def loadUserConfig(self):
self.userConfig = c.getInstance().getConfig("user")
print("Hello " + self.userConfig['name'])
print("Use command 'h' for usage help!")
self.selectClassifier(["u", self.userConfig['classifier']])
self.sound = not bool(self.userConfig['autostart_sound'])
self.soundFirstTime = True
self.toggleSound("")
def loadUserConfig(self):
self.userConfig = c.getInstance().getConfig("user")
print("Hello " + self.userConfig['name'])
print("Use command 'h' for usage help!")
self.selectClassifier(["u", self.userConfig['classifier']])
self.sound = not bool(self.userConfig['autostart_sound'])
self.soundFirstTime = True
self.toggleSound("")
def getClassificator(self, name):
if (name == ""):
raise Exception("No classificator specified, see usage 'h'")
elif (name == "lstm"):
if (name not in self.classificators):
from classifier.lstm.lstm import LSTM
lstmConfig = c.getInstance().getConfig("lstm")
cl = LSTM(self.recorder, lstmConfig)
self.classificators[name] = cl
cl = self.classificators[name]
elif (name == "svm"):
if (name not in self.classificators):
from classifier.svm.svm import SVM
svmConfig = c.getInstance().getConfig("svm")
cl = SVM(self.recorder, svmConfig)
self.classificators[name] = cl
cl = self.classificators[name]
elif (name == "trees"):
if (name not in self.classificators):
from classifier.trees.Trees import Trees
treeConfig = c.getInstance().getConfig("trees")
cl = Trees(self.recorder, treeConfig)
self.classificators[name] = cl
cl = self.classificators[name]
elif (name == "hmm"):
if (name not in self.classificators):
from classifier.hmm.gestureApplication import HMM
# hmmConfig = c.getInstance().getConfig("hmm")
cl = HMM(self.recorder)
self.classificators[name] = cl
cl = self.classificators[name]
elif (name == "kmeans"):
if (name not in self.classificators):
from classifier.k_means.kMeans import KMeans
cl = KMeans(self.recorder)
self.classificators[name] = cl
cl = self.classificators[name]
else:
raise Exception("Classificator not existing")
return cl
def getClassificator(self, name):
if(name == ""):
raise Exception("No classificator specified, see usage 'h'")
elif(name == "lstm"):
if(name not in self.classificators):
from classifier.lstm.lstm import LSTM
lstmConfig = c.getInstance().getConfig("lstm")
cl = LSTM(self.recorder, lstmConfig)
self.classificators[name] = cl
cl = self.classificators[name]
elif(name == "svm"):
if(name not in self.classificators):
from classifier.svm.svm import SVM
svmConfig = c.getInstance().getConfig("svm")
cl = SVM(self.recorder, svmConfig)
self.classificators[name] = cl
cl = self.classificators[name]
elif(name == "trees"):
if(name not in self.classificators):
from classifier.trees.Trees import Trees
treeConfig = c.getInstance().getConfig("trees")
cl = Trees(self.recorder, treeConfig)
self.classificators[name] = cl
cl = self.classificators[name]
elif(name == "hmm"):
if(name not in self.classificators):
from classifier.hmm.gestureApplication import HMM
# hmmConfig = c.getInstance().getConfig("hmm")
cl = HMM(self.recorder)
self.classificators[name] = cl
cl = self.classificators[name]
elif(name == "kmeans"):
if(name not in self.classificators):
from classifier.k_means.kMeans import KMeans
cl = KMeans(self.recorder)
self.classificators[name] = cl
cl = self.classificators[name]
else:
raise Exception("Classificator not existing")
return cl
def createKeyBindings(self):
keys = {}
config = c.getInstance().getConfig("keys")
keys[0] = eval(config.get("key_0"))
keys[1] = eval(config.get("key_1"))
keys[2] = eval(config.get("key_2"))
keys[3] = eval(config.get("key_3"))
keys[4] = eval(config.get("key_4"))
keys[5] = eval(config.get("key_5"))
keys[6] = eval(config.get("key_6"))
keys[7] = eval(config.get("key_7"))
return keys
def createKeyBindings(self):
keys = {}
config = c.getInstance().getConfig("keys")
keys[0] = eval(config.get("key_0"))
keys[1] = eval(config.get("key_1"))
keys[2] = eval(config.get("key_2"))
keys[3] = eval(config.get("key_3"))
keys[4] = eval(config.get("key_4"))
keys[5] = eval(config.get("key_5"))
keys[6] = eval(config.get("key_6"))
keys[7] = eval(config.get("key_7"))
return keys
def __init__(self, config, net):
self.config = config
self.net = net
self.avg = util.getAverage(self.net.datacut, self.net.datafold)
self.datalist = []
self.datanum = 0
self.has32 = False
# Classify2
self.previouspredict = 6
self.predcounter = 0
self.predHistSize = 8
self.predHistHalfUpper = 5
self.predcountertreshold = 5
self.predHistory = util.createArraySix(self.predHistSize, )
# Classify3 method
self.predhistoryforclassify3 = {
0: 0,
1: 0,
2: 0,
3: 0,
4: 0,
5: 0,
6: 0
}
self.classify3start = False
# classify4 & 5
self.start = 0
self.buffer = []
self.liveData = []
self.beginClassify = 0
self.beginMax = 0
self.maxValue = 0
self.maxValueList = []
# For interacting with OS
self.outkeys = None
if (c.getInstance().getOSConfig()['type'] == "posix"):
from systemkeys import SystemKeys
self.outkeys = SystemKeys()
self.classifyMethods = {
1: self.__classify1,
2: self.__classify2,
3: self.__classify3,
4: self.__classify4,
5: self.__classify5
}
def _setConfig(self):
self.config = c.getInstance()
self.checkNameSet()
self.checkOSet()
self.audioConfig = self.config.getAudioConfig()
self.pathsConfig = self.config.getPathsConfig()
self.recordConfig = self.config.getRecordConfig()
self.frequency = float(self.audioConfig['frequency'])
self.amplitude = float(self.audioConfig['amplitude'])
self.framerate = int(self.audioConfig['framerate'])
self.duration = int(self.audioConfig['duration'])
self.bufsize = int(self.audioConfig['buffersize'])
self.path = self.pathsConfig['gesturepath']
def __init__(self, config, net):
self.config = config
self.net = net
self.avg = util.getAverage(self.net.datacut, self.net.datafold)
self.datalist = []
self.datanum = 0
self.has32 = False
# Classify2
self.previouspredict = 6
self.predcounter = 0
self.predHistSize = 8
self.predHistHalfUpper = 5
self.predcountertreshold = 5
self.predHistory = util.createArraySix(self.predHistSize,)
# Classify3 method
self.predhistoryforclassify3 = {0:0, 1:0, 2:0, 3:0, 4:0, 5:0, 6:0}
self.classify3start = False
# classify4 & 5
self.start = 0
self.buffer = []
self.liveData = []
self.beginClassify = 0
self.beginMax = 0
self.maxValue = 0
self.maxValueList = []
# For interacting with OS
self.outkeys = None
if(c.getInstance().getOSConfig()['type'] == "posix"):
from systemkeys import SystemKeys
self.outkeys = SystemKeys()
self.classifyMethods = {1:self.__classify1,
2:self.__classify2,
3:self.__classify3,
4:self.__classify4,
5:self.__classify5
}
def __loadDataFromFile(merge67="false"):
gesturepath = c.getInstance().getPathsConfig()['gesturepath']
g = GestureFileIO(gesturePath=gesturepath, relative=relative)
data = [0] * nClasses
getData = None
if(average == "false"):
getData = g.getGesture3DNormalized
else:
getData = g.getGesture3DDiffAvg
if(merge67 == "true"):
merge67 = True
else:
merge67 = False
for i in classes:
data[i] = getData(i, [], merge67)
if(i == 6 and merge67):
data7 = getData(7, [], merge67)
data[i] = np.append(data[i], data7, axis=0)
print("data " + str(i) + " loaded shape: " + str(np.shape(data[i])))
print("data loaded, now creating dataset")
return data
def getAverage(cut, fold):
global avg
# Use this if reaults are too bad by creating avg on demand with recorded gestures
# the avg from all gestures isn't correct anymore -> hardcoded avg
# avg = [ 0.01405917, 0.01451394, 0.01501277, 0.01552741, 0.01608232, 0.0166838,
# 0.017318, 0.01801043, 0.01876477, 0.01958107, 0.02046903, 0.02144034,
# 0.02251206, 0.02370547, 0.02503494, 0.02651884, 0.02819135, 0.03008983,
# 0.03227148, 0.03479038, 0.03773177, 0.04122532, 0.04543961, 0.05061085,
# 0.05710018, 0.06553213, 0.07690142, 0.09310018, 0.11811639, 0.16215195,
# 0.26233294, 0.83321468, 0.99991455, 0.4672319, 0.2171238, 0.1442126,
# 0.10842227, 0.08702495, 0.07270665, 0.06246511, 0.0547552, 0.04876873,
# 0.04395861, 0.03999377, 0.03669765, 0.03390901, 0.03149998, 0.02941618,
# 0.02760306, 0.02599722, 0.02456548, 0.02327461, 0.02212122, 0.02108119,
# 0.02013257, 0.01926706, 0.01846762, 0.01773872, 0.01707368, 0.01645111,
# 0.01586561, 0.0153235, 0.01481276, 0.01433922]
if avg == None:
gesturepath = c.getInstance().getPathsConfig()['gesturepath']
g = GestureFileIO(gesturePath=gesturepath)
avg = g.getAvgFrequency()
avg = preprocessFrame(avg, cut, fold)
return avg
def getAverage(cut, fold):
global avg
# Use this if reaults are too bad by creating avg on demand with recorded gestures
# the avg from all gestures isn't correct anymore -> hardcoded avg
# avg = [ 0.01405917, 0.01451394, 0.01501277, 0.01552741, 0.01608232, 0.0166838,
# 0.017318, 0.01801043, 0.01876477, 0.01958107, 0.02046903, 0.02144034,
# 0.02251206, 0.02370547, 0.02503494, 0.02651884, 0.02819135, 0.03008983,
# 0.03227148, 0.03479038, 0.03773177, 0.04122532, 0.04543961, 0.05061085,
# 0.05710018, 0.06553213, 0.07690142, 0.09310018, 0.11811639, 0.16215195,
# 0.26233294, 0.83321468, 0.99991455, 0.4672319, 0.2171238, 0.1442126,
# 0.10842227, 0.08702495, 0.07270665, 0.06246511, 0.0547552, 0.04876873,
# 0.04395861, 0.03999377, 0.03669765, 0.03390901, 0.03149998, 0.02941618,
# 0.02760306, 0.02599722, 0.02456548, 0.02327461, 0.02212122, 0.02108119,
# 0.02013257, 0.01926706, 0.01846762, 0.01773872, 0.01707368, 0.01645111,
# 0.01586561, 0.0153235, 0.01481276, 0.01433922]
if avg == None:
gesturepath = c.getInstance().getPathsConfig()['gesturepath']
g = GestureFileIO(gesturePath=gesturepath)
avg = g.getAvgFrequency()
avg = preprocessFrame(avg, cut, fold)
return avg
def __loadDataFromFile(merge67="false"):
gesturepath = c.getInstance().getPathsConfig()['gesturepath']
g = GestureFileIO(gesturePath=gesturepath, relative=relative)
data = [0] * nClasses
getData = None
if (average == "false"):
getData = g.getGesture3DNormalized
else:
getData = g.getGesture3DDiffAvg
if (merge67 == "true"):
merge67 = True
else:
merge67 = False
for i in classes:
data[i] = getData(i, [], merge67)
if (i == 6 and merge67):
data7 = getData(7, [], merge67)
data[i] = np.append(data[i], data7, axis=0)
print("data " + str(i) + " loaded shape: " +
str(np.shape(data[i])))
print("data loaded, now creating dataset")
return data
def show_confusion_matrix():
svm_config = c.getInstance('../../').getConfig("svm")
svm = SVM(None, svm_config)
svm.show_confusion_matrix()
