphData = das.getPeriods(4, startIndex=peakTime)['data']
'''
visualizer = Visualizer.Visualizer(dataFile)
visualizer.visualizeAll(correlated=False)
dataFile2 = da.standardize(dataFile)
import scipy.stats as stats
from sklearn.preprocessing import normalize as skNorm
from scipy import signal
samplingRate = 1000 / 50
dataFile = pd.read_csv(
"../../data/good-backup-10seconds/testing-leftright-JUxdyRarf6RVZv0WAABN-7.csv",
header=0)
#dataFile2 = pd.read_csv("../../data/training-leftright-avkfxrmpauHdDpeaAAAa-3.csv", header=0)
dataAlpha = dataFile['alpha'].values
dataBeta = dataFile['beta'].values
dataGamma = dataFile['gamma'].values
analyzer = DataAnalyzer.AutoAnalyzer(dataFile)
visualizer = Visualizer.Visualizer(dataFile)
res = analyzer.getLastPeakTime(visualize=True)
print(res['time'])
print(res['bpm'])
'''
length = int(samplingRate / (bpm[0]/60))
startIndex = length * 2 # <-- Start extracting the peak from the 2nd period
rates = np.array([70,80,90,100,110,120,130,140])/60 #BPMs to test
piece = dataFile['accZ'][startIndex: startIndex+length*2]
peak = signal.find_peaks_cwt(piece, samplingRate/rates/2)
