def main():
path='/local/attale00/'
allFiles = utils.getAllFiles(path+'Multi-PIE/extracted')
allLabelFiles = utils.getAllFiles(path+'Multi-PIE/db_labels')
np.random.shuffle(allLabelFiles)
#get the labels from the database, for each person
sliceat=250
labelstest=utils.parseLabelFiles(path+'Multi-PIE/db_labels','sex',allLabelFiles[0:sliceat],cutoffSeq='',suffix='')
labelstraining = utils.parseLabelFiles(path+'Multi-PIE/db_labels','sex',allLabelFiles[sliceat:],cutoffSeq='',suffix='')
#now generate the label dict for each file
labsTest={}
labsTraining={}
for f in allFiles:
if labelstest.has_key(f[0:3]+'.labels'):
labsTest[f]=labelstest[f[0:3]+'.labels']
elif labelstraining.has_key(f[0:3]+'.labels'):
labsTraining[f]=labelstraining[f[0:3]+'.labels' ]
testSet = fg.dataContainer(labsTest)
trainingSet = fg.dataContainer(labsTraining)
roi=(0,64,0,64)
ppc=(8,8)
cpb=(8,8)
fg.getHogFeature(testSet,roi,path=path+'Multi-PIE_grayScale64/',ending=None,extraMask = None,pixels_per_cell=ppc,cells_per_block=cpb)
fg.getHogFeature(trainingSet,roi,path=path+'Multi-PIE_grayScale64/',ending=None,extraMask = None,pixels_per_cell=ppc, cells_per_block=cpb)
testSet.targetNum=map(utils.mapSexLabel2Two,testSet.target)
trainingSet.targetNum = map(utils.mapSexLabel2Two,trainingSet.target)
rf1=classifierUtils.standardRF(max_features=np.sqrt(len(testSet.data[0])))
rf2=classifierUtils.standardRF(max_features=np.sqrt(len(trainingSet.data[0])))
rf1.fit(testSet.data,testSet.targetNum)
s=rf1.score(trainingSet.data,trainingSet.targetNum)
trainingSet.classifiedAs=rf1.predict(trainingSet.data)
trainingSet.hasBeenClassified=True
classifierUtils.evaluateClassification(trainingSet,{0:'male',1:'female'})
print 'Score: {}'.format(s)
print '----------other way around ----\n'
rf2.fit(trainingSet.data,trainingSet.targetNum)
s=rf2.score(testSet.data,testSet.targetNum)
testSet.classifiedAs=rf2.predict(testSet.data)
testSet.hasBeenClassified=True
classifierUtils.evaluateClassification(testSet,{0:'male',1:'female'})
print 'Score: {}'.format(s)
def main(mode):
path = "/local/attale00/extracted_pascal__4__Multi-PIE"
path_ea = path + "/color128/"
allLabelFiles = utils.getAllFiles("/local/attale00/a_labels")
labeledImages = [i[0:16] + ".png" for i in allLabelFiles]
# labs=utils.parseLabelFiles(path+'/Multi-PIE/labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
labs = utils.parseLabelFiles(
"/local/attale00/a_labels", "mouth", labeledImages, cutoffSeq=".png", suffix="_face0.labels"
)
testSet = fg.dataContainer(labs)
roi = (50, 74, 96, 160)
X = fg.getAllImagesFlat(path_ea, testSet.fileNames, (128, 256), roi=roi)
# perform ICA
if mode not in ["s", "v"]:
ica = FastICA(n_components=100, whiten=True)
ica.fit(X)
meanI = np.mean(X, axis=0)
X1 = X - meanI
data = ica.transform(X1)
filters = ica.components_
elif mode in ["s", "v"]:
W = np.load("/home/attale00/Desktop/classifiers/ica/filter1.npy")
m = np.load("/home/attale00/Desktop/classifiers/ica/meanI1.npy")
X1 = X - m
data = np.dot(X1, W.T)
for i in range(len(testSet.data)):
testSet.data[i].extend(data[i, :])
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
# fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 3, cells_per_block=(6,2),maskFromAlpha=False)
# fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=10)
testSet.targetNum = map(utils.mapMouthLabels2Two, testSet.target)
rf = classifierUtils.standardRF(max_features=np.sqrt(len(testSet.data[0])), min_split=5, max_depth=40)
if mode in ["s", "v"]:
print "Classifying with loaded classifier"
classifierUtils.classifyWithOld(
path, testSet, mode, clfPath="/home/attale00/Desktop/classifiers/ica/rf128ICA_1"
)
elif mode in ["c"]:
print "cross validation of data"
print "Scores"
# print classifierUtils.standardCrossvalidation(rf,testSet,n_jobs=5)
# _cvDissect(testSet,rf)
classifierUtils.dissectedCV(rf, testSet)
print "----"
elif mode in ["save"]:
print "saving new classifier"
_saveRF(testSet)
else:
print "not doing anything"
def errorPatch():
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/mouth_img_error_multiPie/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
#
fileNames = labeledImages;
testSet = fg.dataContainer(labs)
fg.getImagePatchStat(testSet,path=path_ea,patchSize=(4,12),overlap = 2)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
clfPath = '/home/attale00/Desktop/classifiers/errorpatches/rferror'
f=file(clfPath,'r')
print 'classifier used: '+ f.name
clf = pickle.load(f)
testSet.classifiedAs=clf.predict(testSet.data)
testSet.probabilities=clf.predict_proba(testSet.data)
return testSet
def main(mode):
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/mouth_img_error/'
#
fileNames = utils.getAllFiles(path_ea);
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
fg_mode = 0
size=(4,12)
overlap=2
#size=(40,120)
fg.getImagePatchStat(testSet,path=path_ea,patchSize=size,overlap = overlap,mode=fg_mode)
print 'feature vector length: {}'.format(len(testSet.data[0]))
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=13,max_depth=40)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
pickle.dump(rValues,open('errorpatch_size_{}'.format(size[0]),'w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf)
else:
print 'not doing anything'
def main(mode):
path = '/local/attale00/extracted_pascal__4__Multi-PIE'
path_ea = path+'/color128/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
#labs=utils.parseLabelFiles(path+'/Multi-PIE/labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
#fileNames = utils.getAllFiles(path_ea);
#labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(50,74,96,160)
#roi=(44,84,88,168)
# eM=np.load('/home/attale00/Desktop/mouthMask.npy')
# m=cv2.resize(np.uint8(eM),(256,256));
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
# dil = cv2.dilate(m,strel)
#
# m=dil>0;
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 8, cells_per_block=(6,2),maskFromAlpha=False)
fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=20)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=5,max_depth=40)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,'s')
elif mode in ['c']:
print 'cross validation of data'
print 'Scores'
#print classifierUtils.standardCrossvalidation(rf,testSet,n_jobs=5)
#_cvDissect(testSet,rf)
classifierUtils.dissectedCV(rf,testSet)
print '----'
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet)
else:
print 'not doing anything'
def multiPiePose():
path_pose = '/local/attale00/poseLabels/multipie'
poseLabel = utils.parseLabelFiles(path_pose,'rotY',labeledImages,cutoffSeq='.png',suffix='.frog')
poses = [(k,np.double(v)*180/np.pi) for k,v in poseLabel.iteritems()]
poses = sorted(poses,key=lambda s: s[1])
return poses
def aflwPose():
path_pose = "/local/attale00/poseLabels/aflw"
poseLabelaflw = utils.parseLabelFiles(path_pose, "rotY", fileNames, cutoffSeq=".png", suffix=".frog")
poses_aflw = [(k, np.double(i) * 180 / np.pi) for k, i in poseLabelaflw.iteritems()]
poses_aflw = sorted(poses_aflw, key=lambda s: s[1])
return poses_aflw
def patches():
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/multiPIE_cropped3/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
fileNames = labeledImages;
testSet = fg.dataContainer(labs)
roi=(0,37,0,115)
roi=None
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(40,120),roi=roi)
W=np.load('/home/attale00/Desktop/classifiers/patches/filter2.npy')
m=np.load('/home/attale00/Desktop/classifiers/patches/meanI2.npy')
X1=X-m
data=np.dot(X1,W.T)
for i in range(len(fileNames)):
testSet.data[i].extend(data[i,:])
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 5, pixels_per_cell=(24,8),cells_per_block=(3,3),maskFromAlpha=False)
fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=20)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
clfPath = '/home/attale00/Desktop/classifiers/patches/rfICAHogColor'
f=file(clfPath,'r')
print 'classifier used: '+ f.name
clf = pickle.load(f)
testSet.classifiedAs=clf.predict(testSet.data)
testSet.probabilities=clf.predict_proba(testSet.data)
return testSet
def main(mode):
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/mouth_img_error/'
#
fileNames = utils.getAllFiles(path_ea);
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
components = 150
roi=None
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(40,120),roi=roi)
# X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(120,40),roi=roi,resizeFactor = .5)
#
# perform ICA
if mode not in ['s','v']:
ica = FastICA(n_components=components,whiten=True)
ica.fit(X)
meanI=np.mean(X,axis=0)
X1=X-meanI
data=ica.transform(X1)
filters=ica.components_
elif mode in ['s','v']:
W=np.load('/home/attale00/Desktop/classifiers/patches/filterMP1.npy')
m=np.load('/home/attale00/Desktop/classifiers/patches/meanIMP1.npy')
X1=X-m
data=np.dot(X1,W.T)
for i in range(len(fileNames)):
testSet.data[i].extend(data[i,:])
print 'feature vector length: {}'.format(len(testSet.data[0]))
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=13,max_depth=40)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
pickle.dump(rValues,open('errorpatch_ica','w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,filters=filters,meanI=meanI)
else:
print 'not doing anything'
def main(nJobs = 1):
path = '/local/attale00/GoodPose/extracted_alpha/grayScale64'
fileNames = utils.getAllFiles(path);
labs=utils.parseLabelFiles('/local/attale00/GoodPose'+'/mouth_labels','mouth',fileNames,cutoffSeq='_0.png',suffix='_face0.labels')
print('-----computing Features-----')
roi2 = (0,32,0,64)
mouthSet = fg.dataContainer(labs)
#load the mask for the mouth room pixels and dilate it
eM=np.load('/home/attale00/Desktop/mouthMask.npy')
m=cv2.resize(np.uint8(eM),(256,256));
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
dil = cv2.dilate(m,strel)
m=dil>0;
#get the features
fg.getHogFeature(mouthSet,roi2,path=path+'/',ending=None,extraMask = None)
#map the string labels to numbers (required by sklearn)
#change the mapping here for different classifiers
mouthSet.targetNum=map(utils.mapMouthLabels2Two,mouthSet.target)
n_estimators = 100
min_split = 10
max_depth = 20
max_features = np.sqrt(len(mouthSet.data[0]))
rf = classifierUtils.standardRF(max_features = max_features)
rf2=classifierUtils.standardRF(max_features=max_features)
score=classifierUtils.standardCrossvalidation(rf2,mouthSet)
rf.fit(mouthSet.data,mouthSet.targetNum)
pickle.dump(rf,open('/home/attale00/Desktop/classifiers/RandomForestMouthclassifier_12','w'))
f=open('/home/attale00/Desktop/classifiers/RandomForestMouthclassifier_12.txt','w')
f.write('Trained on aflw\n')
f.write('Attribute: mouth' )
f.write('Features: getHogFeature(mouthSet,roi2,path=path,ending=None,extraMask = m) on 64*64 grayScale 3 direction bins \n')
f.write('ROI:(0,32,0,64)\n')
f.write('labels: closed, narrow: 0, open, wideOpen: 1\n')
f.write('CV Score: {}\n'.format(score))
f.close()
def main(mode):
path = '/local/attale00/AFLW_ALL'
path_ea = path+'/color256/'
fileNames = utils.getAllFiles(path_ea);
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(88,165,150,362)
#roi=(44,84,88,168)
# eM=np.load('/home/attale00/Desktop/mouthMask.npy')
# m=cv2.resize(np.uint8(eM),(256,256));
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
# dil = cv2.dilate(m,strel)
#
# m=dil>0;
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 4, cells_per_block=(26,9),maskFromAlpha=False)
fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=20)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=5,max_depth=40)
print len(testSet.data)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
#classifierUtils.standardCrossvalidation(rf,testSet,n_jobs=5)
classifierUtils.dissectedCV(rf,testSet)
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet)
else:
print 'not doing anything'
def multiPiePose():
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
path_pose = '/local/attale00/poseLabels/multipie'
poseLabel = utils.parseLabelFiles(path_pose,'rotY',labeledImages,cutoffSeq='.png',suffix='.frog')
poses = np.array([np.double(i)*180/np.pi for i in poseLabel.values()])
return poses
def aflwPose():
path_ea = '/local/attale00/AFLW_cropped/mouth_img_error/'
#
fileNames = utils.getAllFiles(path_ea);
path_pose = '/local/attale00/poseLabels/aflw'
poseLabelaflw = utils.parseLabelFiles(path_pose,'rotY',fileNames,cutoffSeq='.png',suffix='.frog')
poses_aflw = [np.double(i)*180/np.pi for i in poseLabelaflw.values()]
return np.array(poses_aflw)
def main(mode):
path='/local/attale00/'
allFiles = utils.getAllFiles(path+'Multi-PIE/extracted')
allLabelFiles = utils.getAllFiles(path+'Multi-PIE/labels')
#allLabelFiles = utils.getAllFiles(path+'a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
labs=utils.parseLabelFiles(path+'/Multi-PIE/labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
#labs=utils.parseLabelFiles(path+'a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi = (0,32,0,64)
#roi = (128,256,0,256)
eM=np.load('/home/attale00/Desktop/mouthMask.npy')
m=cv2.resize(np.uint8(eM),(256,256));
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
dil = cv2.dilate(m,strel)
m=dil>0;
fg.getHogFeature(testSet,roi,path=path+'Multi-PIE_grayScale64/',ending=None,extraMask = None)
fg.getColorHistogram(testSet,(50,190,110,402),path = path+'/Multi-PIE/extracted/',ending=None,colorspace='lab',range=(1.,255.0),bins = 20)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
#testSet.targetNum = map(utils.mapGlassesLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=5,max_depth=40)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
classifierUtils.dissectedCV(rf,testSet)
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf)
else:
print 'not doing anything'
def main(mode):
path = '/local/attale00/extracted_pascal__4__Multi-PIE'
path_ea = path+'/color256/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
#labs=utils.parseLabelFiles(path+'/Multi-PIE/labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(88,165,150,362)
#roi=(44,84,88,168)
# eM=np.load('/home/attale00/Desktop/mouthMask.npy')
# m=cv2.resize(np.uint8(eM),(256,256));
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
# dil = cv2.dilate(m,strel)
#
# m=dil>0;
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 4, cells_per_block=(26,9),maskFromAlpha=False)
fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=20)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,'v')
elif mode in ['c']:
print 'cross validation of data'
_cross_validate(testSet)
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet)
else:
print 'not doing anything'
def main(mode):
poses = multiPiePose()
#poses = multiPiePoseOriginal()
poseDict=splitByPose(poses,binmax=75, stepsize = 30)
for k,v in poseDict.iteritems():
fn = [i[0] for i in v]
labs=utils.parseLabelFiles(path_label,'mouth',fn,cutoffSeq='.png',suffix='_face0.labels')
labs=dict((lk,lv) for (lk,lv) in labs.iteritems() if not lv.startswith('narr'))
testSet = fg.dataContainer(labs)
fg.getHogFeature(testSet,None,path=path_ea,ending='.png',extraMask = None,orientations = 9, cells_per_block=(3,3),pixels_per_cell=(24,8),maskFromAlpha=False)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = 30,min_split=12,max_depth=70)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
print '------------------- pose {}-----------------'.format(k)
plt.figure()
#obj=classifierUtils.classifyWithOld(path,testSet,mode,clfPath = '/home/attale00/Desktop/classifiers/thesis/poseSplit/pose{}'.format(k))
obj=classifierUtils.classifyWithOld(path,testSet,mode,clfPath = '/home/attale00/Desktop/classifiers/thesis/mirror/rfHogMirror'.format(k))
obj.plot(title='Mirrored,Pose: {}, ntot: {}, nOpen{}'.format(k,len(testSet.data),testSet.targetNum.count(1)))
pickle.dump(obj,open('multiPie_mirror_aggregate{}'.format(k),'w'))
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
plt.title('Pose: {}, n: {}'.format(k,len(v)))
#pickle.dump(rValues,open('patches_pose_hog_{}'.format(k),'w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,identifier = k)
def main(mode):
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/cropped3/'
#
fileNames = utils.getAllFiles(path_ea);
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
roi=None
testSet = fg.dataContainer(labs)
testSetMirror = fg.dataContainer(labs)
for f in range(len(testSetMirror.fileNames)):
testSetMirror.fileNames[f]+='M'
orientations = 9
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = orientations, cells_per_block=(3,3),pixels_per_cell=(24,8),maskFromAlpha=False)
fg.getHogFeature(testSetMirror,roi,path='/local/attale00/AFLW_cropped/mirrored/', ending='.png',orientations = orientations, cells_per_block=(3,3),pixels_per_cell=(24,8))
testSet.addContainer(testSetMirror)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = 30,min_split=12,max_depth=70)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
pickle.dump(rValues,open('patches_cv_hog_{}'.format(orientations),'w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf)
else:
print 'not doing anything'
def texture():
path = '/local/attale00/extracted_pascal__4__Multi-PIE'
path_ea = path+'/color128/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(50,74,96,160)
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(128,256),roi=roi)
W=np.load('/home/attale00/Desktop/classifiers/ica/filter1.npy')
m=np.load('/home/attale00/Desktop/classifiers/ica/meanI1.npy')
X1=X-m
data=np.dot(X1,W.T)
for i in range(len(testSet.data)):
testSet.data[i].extend(data[i,:])
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 3, cells_per_block=(6,2),maskFromAlpha=False)
fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=10)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
clfPath = '/home/attale00/Desktop/classifiers/ica/rf128ICAHOGCOLOR'
f=file(clfPath,'r')
print 'classifier used: '+ f.name
clf = pickle.load(f)
testSet.classifiedAs=clf.predict(testSet.data)
testSet.probabilities=clf.predict_proba(testSet.data)
return testSet
def main(mode):
path = '/local/attale00/GoodPose'
path_ea = path+'/pascal128/'
fileNames = utils.getAllFiles(path+'/targets');
labs=utils.parseLabelFiles(path+'/mouth_labels','mouth',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
#roi=(88,165,150,362)
roi=(44,84,88,168)
eM=np.load('/home/attale00/Desktop/mouthMask.npy')
m=cv2.resize(np.uint8(eM),(256,256));
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
dil = cv2.dilate(m,strel)
m=dil>0;
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 5, cells_per_block=(8,3))
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
_cross_validate(testSet)
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet)
else:
print 'not doing anything'
def main(mode):
path = '/local/attale00/AFLW_ALL/'
poses = aflwPose()
poseDict=splitByPose(poses,binmax=100, stepsize = 40)
for k,v in poseDict.iteritems():
fn = [i[0] for i in v]
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fn,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
fg.getHogFeature(testSet,None,path=path_ea,ending='.png',extraMask = None,orientations = 9, cells_per_block=(3,3),pixels_per_cell=(24,8),maskFromAlpha=False)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = 40,min_split=12,max_depth=70)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
plt.title('Pose: {}, n: {}'.format(k,len(v)))
#pickle.dump(rValues,open('patches_pose_hog_{}'.format(k),'w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,identifier = k)
def main(mode):
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/cropped3/'
#
fileNames = utils.getAllFiles(path_ea);
# minr = 10000;
# for f in fileNames:
# im = cv2.imread(path_ea+f,-1)
# if im.shape[0]!=40 or im.shape[1]!=120:
# print f
# print im.shape
# minr = minr if im.shape[0]>= minr else im.shape[0]
#
# print minr
#
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(0,37,0,115)
roi=None
filters = None
meanI = None
#roi=(44,84,88,168)
# eM=np.load('/home/attale00/Desktop/mouthMask.npy')
# m=cv2.resize(np.uint8(eM),(256,256));
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
# dil = cv2.dilate(m,strel)
#
# m=dil>0;
components = 150
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(40,120),roi=roi)
# X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(120,40),roi=roi,resizeFactor = .5)
#
# perform ICA
if mode not in ['s','v']:
ica = FastICA(n_components=components,whiten=True)
ica.fit(X)
meanI=np.mean(X,axis=0)
X1=X-meanI
data=ica.transform(X1)
filters=ica.components_
elif mode in ['s','v']:
W=np.load('/home/attale00/Desktop/classifiers/patches/filterMP1.npy')
m=np.load('/home/attale00/Desktop/classifiers/patches/meanIMP1.npy')
X1=X-m
data=np.dot(X1,W.T)
for i in range(len(fileNames)):
testSet.data[i].extend(data[i,:])
#orientations = 2
#strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
#fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 5, cells_per_block=(3,3),pixels_per_cell=(24,8),maskFromAlpha=False)
#fg.getColorHistogram(testSet,(0,40,40,80),path=path_ea,ending='.png',colorspace='lab',bins=bins)
#fg.getImagePatchStat(testSet,path=path_ea,patchSize=(4,12))
#fg.getImagePatchStat(testSet,path='/local/attale00/AFLW_cropped/mouth_img_error/',patchSize=(4,12))
#pca
# n_samples, n_features = X.shape
#
# mean_ = np.mean(X, axis=0)
# X -= mean_
# U, S, V = linalg.svd(X)
# explained_variance_ = (S ** 2) / n_samples
# explained_variance_ratio_ = (explained_variance_ /explained_variance_.sum())
# K=V / S[:, np.newaxis] * np.sqrt(n_samples)
# filters=K[:100]
# data=np.dot(X,filters.T)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = 23,min_split=15,max_depth=70,n_estimators=150)
#rf=classifierUtils.standardRF(max_features = 5,min_split=12,max_depth=45)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
pickle.dump(rValues,open('patches_cv_ica_{}'.format(components),'w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,filters=filters,meanI=meanI)
else:
print 'not doing anything'
def main():
logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(message)s")
###############################################################################
# data
path = "/local/attale00/AFLW_ALL/"
path_ea = "/local/attale00/AFLW_cropped/cropped3/"
fileNames = utils.getAllFiles(path_ea)
labs = utils.parseLabelFiles(
path + "/labels/labels", "mouth_opening", fileNames, cutoffSeq=".png", suffix="_face0.labels"
)
testSet = fg.dataContainer(labs)
X = fg.getAllImagesFlat(path_ea, testSet.fileNames, (40, 120), roi=None)
#
# perform ICA
ica = FastICA(n_components=250, whiten=True)
ica.fit(X)
meanI = np.mean(X, axis=0)
X1 = X - meanI
data = ica.transform(X1)
for i in range(len(fileNames)):
testSet.data[i].extend(data[i, :])
DATA = np.zeros((len(testSet.data), len(testSet.data[0])))
for i in range(len(testSet.data)):
DATA[i, :] = np.array(testSet.data[i])
testSet.targetNum = map(utils.mapMouthLabels2Two, testSet.target)
classLabels = np.array(testSet.targetNum)
scores = []
parameters = {
"n_estimators": range(90, 260, 20),
"max_depth": range(50, 110, 20),
"min_samples_split": range(5, 35, 5),
"max_features": range(25, 75, 10),
"min_samples_leaf": range(1, 10, 5),
}
# parameters = {'n_estimators': range(90, 100,20),
# 'max_depth': range(50, 60,20),
# 'min_samples_split':range(5,15,5),
# 'max_features':range(25,35,10),
# 'min_samples_leaf':range(1,10,5)}
print "Performing grid search..."
print "parameters:"
pprint(parameters)
for i in range(80, 260, 15):
mf = i if i < 75 else 75
parameters["max_features"] = range(10, mf, 10)
s = GridSearchUtils.doGridSearch(data[:, 0:i], classLabels, parameters)
temp = (s[0], s[1], i)
scores.append(temp)
scores_s = sorted(scores, key=lambda x: x[0])
print scores_s
pickle.dump(scores_s, open("/local/attale00/gridsearches/ICA_3", "w"))
from sklearn.decomposition import FastICA
import numpy as np
# Display progress logs on stdout
logging.basicConfig(level=logging.INFO,
format='%(asctime)s %(levelname)s %(message)s')
###############################################################################
#data
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/cropped2/'
#
fileNames = utils.getAllFiles(path_ea);
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(0,37,0,115)
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(37,115),roi=roi)
#
# perform ICA
ica = FastICA(n_components=100,whiten=True)
ica.fit(X)
def main(mode):
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/multiPIE_cropped3/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
#labs=utils.parseLabelFiles(path+'/Multi-PIE/labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
labs=dict((k,v) for (k,v) in labs.iteritems() if not v.startswith('narr'))
#
testSet = fg.dataContainer(labs)
roi=(0,37,0,115)
roi=None
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(40,120),roi=roi)
# perform ICA
# if mode not in ['s','v']:
# ica = FastICA(n_components=100,whiten=True)
# ica.fit(X)
# meanI=np.mean(X,axis=0)
# X1=X-meanI
# data=ica.transform(X1)
# filters=ica.components_
#
# elif mode in ['s','v']:
# W=np.load('/home/attale00/Desktop/classifiers/thesis/filtercombined.npy')
# m=np.load('/home/attale00/Desktop/classifiers/thesis/meancombined.npy')
# X1=X-m
# data=np.dot(X1,W.T)
#
# for i in range(len(testSet.fileNames)):
# testSet.data[i].extend(data[i,:])
#
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
#fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 9, pixels_per_cell=(24,8),cells_per_block=(3,3),maskFromAlpha=False)
fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=40)
#fg.getImagePatchStat(testSet,path='/local/attale00/AFLW_cropped/mouth_img_error_multiPie/',patchSize =(4,12))
#pca
# n_samples, n_features = X.shape
#
# mean_ = np.mean(X, axis=0)
# X -= mean_
# U, S, V = linalg.svd(X)
# explained_variance_ = (S ** 2) / n_samples
# explained_variance_ratio_ = (explained_variance_ /explained_variance_.sum())
# K=V / S[:, np.newaxis] * np.sqrt(n_samples)
# filters=K[:100]
# data=np.dot(X,filters.T)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = 27,min_split=13,max_depth=40)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
obj=classifierUtils.classifyWithOld(path,testSet,mode,clfPath = '/home/attale00/Desktop/classifiers/thesis/color_only')
pickle.dump(obj,open('color_only','w'))
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
pickle.dump(rValues,open('patches_mp_','w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,filters=filters,meanI=meanI)
else:
print 'not doing anything'
def main(nJobs = 1):
path = '/local/attale00/GoodPose'
path_ea = path+'/extracted_alpha'
path_adata = path_ea + '/a_data'
filenames = utils.getAllFiles(path+'/targets');
attribute = 'mouth'
attribute_values = utils.parseLabelINIFile(path+'/mouth_labels/labels.ini',attribute);
print('------------Attribute: \t'+attribute+' ---------------')
for i in attribute_values:
print('Value: \t'+i)
print('----------------------------')
print('----------parsing label files------')
labs=utils.parseLabelFiles(path+'/mouth_labels','mouth',filenames,cutoffSeq='.png',suffix='_face0.labels')
print('-----computing Features-----')
#make 10 bin hist for each mouth
mouthSet = fg.getHistogram(10,(0,270,60,452),hrange=(155.,255.0),labelFileDict = labs,path = path_ea+'/grayScale/',ending='_0.png')
mouthSet.targetNum = map(utils.mapMouthLabels2Two,mouthSet.target)
fpr=[]
tpr=[]
fnr = []
for i in xrange(1,101,5):
trainingSet,testSet = mouthSet.splitInTestAndTraining(frac=.6)
(a,b,c)=_classify(trainingSet, testSet,n_estimators=i)
tpr.append(a)
fpr.append(b)
fnr.append(c)
plt.figure()
plt.plot(range(1,101,5),tpr)
plt.xlabel('number of trees')
plt.ylabel('score')
plt.title('.6 training set, max_feature 3, min_split 1' )
fpr=[]
tpr=[]
fnr = []
for i in xrange(1,20):
trainingSet,testSet = mouthSet.splitInTestAndTraining(frac=i*.05)
(a,b,c)=_classify(trainingSet, testSet,n_estimators=70)
tpr.append(a)
fpr.append(b)
fnr.append(c)
plt.figure()
plt.plot(np.arange(0.05,1,0.05),tpr)
plt.xlabel('fraction used as training set')
plt.ylabel('score')
plt.title('ntrees = 70, max_feature 3, min_split 1' )
fpr=[]
tpr=[]
fnr = []
for i in xrange(1,11):
trainingSet,testSet = mouthSet.splitInTestAndTraining(frac=.7)
(a,b,c)=_classify(trainingSet, testSet,n_estimators=70,max_features=i)
tpr.append(a)
fpr.append(b)
fnr.append(c)
plt.figure()
plt.plot(range(1,11),tpr)
plt.xlabel('number of features used')
plt.ylabel('score')
plt.title('ntrees = 70,frac=.7 min_split 1' )
plt.show()
#classifier
#linSVM = svm.SVC(kernel = 'linear',C=1)
#this takes forever: check if that can be true
#scoresLinSVM = cross_validation.cross_val_score(linSVM,data,y=targetNum,n_jobs=-1,verbose = 1)
#implement random forest classifier with verbosity level
#scoresRF = cross_validation.cross_val_score(rf,mouthSet.data,y=mouthSet.targetNum,n_jobs=nJobs,verbose = 1)
#print(scoresRF)
return
def main(mode):
path = '/local/attale00/AFLW_ALL/'
path_ea = '/local/attale00/AFLW_cropped/cropped3/'
#
fileNames = utils.getAllFiles(path_ea);
# minr = 10000;
# for f in fileNames:
# im = cv2.imread(path_ea+f,-1)
# if im.shape[0]!=40 or im.shape[1]!=120:
# print f
# print im.shape
# minr = minr if im.shape[0]>= minr else im.shape[0]
#
# print minr
#
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
roi=(0,37,0,115)
roi=None
#roi=(44,84,88,168)
# eM=np.load('/home/attale00/Desktop/mouthMask.npy')
# m=cv2.resize(np.uint8(eM),(256,256));
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
# dil = cv2.dilate(m,strel)
#
# m=dil>0;
#X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(40,120),roi=roi)
# perform ICA
names_open = []
names_closed = []
for i,f in enumerate(testSet.fileNames):
if testSet.targetNum[i] == 0:
names_closed.append(f)
elif testSet.targetNum[i] == 1:
names_open.append(f)
Xopen = fg.getAllImagesFlat(path_ea,names_open,(40,120))
XClosed = fg.getAllImagesFlat(path_ea,names_closed,(40,120))
if mode not in ['s','v']:
icaopen = FastICA(n_components=100,whiten=True)
icaopen.fit(Xopen)
meanIopen=np.mean(Xopen,axis=0)
X1open=Xopen-meanIopen
dataopen=icaopen.transform(X1open)
filtersopen=icaopen.components_
plottingUtils.showICAComponents(filtersopen,(40,120),4,4)
icaclosed = FastICA(n_components=100,whiten=True)
icaclosed.fit(XClosed)
meanIclosed=np.mean(XClosed,axis=0)
X1closed=XClosed-meanIclosed
dataclosed=icaclosed.transform(X1closed)
filtersclosed=icaclosed.components_
plottingUtils.showICAComponents(filtersclosed,(40,120),4,4)
plt.show()
elif mode in ['s','v']:
W=np.load('/home/attale00/Desktop/classifiers/patches/filterMP1.npy')
m=np.load('/home/attale00/Desktop/classifiers/patches/meanIMP1.npy')
X1=X-m
data=np.dot(X1,W.T)
for i in range(len(fileNames)):
testSet.data[i].extend(data[i,:])
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
#fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 5, cells_per_block=(3,3),pixels_per_cell=(24,8),maskFromAlpha=False)
#fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=20)
#pca
# n_samples, n_features = X.shape
#
# mean_ = np.mean(X, axis=0)
# X -= mean_
# U, S, V = linalg.svd(X)
# explained_variance_ = (S ** 2) / n_samples
# explained_variance_ratio_ = (explained_variance_ /explained_variance_.sum())
# K=V / S[:, np.newaxis] * np.sqrt(n_samples)
# filters=K[:100]
# data=np.dot(X,filters.T)
rf=classifierUtils.standardRF(max_features = 27,min_split=13,max_depth=40)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
classifierUtils.dissectedCV(rf,testSet)
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,filters=filters,meanI=meanI)
else:
print 'not doing anything'
def main(mode):
labelFiles='/local/attale00/aflw_original_labels'
path_ea = '/local/attale00/AFLW_cropped/eyes/'
fileNames = utils.getAllFiles(path_ea)
goods = []
for f in fileNames:
im = cv2.imread(path_ea+f)
if im.shape[0] == 40 and im.shape[1] ==120:
goods.append(f)
fileNames = goods
labs=utils.parseLabelFiles(labelFiles,'glasses',fileNames,cutoffSeq='.png',suffix='.labels')
testSet = fg.dataContainer(labs)
#
# X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(40,120),roi=roi)
#
#
# # perform ICA
# if mode not in ['s','v']:
# ica = FastICA(n_components=100,whiten=True)
# ica.fit(X)
# meanI=np.mean(X,axis=0)
# X1=X-meanI
# data=ica.transform(X1)
# filters=ica.components_
#
# elif mode in ['s','v']:
# W=np.load('/home/attale00/Desktop/classifiers/patches/filter2.npy')
# m=np.load('/home/attale00/Desktop/classifiers/patches/meanI2.npy')
# X1=X-m
# data=np.dot(X1,W.T)
#
# for i in range(len(testSet.fileNames)):
# testSet.data[i].extend(data[i,:])
#
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
fg.getHogFeature(testSet,None,path=path_ea,ending='.png',extraMask = None,orientations = 5, pixels_per_cell=(24,8),cells_per_block=(3,3),maskFromAlpha=False)
#testSet.targetNum=map(lambda x: 1 if x=='light' else 0,testSet.target)
#rf = svm.NuSVC()
#rf = linear_model.SGDClassifier(loss='perceptron', eta0=1, learning_rate='constant', penalty=None)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
obj=classifierUtils.classifyWithOld(path,testSet,mode,clfPath = '/home/attale00/Desktop/classifiers/thesis/color_only')
#pickle.dump(obj,open('color_only','w'))
elif mode in ['c']:
print 'cross validation of data'
testSet.targetNum=map(lambda x: 1 if x=='1' else 0,testSet.target)
rf=classifierUtils.standardRF(max_features = 27,min_split=5,max_depth=60,n_estimators = 500)
rValues = classifierUtils.dissectedCV(rf,testSet)
plt.title('No glasses against rest')
plt.show()
#pickle.dump(rValues,open('patches_mp_','w'))
elif mode in ['save']:
print 'saving new classifier'
testSet.targetNum=map(lambda x: 1 if x=='1' else 0,testSet.target)
rf=classifierUtils.standardRF(max_features = 27,min_split=5,max_depth=60,n_estimators = 500)
_saveRF(testSet,rf)
else:
print 'not doing anything'
def main(mode):
path = '/local/attale00/AFLW_ALL'
path_ea = path+'/color128/'
fileNames = utils.getAllFiles(path_ea);
labs=utils.parseLabelFiles(path+'/labels/labels','mouth_opening',fileNames,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
testSetMirror = fg.dataContainer(labs)
for f in range(len(testSetMirror.fileNames)):
testSetMirror.fileNames[f]+='M'
roi=(50,74,96,160)
X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(128,256),roi=roi)
Y=fg.getAllImagesFlat(path+'/mirror128/',testSet.fileNames,(128,256),roi=roi)
Z=np.concatenate((X,Y),axis=0)
# perform ICA
ica = FastICA(n_components=100,whiten=True)
ica.fit(Z)
meanI=np.mean(Z,axis=0)
X1=X-meanI
Y1=Y-meanI
data=ica.transform(X1)
datam=ica.transform(Y1)
filters=ica.components_
for i in range(len(fileNames)):
testSet.data[i].extend(data[i,:])
testSetMirror.data[i].extend(datam[i,:])
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
#fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 3, cells_per_block=(6,2),maskFromAlpha=False)
#fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=10)
#pca
# n_samples, n_features = X.shape
#
# mean_ = np.mean(X, axis=0)
# X -= mean_
# U, S, V = linalg.svd(X)
# explained_variance_ = (S ** 2) / n_samples
# explained_variance_ratio_ = (explained_variance_ /explained_variance_.sum())
# K=V / S[:, np.newaxis] * np.sqrt(n_samples)
# filters=K[:100]
# data=np.dot(X,filters.T)
testSet.addContainer(testSetMirror)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=5,max_depth=40)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
_classifyWithOld(path,testSet,mode)
elif mode in ['c']:
print 'cross validation of data'
classifierUtils.dissectedCV(rf,testSet)
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet,rf,filters=filters,meanI=meanI)
else:
print 'not doing anything'
def main(nJobs = 1):
path = '/local/attale00'
fileNames = utils.getAllFiles(path+'/targets');
attribute = 'mouth'
attribute_values = utils.parseLabelINIFile(path+'/mouth_labels/labels.ini',attribute);
print('------------Attribute: \t'+attribute+' ---------------')
for i in attribute_values:
print('Value: \t'+i)
print('----------------------------')
print('----------parsing label files------')
labs=utils.parseLabelFiles(path+'/mouth_labels','mouth',fileNames,cutoffSeq='.png',suffix='_face0.labels')
print('-----computing Features-----')
#make 10 bin hist for each mouth
#roi = (40,200,100,200)
roi = (50,190,110,402)
roi2=(128,256,0,256)
mouthSet = fg.dataContainer(labs)
#fg.getHistogram(20,roi,hrange=(0,255),dataC = mouthSet,path = path+'/extracted/gradients/Direction/',ending='_0.png')
eM=np.load('/home/attale00/Desktop/mouthMask.npy')
m=cv2.resize(np.uint8(eM),(256,256));
strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
dil = cv2.dilate(m,strel)
m=dil>0;
# em=m[roi[0]:roi[1],roi[2]:roi[3]]
# m= m !=True
fg.getHogFeature(mouthSet,roi2,path=path+'/targets/grayScaleSmall/',ending=None,extraMask = None)
#fg.getPixelValues(mouthSet,roi,path=path_ea+'/',ending='_0.png',mask =m, scaleFactor = 10)
#fg.getColorHistogram(mouthSet,roi,path=path_ea+'/',ending='_0.png',colorspace=None,range=(1.0,255.0),bins = 20)
mouthSet.targetNum=map(utils.mapMouthLabels2Two,mouthSet.target)
n_estimators = range(10,180,20);
max_features = range(2,22,2)
max_depth = range(5,40,5)
max_depth.append(100)
min_split = range(1,20,2)
score=[]
var = []
for n in n_estimators:
scoresRF = _crossValidate(mouthSet, max_depth = 20,n_estimators =n ,nJobs = nJobs,max_features = np.sqrt(len(mouthSet.data[0])),min_split = 5)
score.append(scoresRF.mean())
var.append(scoresRF.std())
print scoresRF
plt.errorbar(n_estimators,score,yerr=var)
plt.xlabel('number of trees')
# plt.ylabel('cross val score')
# mouthSet2 = fg.dataContainer(labs)
# roi=(256,512,0,512)
# fg.getColorHistogram(mouthSet2,roi,path=path+'/targets/',ending=None,colorspace='lab',range=(1.,255.0),bins = 20)
# mouthSet2.targetNum=map(utils.mapMouthLabels2Two,mouthSet2.target)
# score=[]
# var = []
# for n in n_estimators:
# scoresRF = _crossValidate(mouthSet2, max_depth = 20,n_estimators =n ,nJobs = nJobs,max_features = np.sqrt(len(mouthSet2.data[0])),min_split = 5)
#
# score.append(scoresRF.mean())
# var.append(scoresRF.std())
#
# print scoresRF
# plt.errorbar(n_estimators,score,yerr=var)
# plt.xlabel('number of trees')
# plt.ylabel('cross val score')
# fg.getColorHistogram(mouthSet,roi,path=path_ea+'/',ending='_0.png',colorspace='lab',range=(100.0,255.0),bins = 20)
#
# score=[]
# var = []
# for n in n_estimators:
# scoresRF = _crossValidate(mouthSet, max_depth = 20,n_estimators =n ,nJobs = nJobs,max_features = np.sqrt(len(mouthSet.data[0])),min_split = 5)
#
# score.append(scoresRF.mean())
# var.append(scoresRF.std())
#
# print scoresRF
# plt.errorbar(n_estimators,score,yerr=var)
# plt.xlabel('number of trees')
# plt.ylabel('cross val score')
# plt.legend(['HOG','LAB','HOG+LAB'])
# plt.title('20bins')
plt.show()
#classifier
#linSVM = svm.SVC(kernel = 'linear',C=1)
#this takes forever: check if that can be true
#scoresLinSVM = cross_validation.cross_val_score(linSVM,data,y=targetNum,n_jobs=-1,verbose = 1)
#implement random forest classifier with verbosity level
# roi_narrow=(60,160,130,382)
# extraMask = np.load('/home/attale00/Desktop/emptyMouthMask.npy')
#
# fg.getMeanAndVariance(roi_narrow,mouthSet,path_ea+'/',extraMask = extraMask,ending='_0.png')
# scoresRF = _crossValidate(mouthSet,max_features = 13)
# print 'Orientation and mean and cov' +str(scoresRF)
return
def main(mode):
path = '/local/attale00/extracted_pascal__4__Multi-PIE'
path_ea = path+'/color128/'
allLabelFiles = utils.getAllFiles('/local/attale00/a_labels')
labeledImages = [i[0:16]+'.png' for i in allLabelFiles]
#labs=utils.parseLabelFiles(path+'/Multi-PIE/labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
labs=utils.parseLabelFiles('/local/attale00/a_labels','mouth',labeledImages,cutoffSeq='.png',suffix='_face0.labels')
testSet = fg.dataContainer(labs)
roi=(50,74,96,160)
# X=fg.getAllImagesFlat(path_ea,testSet.fileNames,(128,256),roi=roi)
#
#
# # perform ICA
# if mode not in ['s','v']:
# ica = FastICA(n_components=50,whiten=True)
# ica.fit(X)
# meanI=np.mean(X,axis=0)
# X1=X-meanI
# data=ica.transform(X1)
# filters=ica.components_
#
# elif mode in ['s','v']:
# W=np.load('/home/attale00/Desktop/classifiers/thesis/filter3.npy')
# m=np.load('/home/attale00/Desktop/classifiers/thesis/meanI3.npy')
# X1=X-m
# data=np.dot(X1,W.T)
#
# for i in range(len(testSet.data)):
# testSet.data[i].extend(data[i,:])
###
#
#
#
#
# strel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
fg.getHogFeature(testSet,roi,path=path_ea,ending='.png',extraMask = None,orientations = 4, cells_per_block=(6,2),maskFromAlpha=False)
fg.getPoseLabel(testSet,pathToPoseFiles='/local/attale00/poseLabels/multipie/')
#fg.getColorHistogram(testSet,roi,path=path_ea,ending='.png',colorspace='lab',bins=40)
testSet.targetNum=map(utils.mapMouthLabels2Two,testSet.target)
rf=classifierUtils.standardRF(max_features = np.sqrt(len(testSet.data[0])),min_split=5,max_depth=40)
if mode in ['s','v']:
print 'Classifying with loaded classifier'
obj=classifierUtils.classifyWithOld(path,testSet,mode,clfPath = '/home/attale00/Desktop/classifiers/thesis/texture/hog_pose')
pickle.dump(obj,open('hog_pose','w'))
elif mode in ['c']:
print 'cross validation of data'
rValues = classifierUtils.dissectedCV(rf,testSet)
pickle.dump(rValues,open('texture_mp_','w'))
elif mode in ['save']:
print 'saving new classifier'
_saveRF(testSet)
else:
print 'not doing anything'
