def trainClassifier(self):
"""Trainng the classifier begins by aligning any images in the
img directory and putting them into the aligned images
directory. Each of the aligned face images are passed through the
neural net and the resultant embeddings along with their
labels (names of the people) are used to train the classifier
which is saved to a pickle file as a character stream"""
print("trainClassifier called")
path = "/home/hatice/PycharmProjects/newOpenface/mini/cache.t7"
try:
os.remove(path) # Remove cache from aligned images folder
except:
print("Failed to remove cache.t7. Could be that it did not existed in the first place.")
pass
start = time.time()
OUTER_EYES_AND_NOSE = [36, 45, 33]
aligndlib.alignMain("img/","mini/","outerEyesAndNose","/home/hatice/PycharmProjects/newOpenface/models/dlib/shape_predictor_68_face_landmarks.dat",args.imgDim)
print("Aligning images for training took {} seconds.".format(time.time() - start))
done = False
start = time.time()
done = self.generate_representation()
if done is True:
print("Representation Generation (Classification Model) took {} seconds.".format(time.time() - start))
start = time.time()
# Train Model
self.train("/home/hatice/PycharmProjects/newOpenface/generated-embeddings/","LinearSvm",-1)
print("Training took {} seconds.".format(time.time() - start))
else:
print("Generate representation did not return True")
def trainClassifier(self):
"""Trainng the classifier begins by aligning any images in the
training-images directory and putting them into the aligned images
directory. Each of the aligned face images are passed through the
neural net and the resultant embeddings along with their
labels (names of the people) are used to train the classifier
which is saved to a pickle file as a character stream"""
logger.info("trainClassifier called")
path = fileDir + "/aligned-images/cache.t7"
try:
os.remove(path) # Remove cache from aligned images folder
except:
logger.info("Failed to remove cache.t7. Could be that it did not existed in the first place.")
pass
start = time.time()
aligndlib.alignMain("training-images/","aligned-images/","outerEyesAndNose",args.dlibFacePredictor,args.imgDim)
logger.info("Aligning images for training took {} seconds.".format(time.time() - start))
done = False
start = time.time()
done = self.generate_representation()
if done is True:
logger.info("Representation Generation (Classification Model) took {} seconds.".format(time.time() - start))
start = time.time()
# Train Model
self.train("generated-embeddings/","LinearSvm",-1)
logger.info("Training took {} seconds.".format(time.time() - start))
else:
logger.info("Generate representation did not return True")
def trainClassifier(self):
"""Trainng the classifier begins by aligning any images in the
training-images directory and putting them into the aligned images
directory. Each of the aligned face images are passed through the
neural net and the resultant embeddings along with their
labels (names of the people) are used to train the classifier
which is saved to a pickle file as a character stream"""
path = fileDir + "/aligned-images/cache.t7"
try:
os.remove(path) # Remove cache from aligned images folder
except:
print "Tried to remove cache.t7"
pass
start = time.time()
aligndlib.alignMain("training-images/", "aligned-images/",
"outerEyesAndNose", args.dlibFacePredictor,
args.imgDim)
print("\nAligning images took {} seconds.".format(time.time() - start))
done = False
start = time.time()
done = self.generate_representation()
if done is True:
print(
"Representation Generation (Classification Model) took {} seconds."
.format(time.time() - start))
start = time.time()
# Train Model
self.train("generated-embeddings/", "LinearSvm", -1)
print("Training took {} seconds.".format(time.time() - start))
else:
print("Generate representation did not return True")
def trainClassifier(self, database):
"""Trainng the classifier begins by aligning any images in the
training-images directory and putting them into the aligned images
directory. Each of the aligned face images are passed through the
neural net and the resultant embeddings along with their
labels (names of the people) are used to train the classifier
which is saved to a pickle file as a character stream"""
logger.info("trainClassifier called")
path = fileDir + "/aligned-images/cache.t7"
try:
os.remove(path) # Remove cache from aligned images folder
except:
logger.info(
"Failed to remove cache.t7. Could be that it did not existed in the first place."
)
pass
start = time.time()
aligndlib.alignMain("training-images/", "aligned-images/",
"outerEyesAndNose", args.dlibFacePredictor,
args.imgDim)
logger.info("Aligning images for training took {} seconds.".format(
time.time() - start))
done = False
start = time.time()
done = self.generate_representation()
if done is True:
logger.info(
"Representation Generation (Classification Model) took {} seconds."
.format(time.time() - start))
start = time.time()
# Train Model
o_fname, n_fname, o_fname2, n_fname2 = self.train(
"generated-embeddings/", -1, database)
logger.info("Training took {} seconds.".format(time.time() -
start))
else:
logger.info("Generate representation did not return True")
return o_fname, n_fname, o_fname2, n_fname2, done
def trainClassifier(self):
self.trainingEvent.clear(
) #event used to hault face_processing threads to ensure no threads try access .pkl file while it is being updated
path = self.fileDir + "/aligned-images/cache.t7"
try:
os.remove(path) # remove cache from aligned images folder
except:
print "Tried to remove cache.t7"
pass
start = time.time()
aligndlib.alignMain("training-images/", "aligned-images/",
"outerEyesAndNose", self.args.dlibFacePredictor,
self.args.imgDim)
print("\nAligning images took {} seconds.".format(time.time() - start))
done = False
start = time.time()
done = self.generate_representation()
if done is True:
print(
"Representation Generation (Classification Model) took {} seconds."
.format(time.time() - start))
start = time.time()
#Train Model
self.train("generated-embeddings/", "LinearSvm", -1)
print("Training took {} seconds.".format(time.time() - start))
else:
print("Generate representation did not return True")
self.trainingEvent.set() #threads can continue processing
return True
clf_final)])
self.clf.fit(embeddings, labelsNum) #link embeddings to labels
fName = "{}classifier.pkl".format(workDir)
print("Saving classifier to '{}'".format(fName))
with open(fName, 'w') as f:
pickle.dump(
(self.le, self.clf), f
) # Creates character stream and writes to file to use for recognition
#
# def getSquaredl2Distance(self,rep1,rep2):
# """Returns number between 0-4, Openface calculated the mean between
# similar faces is 0.99 i.e. returns less than 0.99 if reps both belong
# to the same person"""
#
# d = rep1 - rep2
if __name__ == "__main__":
a = FaceRecogniser()
# a.trainClassifier()
aligndlib.alignMain(
"input/", "output/", "innerEyesAndBottomLip",
"/home/hatice/PycharmProjects/newOpenface/models/dlib/shape_predictor_68_face_landmarks.dat",
args.imgDim)
a.recognize_face("output/bugraabi.png")
