def obtain_test_data():
''' Obtain testing data for the classifier'''
featureList = []
tweets = []
with open('../data/preprocessedTesting.data','rb') as f:
reader = csv.reader(f, delimiter='\t')
l = list(reader)
for row in l:
sentiment = row[2]
tweet = row[3]
tweet = tweet[:-2] # ignoring hashcounts
featureVector = featureExtraction.getFeatureVector(tweet)
# featureList.extend(featureVector)
tweets.append((featureVector, sentiment))
with open('../data/preprocessedTraining.data','rb') as f:
reader = csv.reader(f, delimiter='\t')
l = list(reader)
for row in l:
sentiment = row[2]
tweet = row[3]
tweet = tweet[:-2] # ignoring hashcounts
featureVector = featureExtraction.getFeatureVector(tweet)
featureList.extend(featureVector)
# Remove featureList duplicates
featureList = list(set(featureList))
result = featureExtraction.getFeatureListAndLabels(tweets, featureList)
return result
def obtain_test_data():
''' Obtain testing data for the classifier'''
featureList = []
tweets = []
with open('../data/preprocessedTesting.data', 'rb') as f:
reader = csv.reader(f, delimiter='\t')
l = list(reader)
for row in l:
sentiment = row[2]
tweet = row[3]
tweet = tweet[:-2] # ignoring hashcounts
featureVector = featureExtraction.getFeatureVector(tweet)
# featureList.extend(featureVector)
tweets.append((featureVector, sentiment))
with open('../data/preprocessedTraining.data', 'rb') as f:
reader = csv.reader(f, delimiter='\t')
l = list(reader)
for row in l:
sentiment = row[2]
tweet = row[3]
tweet = tweet[:-2] # ignoring hashcounts
featureVector = featureExtraction.getFeatureVector(tweet)
featureList.extend(featureVector)
# Remove featureList duplicates
featureList = list(set(featureList))
result = featureExtraction.getFeatureListAndLabels(tweets, featureList)
return result
def getFeatures(folder):
features = []
with open(os.path.join(folder,"calibration.csv")) as csvfile:
reader = csv.reader(csvfile)
low = [ float(x) for x in reader.next()]
high = [ float(x) for x in reader.next()]
green = fe.colourFilter(tuple(low),tuple(high))
low = [ float(x) for x in reader.next()]
high = [ float(x) for x in reader.next()]
blue = fe.colourFilter(tuple(low),tuple(high))
for f in os.listdir(folder):
if os.path.splitext(f)[1] == ".ppm":
imbgr = cv2.imread(os.path.join(folder,f))
hull = green.getColourHull(imbgr)
features.append(fe.getFeatureVector(hull,['central']))
hull = blue.getColourHull(imbgr)
features.append(fe.getFeatureVector(hull,['central']))
return features
# if temp<minimumValue:
# minimumValue=temp
isLog = True
featureMatrix = []
labelMatrix = []
featureVector = []
for index in csvData.keys():
print(len(csvData[index]))
if isLog:
featureLog = './' + index + '_feature.csv'
fl = open(featureLog, 'w')
if True:
actionLength = len(csvData[index])
for i in range(actionLength):
featureVector = getFeatureVector(csvData[index][i])
# if len(featureVector)!=21:
# print('wrong length -- discarded')
# continue
# print(len(featureVector))
featureMatrix.append(featureVector)
labelMatrix.append(labelSwitch(index))
if isLog:
for item in featureVector:
fl.write(str(item))
fl.write(',')
fl.write(str(index))
fl.write('\n')
# if index=='fi':
# print(len(featureVector))
# f = open(path, 'w')
# for i in range(len(signalSegment[2])):
# f.write(str(signalSegment[0][i]))
# f.write(',')
# f.write(str(signalSegment[1][i]))
# f.write(',')
# f.write(str(signalSegment[2][i]))
# f.write('\n')
# f.close()
# f = open(path, 'a')
# fl = open(featureLog, 'a')
# pl = open(predictionLog, 'a')
### SVM
featureVector = feature.getFeatureVector(signalSegment)
featureArray = np.array(featureVector)
featureArray = featureArray.reshape((1,len(featureVector)))
featureArray = tools.standardise(featureArray,meanValue,stdValue)
svmOutput = model.predict(featureArray)
print(rr.printOutput(svmOutput))
f=open('result.txt','w')
f.write(str(svmOutput))
f.close()
### ANN:
'''featureVector = feature.getFeatureVector(signalSegment)
print(len(featureVector))
featureArray = np.array(featureVector)
featureArray = featureArray.reshape((1,len(featureVector)))
labels = pickle.load(modelfile)
running = deque([])
with open(sys.argv[1]) as csvfile:
reader = csv.reader(csvfile)
low = [ float(x) for x in reader.next()]
high = [ float(x) for x in reader.next()]
green = fe.colourFilter(tuple(low),tuple(high))
while 1:
try:
imbgr = fe.get_video()
hull = green.getColourHull(imbgr)
if len(hull):
features = fe.getFeatureVector(hull,['central'])
imbgrclass = model.predict([features])
imbgrclass = imbgrclass[0]
if len(running) < 10:
sign = imbgrclass
running.append(sign)
else:
_ = running.popleft()
running.append(imbgrclass)
sign = median(running)
sign = int(round(sign))
letter = labels[sign]