def __init__(self):
#data_manager 생성
self.data_manager = Data_manager()
train_sentences = self.data_manager.get_train_sentences()
train_tags = self.data_manager.get_train_tags()
#학습 문장으로 word2_vec_embedder 생성
self.word2vec_embedder = Word2vec_embedder(train_sentences)
#word2vec 방식으로 embedding 된 vector_set
with open('trainDataVecs.pkl', 'rb') as f:
trainDataVecs = pickle.load(f)
#embedding 된 vector_set과 정답지로 Sentiment_analysis 모델 생성
self.sentiment_analysis_model = Sentiment_analysis_model(
trainDataVecs, train_tags)
def __init__(self, parent):
self.parent = parent
self.model = Model()
self.view = View(parent)
self.data_manager = Data_manager()
self.view.setup()
pub.subscribe(self.button_1_pressed, "BUTTON_1_PRESSED")
pub.subscribe(self.button_2_pressed, "BUTTON_2_PRESSED")
pub.subscribe(self.button_3_pressed, "BUTTON_3_PRESSED")
pub.subscribe(self.load_image, "LOAD_IMAGE")
pub.subscribe(self.image_loaded, "IMAGE_LOADED")
pub.subscribe(self.barthel_data_sent, "BARTHEL_DATA_SENT")
pub.subscribe(self.update_barthel, "UPDATE_BARTHEL")
pub.subscribe(self.error_barthel, "ERROR_BARTHEL")
pub.subscribe(self.emina_data_sent, "EMINA_DATA_SENT")
pub.subscribe(self.update_emina, "UPDATE_EMINA")
pub.subscribe(self.error_emina, "ERROR_EMINA")
pub.subscribe(self.analyse_image, "ANALYSE_IMAGE")
pub.subscribe(self.ask_mask_confirmation, "ASK_MASK_CONFIRMATION")
pub.subscribe(self.pre_segmentation_confirmated, "PRE_SEGMENTATION_CONFIRMATED")
pub.subscribe(self.segmentation_gui, "SEGMENTATION_GUI")
pub.subscribe(self.roi_granulation, "ROI_GRANULATION")
pub.subscribe(self.roi_necrosis, "ROI_NECROSIS")
pub.subscribe(self.roi_slough, "ROI_SLOUGH")
pub.subscribe(self.ask_roi_confirmation, "ASK_ROI_CONFIRMATION")
pub.subscribe(self.roi_confirmated, "ROI_CONFIRMATED")
pub.subscribe(self.ask_perimeter, "ASK_PERIMETER")
pub.subscribe(self.update_perimeter_count, "UPDATE_PERIMETER_COUNT")
pub.subscribe(self.update_granulation_count, "UPDATE_GRANULATION_COUNT")
pub.subscribe(self.update_necrosis_count, "UPDATE_NECROSIS_COUNT")
pub.subscribe(self.update_slough_count, "UPDATE_SLOUGH_COUNT")
pub.subscribe(self.tot_ple_ko, "TOT_PLE_KO")
pub.subscribe(self.data_ko, "DATA_KO")
pub.subscribe(self.data_ok, "DATA_OK")
pub.subscribe(self.flash_reduction, "FLASH_REDUCTION")
pub.subscribe(self.flash_confirmated, "FLASH_CONFIRMATED")
pub.subscribe(self.image_accepted, "IMG_ACCEPTED")
pub.subscribe(self.data_files_ko, "DATA_FILES_KO")
pub.subscribe(self.data_n_elements, "DATA_N_ELEMENTS")
pub.subscribe(self.ask_image_i, "ASK_IMAGE_i")
pub.subscribe(self.load_image_i, "IMAGE_LOAD_i")
pub.subscribe(self.load_metadata_i, "METADATA_LOAD_i")
def update_model(self):
#data_manager 생성
self.data_manager = Data_manager()
self.data_manager.update_train_data_set_from_report_data('report.txt')
train_sentences = self.data_manager.get_train_sentences()
train_tags = self.data_manager.get_train_tags()
print(len(train_sentences))
print(len(train_tags))
# 학습 문장으로 word2_vec_embedder 생성
self.word2vec_embedder.make_model(train_sentences)
#word2vec 방식으로 embedding 된 vector_set
trainDataVecs = word2vec_embbeding_module.getAvgFeatureVecs(
train_sentences, self.word2vec_embedder.model,
self.word2vec_embedder.num_features)
with open('trainDataVecs.pkl', 'wb') as f:
pickle.dump(trainDataVecs, f)
#embedding 된 vector_set과 정답지로 Sentiment_analysis 모델 생성
self.sentiment_analysis_model.set_x_train(trainDataVecs)
self.sentiment_analysis_model.set_y_train(train_tags)
self.sentiment_analysis_model.make_model()
class Sentiment_analysis:
def __init__(self):
#data_manager 생성
self.data_manager = Data_manager()
train_sentences = self.data_manager.get_train_sentences()
train_tags = self.data_manager.get_train_tags()
#학습 문장으로 word2_vec_embedder 생성
self.word2vec_embedder = Word2vec_embedder(train_sentences)
#word2vec 방식으로 embedding 된 vector_set
with open('trainDataVecs.pkl', 'rb') as f:
trainDataVecs = pickle.load(f)
#embedding 된 vector_set과 정답지로 Sentiment_analysis 모델 생성
self.sentiment_analysis_model = Sentiment_analysis_model(
trainDataVecs, train_tags)
def Sentiment_analysis_predict_pos_neg(self, sentence):
# sentence, score = self.sentiment_analysis_model.predict_pos_neg_by_loaded_model(sentence,self.word2vec_embedder.model,self.word2vec_embedder.num_features)
sentence, score = self.sentiment_analysis_model.predict_pos_neg(
sentence, self.word2vec_embedder.model,
self.word2vec_embedder.num_features)
return sentence, score
def update_model(self):
#data_manager 생성
self.data_manager = Data_manager()
self.data_manager.update_train_data_set_from_report_data('report.txt')
train_sentences = self.data_manager.get_train_sentences()
train_tags = self.data_manager.get_train_tags()
print(len(train_sentences))
print(len(train_tags))
# 학습 문장으로 word2_vec_embedder 생성
self.word2vec_embedder.make_model(train_sentences)
#word2vec 방식으로 embedding 된 vector_set
trainDataVecs = word2vec_embbeding_module.getAvgFeatureVecs(
train_sentences, self.word2vec_embedder.model,
self.word2vec_embedder.num_features)
with open('trainDataVecs.pkl', 'wb') as f:
pickle.dump(trainDataVecs, f)
#embedding 된 vector_set과 정답지로 Sentiment_analysis 모델 생성
self.sentiment_analysis_model.set_x_train(trainDataVecs)
self.sentiment_analysis_model.set_y_train(train_tags)
self.sentiment_analysis_model.make_model()
sheet = "https://docs.google.com/spreadsheets/d/1ecgbu4-tBPrXJtZzMFKN1YQdBdJ9aE5tKWLGBdDhVE0/edit?usp=sharing"
# Instance ofg the google sheedt class
my_sheet = Google_sheets(sheet)
# Save the list of google sheet data
data = my_sheet.get_data()
# Path of the project
current_path = os.path.dirname(__file__)
# Path of the csv file with local data
csv_file_path = os.path.join(current_path, "data.csv")
# Instance of the data manager class
my_data_manager = Data_manager(csv_file_path)
# Compare the local data and the new data. return new registers
new_data = my_data_manager.compare_data(data)
# Loop for each new register of data
for row in new_data:
# EMAIL CONECTION
# Local email account to send error message
email_server = "*****@*****.**"
password_server = "Superman2019"
# Email information of the client
email_user = row["email"]
len(train_sentences)
len(test_sentences)
test_sentences[50000]
test_label = [int(x) for x in test_label]
# test_sentences[50000:50010]
with open('trainDataVecs.pkl', 'wb') as f:
pickle.dump(trainDataVecs, f)
with open('testDataVecs.pkl', 'wb') as f:
pickle.dump(testDataVecs, f)
len(testDataVecs)
sentiment_analysis.sentiment_analysis_model.model_evaluate(
testDataVecs, test_label)
sentiment_analysis.Sentiment_analysis_predict_pos_neg(test_text[1])[1]
data_manager = Data_manager()
type(test_label[0])
TP = 0
FP = 0
FN = 0
TN = 0
for i in range(len(test_text)):
result = sentiment_analysis.Sentiment_analysis_predict_pos_neg(
test_text[i])[1]
if (test_label_[i] == 1 and result >= 0):
TP = TP + 1
if (test_label_[i] == 1 and result < 0):
FN = FN + 1
if (test_label_[i] == 0 and result >= 0):
FP = FP + 1
if (test_label_[i] == 0 and result < 0):
class Controller:
"""
A class used to handle requests
...
Attributes
----------
parent : Tk
root window
model : Model
a class to process and manage data structures
view : View
a class for data representation and user interaction
data_manager : Data_manager
a class used to manage the files's data reading/writing
Methods
-------
button_1_pressed()
Prints that button_1 from view has been pressed.
button_2_pressed()
Prints that button_2 from view has been pressed and calls the function.
load_data from Data_manager.
button_3_pressed(data)
Prints that button_3 from view has been pressed.
Calls the functions to check metadata and images before saving.
load_image()
Calls the function to load a pressure injury's image.
WARNING: For optimal visualization, images must be '560x390'
image_loaded(image_tk)
Defines a Pressure_img object and saves its image.
Calls the update_image from View to update the label with the loaded image.
Calls the function from View to show "process" button.
barthel_data_sent(data)
Calls the Model function to calculate barthel from data sent.
emina_data_sent(data)
Calls the Model function to calculate emina from data sent.
update_barthel(data)
Calls the update_barthel from View to update the Scale with the value.
update_emina(data)
Calls the update_emina from View to update the Scale with the value.
error_barthel()
View request to check barthel fields.
error_emina()
View request to check emina fields.
analyse_image()
Checks if Pressure_img has been processed and calls processing function if not.
ask_mask_confirmation(img_cv2_mask, scale_factor)
Calls the View function to ask user confirmation about a mask.
ask_roi_confirmation(img_cv2_mask, scale_factor, tissue, scale_factor)
Calls the View function to ask user confirmation about a roi.
pre_segmentation_confirmated(img_imgtk_mask, img_cv2_mask)
Calls the Pressure_img function for the first image segmentation.
segmentation_gui(img_imgtk_mask, img_cv2_mask)
Updates the Pressure_img mask image.
Calls the View function to show GUI for img segmentation to the user.
roi_granulation()
Updates the Pressure_img mask image
Calls the Pressure_img function to allow the user to select the granulation roi.
roi_necrosis()
Updates the Pressure_img mask image
Calls the Pressure_img function to allow the user to select the necrosis roi.
roi_slough()
Updates the Pressure_img mask image
Calls the Pressure_img function to allow the user to select the slough roi.
roi_confirmated()
Calls the Pressure_img function to update granulation/necrosis/slough fields.
ask_perimeter()
Checks if perimeter has been cropped and calls the Pressure_img function if not.
update_perimeter_count()
Updates the View of perimeter label.
update_granulation_count(number)
Updates the View of granulation's label with the number of roi's already selected.
update_necrosis_count(number)
Updates the View of necrosis's label with the number of roi's already selected.
update_slough_count(number)
Updates the View of slough's label with the number of roi's already selected.
tot_ple_ko()
Calls a View function to warn the user that all metadata fields must be filled.
data_ko(error)
Calls a View function to warn the user what field has the wrong input data.
data_ok()
Sets Pressure_img loaded boolean to False.
Calls View function to reset loaded image label.
Calls the function to save all the data entered by the user.
flash_reduction(img_cv2_mask)
Checks if flash reduction has been called and calls
Pressure_img function to reduce flash if not.
Calls the View function to ask user confirmation.
flash_confirmated(img_cv2_flash)
Updates Pressure_img mask and flash_reduced boolean.
Calls the View function to update image label.
image_accepted()
Updates Pressure_img processed boolean to True.
save_data_file()
Calls the Data_manager function to save user introduced data and images.
data_files_ko()
Calls the View function to warn the user about a file manager error.
data_n_elements(num)
Calls the View function to update the element's label counter with a new value.
load_image_i(image_tk)
Calls the View function to load the image_tk to the p2_label_img.
load_metadata_i(metadata)
Calls the View function to load the metadata to the p2_label_metadata.
ask_image_i(i)
Calls the Data_manager functions to read and load the image and metadata "i".
"""
def __init__(self, parent):
self.parent = parent
self.model = Model()
self.view = View(parent)
self.data_manager = Data_manager()
self.view.setup()
pub.subscribe(self.button_1_pressed, "BUTTON_1_PRESSED")
pub.subscribe(self.button_2_pressed, "BUTTON_2_PRESSED")
pub.subscribe(self.button_3_pressed, "BUTTON_3_PRESSED")
pub.subscribe(self.load_image, "LOAD_IMAGE")
pub.subscribe(self.image_loaded, "IMAGE_LOADED")
pub.subscribe(self.barthel_data_sent, "BARTHEL_DATA_SENT")
pub.subscribe(self.update_barthel, "UPDATE_BARTHEL")
pub.subscribe(self.error_barthel, "ERROR_BARTHEL")
pub.subscribe(self.emina_data_sent, "EMINA_DATA_SENT")
pub.subscribe(self.update_emina, "UPDATE_EMINA")
pub.subscribe(self.error_emina, "ERROR_EMINA")
pub.subscribe(self.analyse_image, "ANALYSE_IMAGE")
pub.subscribe(self.ask_mask_confirmation, "ASK_MASK_CONFIRMATION")
pub.subscribe(self.pre_segmentation_confirmated, "PRE_SEGMENTATION_CONFIRMATED")
pub.subscribe(self.segmentation_gui, "SEGMENTATION_GUI")
pub.subscribe(self.roi_granulation, "ROI_GRANULATION")
pub.subscribe(self.roi_necrosis, "ROI_NECROSIS")
pub.subscribe(self.roi_slough, "ROI_SLOUGH")
pub.subscribe(self.ask_roi_confirmation, "ASK_ROI_CONFIRMATION")
pub.subscribe(self.roi_confirmated, "ROI_CONFIRMATED")
pub.subscribe(self.ask_perimeter, "ASK_PERIMETER")
pub.subscribe(self.update_perimeter_count, "UPDATE_PERIMETER_COUNT")
pub.subscribe(self.update_granulation_count, "UPDATE_GRANULATION_COUNT")
pub.subscribe(self.update_necrosis_count, "UPDATE_NECROSIS_COUNT")
pub.subscribe(self.update_slough_count, "UPDATE_SLOUGH_COUNT")
pub.subscribe(self.tot_ple_ko, "TOT_PLE_KO")
pub.subscribe(self.data_ko, "DATA_KO")
pub.subscribe(self.data_ok, "DATA_OK")
pub.subscribe(self.flash_reduction, "FLASH_REDUCTION")
pub.subscribe(self.flash_confirmated, "FLASH_CONFIRMATED")
pub.subscribe(self.image_accepted, "IMG_ACCEPTED")
pub.subscribe(self.data_files_ko, "DATA_FILES_KO")
pub.subscribe(self.data_n_elements, "DATA_N_ELEMENTS")
pub.subscribe(self.ask_image_i, "ASK_IMAGE_i")
pub.subscribe(self.load_image_i, "IMAGE_LOAD_i")
pub.subscribe(self.load_metadata_i, "METADATA_LOAD_i")
def button_1_pressed(self):
"""
Prints that button_1 from view has been pressed.
"""
print("controller - Botó 1!")
def button_2_pressed(self):
"""
Prints that button_2 from view has been pressed and calls the function.
load_data from Data_manager.
"""
print("controller - Botó 2!")
self.data_manager.load_data()
def button_3_pressed(self, data):
"""
Prints that button_3 from view has been pressed.
Calls the functions to check metadata and images before saving.
Parameters
----------
data : list
a list with all the metadata field's information written by the user
"""
print("controller - Botó 3!")
try:
if self.pressure_img.loaded:
if self.pressure_img.processed:
self.model.getData(data)
else:
self.view.popupmsg("És necessari processar la imatge.")
else:
self.view.popupmsg("És necessari carregar una imatge")
except:
self.view.popupmsg("És necessari carregar una imatge")
def load_image(self):
"""
Calls the function to load a pressure injury's image.
WARNING: For optimal visualization, images must be '560x390'
"""
print("controller - carregar imatge")
self.pressure_img.path = self.model.carregar_imatge()
def image_loaded(self, image_tk):
"""
Defines a Pressure_img object and saves its image.
Calls the update_image from View to update the label with the loaded image.
Calls the function from View to show "process" button.
Parameters
----------
image_tk : PIL Image
image ready to be loaded in a label
"""
self.pressure_img = Pressure_img()
self.pressure_img.img_origin = self.model.img_original
self.pressure_img.loaded = True
self.view.update_image(image_tk)
self.view.botoImg()
def barthel_data_sent(self, data):
"""
Calls the Model function to calculate barthel from data sent.
Parameters
----------
data : list
list with all the field's values selected by the user
"""
print("controller - barthel_data_sent!")
self.model.calculateBarthel(data)
def emina_data_sent(self, data):
"""
Calls the Model function to calculate emina from data sent.
Parameters
----------
data : list
list with all the field's values selected by the user
"""
print("controller - emina_data_sent!")
self.model.calculateEmina(data)
def update_barthel(self, data):
"""
Calls the View function update_barthel to update barthel value.
Parameters
----------
data : int
value of the barthel scale
"""
print("controller - update_barthel")
self.view.update_barthel(data)
def update_emina(self, data):
"""
Calls the View function update_emina to update emina value.
Parameters
----------
data : int
value of the emina scale
"""
print("controller - update_emina")
self.view.update_emina(data)
def error_barthel(self):
"""
View request to check barthel fields.
"""
print("controller - error barthel")
self.view.popupmsg("S'han d'omplir tots els camps")
def error_emina(self):
"""
View request to check emina fields.
"""
print("controller - error emina")
self.view.popupmsg("S'han d'omplir tots els camps")
def analyse_image(self):
"""
Checks if Pressure_img has been processed and calls processing function if not.
"""
print("controller - analyse_image!")
if self.pressure_img.processed == False:
self.pressure_img.crop_image(self.pressure_img.path)
else:
self.view.popupmsg("La imatge ja ha estat processada.")
def ask_mask_confirmation(self, img_cv2_mask, scale_factor):
"""
Calls the View function to ask user confirmation about an image's mask.
Parameters
----------
img_cv2_mask : image cv2
image that requires user confirmation
scale_factor : int
image resize value (default = 100)
"""
print("controller - ask_mask_confirmation!")
try:
self.view.ask_mask_confirmation(img_cv2_mask, scale_factor)
except:
self.view.popupmsg("Alguna cosa ha fallat. Torna-ho a intentar!")
def ask_roi_confirmation(self, img_cv2_mask, img_cv2_roi, tissue, scale_factor):
"""
Calls the View function to ask user confirmation about a image's roi.
Parameters
----------
img_cv2_mask : image cv2
image before cropping roi
img_cv2_roi : image cv2
image that requires user confirmation
tissue : String
tissue of the roi
scale_factor : int
image resize value (default = 100)
"""
print("controller - ask_roi_confirmation!")
try:
self.view.ask_roi_confirmation(img_cv2_mask, img_cv2_roi, tissue, scale_factor)
except:
self.view.popupmsg("Alguna cosa ha fallat. Torna-ho a intentar!")
def pre_segmentation_confirmated(self, img_imgtk_mask, img_cv2_mask):
"""
Calls the Pressure_img function for the first image segmentation.
Parameters
----------
img_imgtk_mask : PIL Image
image before cropping roi
img_cv2_mask : image cv2
image that requires user confirmation
"""
print("controller - pre_segmentation_confirmated!")
scale_factor = 100
self.pressure_img.begin_segmentation(img_imgtk_mask=img_imgtk_mask, img_cv2_mask=img_cv2_mask, scale_factor= scale_factor)
def segmentation_gui(self, img_imgtk_mask, img_cv2_mask):
"""
Updates the Pressure_img mask image.
Parameters
----------
img_imgtk_mask : PIL Image
image before cropping roi
img_cv2_mask : image cv2
image that requires user confirmation
"""
print("controller - segmentation_gui!")
self.pressure_img.close_all()
self.pressure_img.mask = img_cv2_mask
self.view.segmentation_gui(img_imgtk_mask, img_cv2_mask)
def roi_granulation(self):
"""
Updates the Pressure_img mask image
Calls the Pressure_img function to allow the user to select the granulation roi.
"""
print("controller - roi_granulation!")
img_cv2_mask = self.pressure_img.mask
self.pressure_img.roi_crop(img_cv2_mask, "Granulation")
def roi_necrosis(self):
"""
Updates the Pressure_img mask image
Calls the Pressure_img function to allow the user to select the necrosis roi.
"""
print("controller - roi_necrosis!")
img_cv2_mask = self.pressure_img.mask
self.pressure_img.roi_crop(img_cv2_mask, "Necrosis")
def roi_slough(self):
"""
Updates the Pressure_img mask image
Calls the Pressure_img function to allow the user to select the slough roi.
"""
print("controller - roi_slough!")
img_cv2_mask = self.pressure_img.mask
self.pressure_img.roi_crop(img_cv2_mask, "Slough")
def roi_confirmated(self, img_cv2_roi, tissue):
"""
Calls the Pressure_img function to update granulation/necrosis/slough fields.
Parameters
----------
img_cv2_roi : image cv2
roi selected by the user
tissue : String
tissue of the roi
"""
print("controller - roi_confirmated!")
self.pressure_img.save_mask_data(img_cv2_roi, tissue)
def ask_perimeter(self):
"""
Checks if perimeter has been cropped and calls the Pressure_img function if not.
"""
print("controller - ask_perimeter!")
img_cv2_mask = self.pressure_img.mask
if self.pressure_img.perimetre_done:
self.view.popupmsg("El perímetre ja ha estat seleccionat")
else:
self.pressure_img.roi_crop(img_cv2_mask, "Perimeter")
def update_perimeter_count(self):
"""
Updates the View of perimeter label.
"""
print("controller - update_perimeter_count!")
self.view.update_perimeter_count()
def update_granulation_count(self, number):
"""
Updates the View of granulation's label with the number of roi's already selected.
Parameters
----------
number : int
number of granulation rois already selected
"""
print("controller - update_granulation_count!")
self.view.update_granulation_count(number)
def update_necrosis_count(self, number):
"""
Updates the View of necrosis's label with the number of roi's already selected.
Parameters
----------
number : int
number of necrosis rois already selected
"""
print("controller - update_necrosis_count!")
self.view.update_necrosis_count(number)
def update_slough_count(self, number):
"""
Updates the View of slough's label with the number of roi's already selected.
Parameters
----------
number : int
number of slough rois already selected
"""
print("controller - update_slough_count!")
self.view.update_slough_count(number)
def tot_ple_ko(self):
"""
Calls a View function to warn the user that all metadata fields must be filled.
"""
print("controller - tot_ple_ko!")
self.view.popupmsg("S'han d'omplir tots els camps")
def data_ko(self, error):
"""
Calls a View function to warn the user what field has the wrong input data.
Parameters
----------
error : list
wrong input data field's name
"""
self.view.show_error(error)
print("controller - data_ko")
def data_ok(self):
"""
Sets Pressure_img loaded boolean to False.
Calls View function to reset loaded image label.
Calls the function to save all the data entered by the user.
"""
print("controller - data_ok")
self.pressure_img.loaded = False
self.view.reset_view()
try:
self.save_data_file()
self.view.popupmsg("Procés finalitzat amb èxit. Prem OK per continuar.")
except:
self.view.popupmsg("Error de gestió de fitxers.")
def flash_reduction(self, img_cv2_mask):
"""
Checks if flash reduction has been called and calls
Pressure_img function to reduce flash if not.
Calls the View function to ask user confirmation.
Parameters
----------
img_cv2_mask : image cv2
image selected by user to reduce its flash
"""
print("controller - flash_reduction!")
if self.pressure_img.flash_reduced == False:
try:
img_flash_reduced = self.pressure_img.flash_reduction(img_cv2_mask)
self.view.ask_flash_confirmation(img_cv2_mask, img_flash_reduced)
except:
self.view.popupmsg("Alguna cosa ha fallat. Torna-ho a intentar!")
else:
self.view.popupmsg("Ja s'ha aplicat la reducció.")
def flash_confirmated(self, img_cv2_flash):
"""
Updates Pressure_img mask and flash_reduced boolean.
Calls the View function to update image label.
Parameters
----------
img_cv2_mask : image cv2
image selected and validated by user to reduce its flash
"""
print("controller - flash_confirmated!")
self.pressure_img.flash_reduced = True
self.pressure_img.mask = img_cv2_flash
self.pressure_img.img = img_cv2_flash.copy()
self.view.update_flash_label(img_cv2_flash)
def image_accepted(self):
"""
Updates Pressure_img processed boolean to True.
"""
print("controller - image_accepted!")
self.pressure_img.processed = True
def save_data_file(self):
"""
Calls the Data_manager function to save user introduced data and images.
"""
print("controller - save_data_file")
self.data_manager.save_data(self.model.metadata, self.pressure_img)
def data_files_ko(self):
"""
Calls the View function to warn the user about a file manager error.
"""
print("controller - data_files_ko")
self.view.popupmsg("Error de gestió dels fitxers.")
def data_n_elements(self, num):
"""
Calls the View function to update the element's label counter with a new value.
Parameters
----------
num : int
number of images found in the storage directory
"""
print("controller - data_n_elements")
self.view.update_data_n_elements(num)
def load_image_i(self, img_tk):
"""
Calls the View function to load the image_tk to the p2_label_img.
Parameters
----------
img_tk : PIL Image
image ready to be loaded to a label
"""
print("controller - load_img_i")
self.view.load_image_i(img_tk)
def load_metadata_i(self, metadata):
"""
Calls the View function to load the metadata to the p2_label_metadata.
Parameters
----------
metadata : JSON Object
data sent to be loaded into a label
"""
print("controller - load_metadata_i")
self.view.load_metadata_i(metadata)
def ask_image_i(self, i):
"""
Calls the Data_manager functions to read and load the image and metadata "i".
Parameters
----------
i : int
id of the image and metadata that have to be read
"""
print("controller - ask_img_i")
self.data_manager.load_img_i(i)
self.data_manager.load_metadata_i(i)
def main():
# set up our command line arguments
parser = argparse.ArgumentParser(
description=
'Set arguments for training of pytorch image recognotion model')
parser.add_argument('data_directory',
type=str,
help='select directory for images to train on')
parser.add_argument('--save_dir',
type=str,
default='model_checkpoint.pth',
help='select directory for saving of model')
parser.add_argument('--arch',
type=str,
default='densenet121',
help='select architecture')
parser.add_argument('--learning_rate',
type=float,
default='.003',
help='select learning rate')
parser.add_argument(
'--hidden_units',
type=int,
default='256',
help='select number of hidden units for each of 3 layers')
parser.add_argument('--epochs',
type=int,
default=5,
help='set number of epochs for training')
parser.add_argument('--dropout',
type=float,
default=.2,
help='change of dropout')
parser.add_argument(
'--print_every',
type=int,
default=10,
help='How often should the model print out training logs')
parser.add_argument(
'--output_size',
type=int,
default=102,
help=
'How many categories should it predict, should be equal to training data'
)
parser.add_argument('--gpu',
type=bool,
default=False,
help='Should pytorch use a gpu for better performance')
# gets our arguments from the command line
in_arg = parser.parse_args()
# instantiates the data_manager class
data_manager = Data_manager()
# gets our dataloaders from the data_manager class, using our arg input as the directory
dataloaders = data_manager.get_dataloaders(in_arg.data_directory)
# creates a Model class
model = Model()
# creates a new model to train on
model.create_model(in_arg.arch, in_arg.hidden_units, in_arg.output_size,
in_arg.dropout)
# sends our model to the gpu if requested
model.use_gpu(in_arg.gpu)
# trains our model
model.train(in_arg.epochs,
dataloaders,
learning_rate=in_arg.learning_rate,
print_every_step=in_arg.print_every)
# tests our trained models accuracy
model.validate(dataloaders)
# saves our trained model for use later
model.save_model(data_manager.data_datasets['train'].class_to_idx,
in_arg.save_dir)