def train(self, readFromFile=True, path="./training3/"):
"""
train the image processor
:return: True if the training succeeds, false otherwise.
"""
self.knn = ColorKnnClassifier()
self.ratioKnns = []
for i in range(7):
self.ratioKnns.append(LetterKnnClassifier())
if readFromFile:
f = open(path, 'r')
for l in f:
l = l.split(",")
x = [float(i) for i in l[1:]]
self.ratioKnns[cc.getColorFromId(int(l[0]))].train.append((int(l[0]), x))
f.close()
f = open("color_data.csv", 'r')
for l in f:
l = l.split(',')
x = [float(i) for i in l[1:]]
self.knn.train.append((int(l[0]), x))
f.close()
else:
fs = listdir(path)
for f in fs:
s = f.split(".")
if len(s) < 2:
continue
s = f.split("_")
if len(s) < 2:
continue
c1 = const.getConstantFromString(s[0])
c2 = cc.getColorFromId(c1)
if c2 is not None:
self.ratioKnns[c2].loadTrainingImage(path + f, c1)
print "LetterClass. " + str(c2) + ":" + str(
cc.getColorFromNumber(c2)) + " trained: " + f + " as " + str(const.getStringFromNumber(c1))
self.knn.trainModel()
for r in self.ratioKnns:
r.trainModel()
print "Completed training"
print "------------------"
def run(self):
r = self.processor.knn.testFromImage(self.sec, show=False, seed=7)
print int(r[0]), cc.getColorFromNumber(r[0]),
v = self.processor.ratioKnns[int(r[0])].testFromImage(self.sec, show=False, seed=5)
x = const.getStringFromNumber(v[0])
print x, self.row, self.col
self.processor.lineLocks[self.row].acquire()
self.processor.output[self.row][self.col] = x
self.processor.lineLocks[self.row].release()
def run(self):
r = self.processor.knn.testFromImage(self.sec, show=False, seed=7)
print int(r[0]), cc.getColorFromNumber(r[0]),
v = self.processor.ratioKnns[int(r[0])].testFromImage(self.sec,
show=False,
seed=5)
x = const.getStringFromNumber(v[0])
print x, self.row, self.col
self.processor.lineLocks[self.row].acquire()
self.processor.output[self.row][self.col] = x
self.processor.lineLocks[self.row].release()
for f in fs:
s = f.split(".")
if len(s) < 2:
continue
s = f.split("_")
if len(s) < 2:
continue
c1 = const.getConstantFromString(s[0])
c2 = cc.getColorFromId(c1)
if c2 is not None:
ratioKnns[c2].loadTrainingImage(path+f, c1)
print "LetterClass. "+str(c2)+":"+str(cc.getColorFromNumber(c2))+" trained: "+f+ " as "+str(const.getStringFromNumber(c1))
knn.populateData()
knn.trainModel()
for r in ratioKnns:
r.trainModel()
print "done training"
print len(knn.train)
print [len(r.train) for r in ratioKnns]
print "starting test:"
s = f.split(".")
if len(s) < 2:
continue
s = f.split("_")
if len(s) < 2:
continue
c1 = const.getConstantFromString(s[0])
c2 = cc.getColorFromId(c1)
if c2 is not None:
ratioKnns[c2].loadTrainingImage(path + f, c1)
print "LetterClass. " + str(c2) + ":" + str(cc.getColorFromNumber(c2)) + " trained: " + f + " as " + str(
const.getStringFromNumber(c1)
)
knn.populateData()
knn.trainModel()
for r in ratioKnns:
r.trainModel()
print "done training"
print len(knn.train)
print [len(r.train) for r in ratioKnns]
print "starting test:"
def train(self, readFromFile=True, path="./training3/"):
"""
train the image processor
:return: True if the training succeeds, false otherwise.
"""
self.knn = ColorKnnClassifier()
self.ratioKnns = []
for i in range(7):
self.ratioKnns.append(LetterKnnClassifier())
if readFromFile:
f = open(path, 'r')
for l in f:
l = l.split(",")
x = [float(i) for i in l[1:]]
self.ratioKnns[cc.getColorFromId(int(l[0]))].train.append(
(int(l[0]), x))
f.close()
f = open("color_data.csv", 'r')
for l in f:
l = l.split(',')
x = [float(i) for i in l[1:]]
self.knn.train.append((int(l[0]), x))
f.close()
else:
fs = listdir(path)
for f in fs:
s = f.split(".")
if len(s) < 2:
continue
s = f.split("_")
if len(s) < 2:
continue
c1 = const.getConstantFromString(s[0])
c2 = cc.getColorFromId(c1)
if c2 is not None:
self.ratioKnns[c2].loadTrainingImage(path + f, c1)
print "LetterClass. " + str(c2) + ":" + str(
cc.getColorFromNumber(
c2)) + " trained: " + f + " as " + str(
const.getStringFromNumber(c1))
self.knn.trainModel()
for r in self.ratioKnns:
r.trainModel()
print "Completed training"
print "------------------"
