def Load_pickle():
pickle = Pickle()
label = pickle.load_label_pickle("label_pickle")
feature = pickle.load_feature_pickle("feature_pickle")
return feature,label
def __init__(self, thread):
logging.basicConfig(filename='logging.log', level=logging.DEBUG,
format='%(asctime)s %(message)s')
if thread:
self.my_lock = threading.Lock()
self.my_semaphore = threading.Semaphore(SEMAPHORE_MAX_NUM)
else:
self.my_lock = multiprocessing.Lock()
self.my_semaphore = multiprocessing.Semaphore(SEMAPHORE_MAX_NUM)
Pickle.__init__(self)
def delete_value(self, key):
logging.info("deleting from database")
self.my_lock.acquire()
for user in range(SEMAPHORE_MAX_NUM):
self.my_semaphore.acquire()
Pickle.delete_value(self, key)
for user in range(SEMAPHORE_MAX_NUM):
self.my_semaphore.release()
self.my_lock.release()
logging.info("finished deleting from database")
def set_value(self, key, val):
logging.info("writing to database")
self.my_lock.acquire()
for i in range(SEMAPHORE_MAX_NUM):
self.my_semaphore.acquire()
Pickle.set_value(self, key, val)
for i in range(SEMAPHORE_MAX_NUM):
self.my_semaphore.release()
self.my_lock.release()
logging.info("finished writing to database")
def get_value(self, key):
logging.info("reading from database")
self.my_semaphore.acquire()
val = Pickle.get_value(self, key)
self.my_semaphore.release()
logging.info("finished reading from database")
return val
def Save_pickle():
dataset = Dataset()
save = dataset.read_training_dataset("train")
label, feature = dataset.label_feature(save,"Emotion","Pixels")
pickle = Pickle()
pickle.save_label_pickle(label)
pickle.save_feature_pickle(feature)
metrics=['accuracy'])
epochs = 1000
batch_size = 40
model.fit(feature,
label,
batch_size=batch_size,
epochs=epochs,
validation_split=0.2,
callbacks=callbacks)
if __name__ == "__main__":
pickle = Pickle()
## loading dataset
label = pickle.load_label_pickle("label_pickle")
feature = pickle.load_feature_pickle("feature_pickle")
# converting numbers btw 0-1 and converting format to float50
for count in range(len(feature)):
feature[count] = feature[count] / 255.0
feature[count] = feature[count].astype("float32")
## convertin feature and label to numpy array
feature = numpy.array(feature).reshape(-1, 48, 48, 1)
label = numpy.array(label)
cnn_model(feature, label)