class MatchTests:
def __init__(self, samStart=11, samEnd=36):
self.data = GetData(sampleStart=samStart, sampleEnd=samEnd)
self.avgs = self.data.GetAveraged()
self.samStart = samStart
def ClassifyTests(
self,
testDir="/home/brian/PycharmProjects/firstAttempt/HandLetters/"):
matches = []
testImages = [
(self.data.NormalizeImage(cv2.imread(testDir + "/" + j, 0)), j)
for j in os.listdir(testDir)
] #tuble(img,name)
for i in testImages:
name = i[1]
img = i[0]
bestMatch = 0
print type(self.avgs[0])
errorArray = self.avgs[0].astype(np.float_) - img.astype(np.float_)
errorArray = errorArray * errorArray
minError = sum(sum(errorArray))
for j in xrange(len(self.avgs)):
errorArray = self.avgs[j].astype(np.float_) - img.astype(
np.float_)
errorArray = errorArray * errorArray
error = sum(sum(errorArray))
if error < minError:
minError = error
bestMatch = j
matches.append(name + "was identified to be: " +
self.data.GetLetters()[bestMatch + self.samStart])
return matches
def __init__(self, sampStart=11, sampEnd=36):
data = GetData(sampleStart=sampStart, sampleEnd=sampEnd)
self.sampStart = sampStart
self.sampEnd = sampEnd
self.imgs = data.GetListImgs()
self.letters = data.GetLetters()
avgs = data.GetAveraged()
self.avgs = avgs[sampStart - 1:sampEnd]
self.kguesses = np.array([list(j.flatten()) for j in avgs])
self.k = sampEnd - sampStart + 1
self.data = data
if len(layers1) != len(layers2):
return False
else:
for i in range(len(layers1)):
w1 = layers1[i].get_weights()
w2 = layers2[i].get_weights()
if not np.array_equal(w1, w2):
return False
return True
# defining the folders path train and test
TRAINING_DATASET_FOLDER_NAME = '3_preprocessed_1_dataset train'
X_train, Y_train, _ = GetData(TRAINING_DATASET_FOLDER_NAME, limit_value=1)
Y_train = np_utils.to_categorical(Y_train, 2)
X_train = X_train.astype('float32')
X_train /= np.max(X_train)
width = 80
height = 80
depth = 2
num_classes = 2
# load the model architecture from file
a = read_model("models/model01.txt")
modelObject = ModelBuilder.ModelBuilder(a, (height, width, depth))
model = modelObject.model
model.compile(
horizontalalignment="center",
color="white" if cm[i, j] > thresh else "black")
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
a = read_model("models/model08.txt")
modelObject = ModelBuilder(a, (80, 80, 2))
model = modelObject.model
model.load_weights(
'trained_model/2018-07-04 22:29:20/model08.txt_2018-07-05 12:24:21.h5')
#plot_model(model, to_file='model_graph.png', show_shapes=True, show_layer_names=True)
(X_test, y_test, _) = GetData('lfw-whofitinram_p80x80')
Y_test = np_utils.to_categorical(y_test, num_classes)
X_test = X_test.astype('float32')
X_test /= np.max(X_test) # Normalise data to [0, 1] range
model.compile(
loss='categorical_crossentropy', # using the cross-entropy loss function
optimizer='adam', # using the Adam optimiser
metrics=['accuracy'])
model.summary()
#evaluate the mode
print(model.evaluate(X_test, Y_test))
predictions = model.predict(X_test)
#========================PRINT CONFUSION MATRIX==============================
#predict labels
bp = 'trained_model/'
NUM_CLASSES = 2
TEST_DATASET_FOLDER_NAME = '3_preprocessed_1_dataset train'
MAX_IMAGES_TO_PLOT = 36
NUM_PRINTED_PAGES = 3
MODEL_TO_LOAD = '2018-05-06 15:25:00.h5'
a = read_model("models/model1.txt")
modelObject = ModelBuilder.ModelBuilder(a, (80, 80, 2))
model = modelObject.model
model.load_weights(bp + MODEL_TO_LOAD)
#model = load_model(bp + MODEL_TO_LOAD)
(X_test, y_test, _) = GetData(TEST_DATASET_FOLDER_NAME, 5)
Y_test = np_utils.to_categorical(y_test, NUM_CLASSES)
X_test = X_test.astype('float32')
X_test /= np.max(X_test) # Normalise data to [0, 1] range
model.compile(
loss='categorical_crossentropy', # using the cross-entropy loss function
optimizer='adam', # using the Adam optimiser
metrics=['accuracy'])
print(model.evaluate(X_test, Y_test))
sum = 0
for i in range(len(y_test)):
sum += y_test[i]
print(sum)
def CrossValidate(k,
models,
models_array_name,
dataset_folder_name,
batch_size,
num_epochs=200,
chat_id="undefined",
folders_at_the_same_time=20,
max_num_of_validation_folders=12,
validate_every=5,
validation_treshold=0):
avg_val_accuracy_models = []
total_num_folders = len(os.listdir(dataset_folder_name))
folders_each_validation = total_num_folders // k if total_num_folders < max_num_of_validation_folders else max_num_of_validation_folders
timestamp = current_datetime()
path = 'Crossvalidation Results/crossvaliationresults_' + timestamp + '.txt'
# folders where models will be saved
os.mkdir("trained_model/" + timestamp)
with open(path, 'w') as the_file:
the_file.write(current_datetime() + '\n')
for i in range(len(models)):
print("\n validating model: " + models_array_name[i])
sum_model_validations_acc = 0
to_avoid_validation = []
with open(path, 'a') as the_file:
the_file.write('\n \n \n \n model: ' + str(i))
with open(path, 'a') as the_file:
models[i].summary(
print_fn=lambda x: the_file.write('\n' + x + '\n'))
# send a message on telegram when the training of another model is starting
if chat_id != 'undefined':
telegram_send_msg("START TRAINING {}".format(models_array_name[i]))
for j in range(k):
print("\n validation round " + str(j))
(X_validation, Y_validation, validation_folders_list) = GetData(
dataset_folder_name, limit_value=folders_each_validation)
X_validation = X_validation.astype('float32')
X_validation /= np.max(X_validation)
Y_validation = np_utils.to_categorical(Y_validation, 2)
to_avoid_validation = to_avoid_validation + validation_folders_list
validation_history = Train.SingletonTrain().Train(
models[i],
training_dataset_folder_name=dataset_folder_name,
epochs=num_epochs,
batch_size=batch_size,
training_folders_count=folders_at_the_same_time,
validation_x=X_validation,
validation_y=Y_validation,
to_avoid=validation_folders_list,
validate_every=validate_every,
subfolder_name=timestamp,
enable_telegram_bot=(chat_id != "undefined"),
save_model=models_array_name[i],
validation_treshold=validation_treshold)
if len(validation_history) > 0:
sum_model_validations_acc += (validation_history[-1])[1]
avg_val_accuracy_models += [sum_model_validations_acc / k]
with open(path, 'a') as the_file:
the_file.write('\n validation results ' +
str(sum_model_validations_acc / k))
def __init__(self, samStart=11, samEnd=36):
self.data = GetData(sampleStart=samStart, sampleEnd=samEnd)
self.avgs = self.data.GetAveraged()
self.samStart = samStart