utils_dagm.py

# -*- coding: utf-8 -*-
"""
Utilities for DAGM  images experimentation. 

@author: josemiguelarrieta
"""
import cv2
import numpy as np
import math
import pickle
from shutil import copyfile

def add_salt(image,cl_number):
    """
    Add salt to an Image.
        
    Input data:
        image = image to add salt.
        cl_number = class number of image.
    Output:
        Image with salt.
    """
    
    (Blue, Green, Red) = cv2.split(image)
    length = len(image)
    number_salt = 25000    
    coords = [np.random.randint(0,length,number_salt), np.random.randint(0,length,number_salt)]
    valid_coords = np.array(coords)
    valid_coords.tolist()
    if cl_number == 5:
        color = 255
    else:
        color = 0
    Red[valid_coords.tolist()]= color
    Green[valid_coords.tolist()]= color
    Blue[valid_coords.tolist()]= color
    image_salted = cv2.merge([Blue, Green, Red])
    return image_salted

def add_blur(image2blur,cl_number): 
    """
    Blur and Image
        
    Input data:
        image2blur = image to be blured.
        cl_number = class number of image.
        
    Output:
        Image blurred.
    """
    
    #Select blur parameter acording to class number. 
    if cl_number == 1 or cl_number == 5:
        blur_parameter = 9
    elif cl_number == 2 or cl_number == 3:
        blur_parameter = 11
    else:
        blur_parameter = 13
    blured = cv2.medianBlur(image2blur, blur_parameter)
    return blured

def add_defect_B(c11, c22, A, B, blured, image):
    """
    Add defect type B(Blured) to image
    
    Input:
        c11, c22, A, B: cordinates where defect should be placed
        blured: image blured.
        image : image to add defect.
    
    Output: 
        image = image with defect B
    
    """
    #Select ROI from image
    (cX, cY) = (blured.shape[1] // 2, blured.shape[0] // 2)
    mask = np.zeros(blured.shape[:2], dtype = "uint8")
    cv2.ellipse(mask, (c11,c22), (B/2,A/2),0,0,360,255, -1) 
    masked = cv2.bitwise_and(blured, blured, mask = mask)
    #cv2.imshow("Mask", mask)
    #cv2.imshow("Mask Applied to Image", masked)
    blured = masked
    img2 =  blured
    #create a ROI
    rows,cols,channels = img2.shape
    roi = image[0:rows, 0:cols]
    # Now create a mask  and create its inverse mask also
    img2gray = cv2.cvtColor(img2,cv2.COLOR_BGR2GRAY)
    ret, mask = cv2.threshold(img2gray, 10, 255, cv2.THRESH_BINARY)
    mask_inv = cv2.bitwise_not(mask)
    # Now black-out the area of logo in ROI
    img1_bg = cv2.bitwise_and(roi,roi,mask = mask_inv)
    # Take only region of interest from logo image.
    img2_fg = cv2.bitwise_and(img2,img2,mask = mask)
    # Put logo in ROI and modify the main image
    dst = cv2.add(img1_bg,img2_fg)
    image[0:rows, 0:cols ] = dst
    return image
    
def Defect(X,Y,window_mask):
    """
    Input:
    
    Output: 
    
    """
    if (float(np.count_nonzero(window_mask))/(X*Y)>0.10):
        return True
    else:
        return False

def defect_B_rect_ROI(x1, x, y1, y2):
    """
    Roi rectangle delimitation creation for defect B.
    
    Input: 
        x1, x, y1, y2 : Big Roi coordinates on delimitation for both defects.
    
    Output:
        d1, d2,  d3, d4 : Coordinates for little Roi Delimitation on defect B.
    """
    #Second defect rectangle dimentions
    d1 = x1
    d2 = y1
    d3 = x1+int(x)
    if y2 < y1+int(x)*2:
        d4 = y2
    else:
        d4 = y1+int(x)*2
    return d1, d2,  d3, d4

# Ellipse inside Rectangle [No rotation]
def ellipse_inside_rect(x1,y1,x,d4):
    """
    Return Coordinates of Ellipse inside given rectangle.
    
    Input: 
        x1,y1,x,d4 : Rectangle Coordinates
    
    Output: 
        c11, c22, A, B : Ellipse Coordinates (No rotation)
    
    """
    x11 = x1
    y11 = y1
    x22 = x1 + int(x)
    y22 = d4
    B = x22 - x11
    A = y22 - y11
    c11 = B/2 + x11
    c22 = A/2 + y11
    return c11, c22, A, B
    
def get_coordinates_crop(x1,x2,y1,y2,length,width): 
    """
    Get coordinates to crop ROI in image, useful when ROI es greater than
    image dimentions. 
        
    Input:
    x1,x2,y1,y2:  Coordinates of ROI in image.
    length,width: Dimentions of image
    
    Output: 
    start_x, end_x, start_y, end_y:  Dimentions of ROI cropped to image. 
    """
    if (x1 < 0):
        start_x = 0
    elif (x1 > length):
        start_x = length
    else:
        start_x = x1
    if (x2 < 0):
        end_x = 0
    elif (x2 > width):
        end_x = width
    else:
        end_x = x2    
    if (y1 < 0):
        start_y = 0
    elif (y1 > length):
        start_y = length
    else:
        start_y = y1
    if (y2 < 0):
        end_y = 0
    elif (y2 > width):   
        end_y = width
    else:
        end_y = y2        
    return start_x, end_x, start_y, end_y
    
def get_labels_defectA(path,class_number,num,exp='',defect='',degrees=False):
    """
    Input:
        path = Path where labels are located.
        class_number = class number
        num = number of image
        exp = check if is for an experiment or not.
        defect = Type of defect.
        degrees = check if necessary to convert to degrees.
    
    Output:
        gt : Dictionary with ground truth for defect A.
        
    Note: Paths are fixed.(Fix)
    """
    i = num - 1
    if exp == True:
        if defect == 'A' or defect =='AB':
            pathF = path+str(class_number)+'/'+'Cl_'+str(class_number)+'_'+defect
            f = open(pathF+'/'+'labels_defect_A.txt', 'r')
    else: 
        pathF = path+str(class_number)+'_def'
        f = open(pathF+'/'+'labels.txt', 'r')
    lines = f.readlines()
    gt = ground_truth_defectA(lines,i,degrees=degrees)
    return gt
              
def get_labels_defectB(path,class_number,num,exp='',defect=''):
    """
    Input:
        path = Path where labels are located.
        class_number = class number
        num = number of image
        exp = check if is for an experiment or not.
        defect = Type of defect.
        degrees = check if degrees already on.
    
    Output:
        gt =  Dictionary with ground truth for defect B.
        
    Note: Paths are fixed.(Fix)
    """
    i = num - 1
    if exp == True:
        if defect == 'B' or defect =='AB':
            pathF = path+str(class_number)+'/'+'Cl_'+str(class_number)+'_'+defect
            f = open(pathF+'/'+'labels_defect_B.txt', 'r')
    else: 
        pathF = path+str(class_number)+'_defAB'
        f = open(pathF+'/'+'labels_defect_B.txt', 'r')
    lines = f.readlines()
    gt = ground_truth_defectB(lines,i)
    return gt
    
def get_roi_rect(path,class_number,num,exp='',defect=''):
    """
    Get Roi Coordinates of defects
    
    Input:
        path = Path where labels are located.
        class_number = class number
        num = number of image
        exp = check if is for an experiment or not.
        defect = Type of defect.
        
        Output:
        
        Note: Paths are fixed.(Fix)
    """
    line_number = num -1 
    if exp == True:
        pathF = path+str(class_number)+'/'+'Cl_'+str(class_number)+'_'+defect
        f = open(pathF+'/'+'ROI.txt', 'r')  
    else:
        pathF = path+str(class_number)+'_defAB'   
        f = open(pathF+'/'+'ROI.txt', 'r')
    lines = f.readlines()
    line = lines[line_number].split("\t") 
    number = int(line[0])
    x1 = int(float(line[1]))
    y1 = int(float(line[2]))
    x2 = int(float(line[3]))
    y2 = int(float(line[4]))
    return {'number':number, 'x1':x1, 
            'y1':y1, 'x2':x2, 
            'y2':y2}
    
def ground_truth_defectB(lines,line_number):
    """
    Input:
        lines = All lines on file .txt
        line_number = specific line on lines to read.
    Output:
        dictionary with labels
    
    Note: Paths are fixed.(Fix)
    """
    line = lines[line_number].split("\t") #Change i to Numbers 
    number = int(line[0])
    x1 = int(float(line[1]))
    y1 = int(float(line[2]))
    x2 = int(float(line[3]))
    y2 = int(float(line[4]))
    return {'number':number, 'x1':x1, 
            'y1':y1, 'x2':x2, 
            'y2':y2}
    
def ground_truth_defectA (lines,line_number,degrees=False):
    """
    Input: 
        lines = All lines on file .txt
        line_number = specific line on lines to read.
        degrees = check if degrees already on.
    
    Output:
        dictionary with labels
    
    """
    line = lines[line_number].split("\t") #Change i to Numbers 
    number = int(line[0])
    semi_major_ax = int(float(line[1]))
    semi_minor_ax = int(float(line[2]))
    if degrees==True:
        rotation_angle = int(line[3])
    else:
        rotation_angle = math.degrees(float(line[3]))
    x_position_center = int(float(line[4]))
    y_position_center = int(float(line[5]))
    return {'number':number, 'semi_major_ax':semi_major_ax, 
            'semi_minor_ax':semi_minor_ax, 'rotation_angle':rotation_angle, 
            'x_position_center':x_position_center, 'y_position_center':y_position_center}
    
def load_image_dagm(path,num_image,class_number,defect = '_def',exp = ''):
    """
    Input: 
        path = Path where images located.
        num_image = number of image
        class_number = class number
        defect = defect type
        exp = if it is for experiment
    
    Output: 
        image = Returned image
    """
    
    if exp == True:
        pathF = path+str(class_number)+'/Cl_'+str(class_number)+'_'+defect
        if defect =='NO':
            defect = ''
        filename = 'Cl_'+str(class_number)+'_'+str(num_image)+'_'+defect+'.png'
        image = cv2.imread(pathF+'/'+filename)
    else:
        filename = str(num_image)+'.png'
        pathF = path+str(class_number)+defect
        image = cv2.imread(pathF+'/'+filename)    
    return image

def load_labels_dagm(path,class_number,num,exp=''):
    """
    Alternative function to load labels A.
    
    see get_labels_defectA
    """
    i = num - 1
    pathF = path+str(class_number)+'_def'
    f = open(pathF+'/'+'labels.txt', 'r')
    lines = f.readlines()
    gt = ground_truth_defectA(lines,i)
    return gt
    
def load_list_selected_images(defect,class_number,number_experimet):
    """
    Input:
        defect = defect type
        class_number = Class number
        number_experimet = Number of experiment
        
    Output:
        lista: Lista of selected images
    
    """
    if defect == '':
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_NO/'
    elif defect == 'A':
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_A/'
    else:
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_'+str(defect)+'/'
    if defect == '':
        list_name = 'list_no_defect.pckl'
    elif defect == 'A':
        list_name = 'list_defect_A.pckl'
    elif defect == 'AB':
        list_name = 'list_defect_AB.pckl'
    elif defect == 'B':
        list_name = 'list_defect_B.pckl'
    f = open(dst+list_name, 'rb')
    lista = pickle.load(f)
    f.close()
    return lista

def LoadDataEvaluation(ClassNumber,LabelType):
    if LabelType == 'SL':
        path_data = '/Users/josemiguelarrieta/Documents/SIVA/ExperimentsData2/defectAB/'+'class'+str(ClassNumber)+'/'+LabelType+'/'
        f = open(path_data+'BagsSL.pckl', 'rb') 
        BagsSL = pickle.load(f)
        f.close()
        f = open(path_data+'BagLabelsSL.pckl', 'rb') 
        BagLabelsSL = pickle.load(f)
        f.close()
        f = open(path_data+'InstanceBagLabelSL.pckl', 'rb') 
        InstanceBagLabelSL = pickle.load(f)
        f.close()
        X = BagsSL
        Y = BagLabelsSL
        Z = InstanceBagLabelSL
    if LabelType == 'ML':
        path_data = '/Users/josemiguelarrieta/Documents/SIVA/ExperimentsData2/defectAB/'+'class'+str(ClassNumber)+'/'+LabelType+'/'
        f = open(path_data+'BagsML.pckl', 'rb') 
        BagsML = pickle.load(f)
        f.close()
        f = open(path_data+'BagLabelsML.pckl', 'rb') 
        BagLabelsML = pickle.load(f)
        f.close()
        f = open(path_data+'InstanceBagLabelML.pckl', 'rb') 
        InstanceBagLabelML = pickle.load(f)
        f.close()
        X = BagsML
        Y = BagLabelsML
        Z = InstanceBagLabelML
    return X,Y,Z

def move_selected_images(lista,defect,class_number,number_experimet):
    """
    Input:
        lista = Lista of selected images
        defect = Type of defect.
        class_number = Class number
        number_experimet = Number of experiment
    
    Output:
        return 1 when finish.
    
    Note: Paths are fixed.(Fix)
    """
    path = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'
    if defect =='':
        src = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(class_number)+'/'
        addon = ''
    elif defect == 'A':
        src = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(class_number)+'_def/'
        addon = ''
    else:
        src = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(class_number)+'_def'+defect+'/'
        addon = '_'
    if defect == '':
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_NO/'
    else:
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_'+str(defect)+'/'
    file_type = '.png'
    contador = 1
    for i in range (0,len(lista)):
        print i
        if defect == '':
            filename = str(lista[i])+file_type
        elif defect == 'A':
            filename = str(lista[i])+file_type
        else :
            filename = str(lista[i])+addon+defect+file_type
        filename_dest = 'Cl_'+ str(class_number)+'_'+str(i+1)+'_'+defect+file_type
        copyfile(src+filename, dst+filename_dest)
        if defect == 'A': 
            cord_defect_A = get_labels_defectA(path,class_number,int(lista[i]))
            cord_roi = get_roi_rect(path,class_number,int(lista[i]))
            write_labels_defectA(class_number,lista[i],cord_defect_A,reason='new_experiment',new_num=contador,defect=defect)
            write_labels_expROI(class_number,lista[i],cord_roi['x1'],cord_roi['y1'],cord_roi['x2'],cord_roi['y2'],reason='new_experiment',new_num=contador,defect=defect)
        elif defect == 'B':
             cord_defect_B = get_labels_defectB(path,class_number,int(lista[i]))
             cord_roi = get_roi_rect(path,class_number,int(lista[i]))
             write_labels_defectB(class_number,lista[i],cord_defect_B['x1'],cord_defect_B['y1'],cord_defect_B['x2'],cord_defect_B['y2'],reason='new_experiment',new_num=contador,defect=defect)
             write_labels_expROI(class_number,lista[i],cord_roi['x1'],cord_roi['y1'],cord_roi['x2'],cord_roi['y2'],reason='new_experiment',new_num=contador,defect=defect)
        elif defect == 'AB': 
            cord_defect_A = get_labels_defectA(path,class_number,int(lista[i]))
            cord_defect_B = get_labels_defectB(path,class_number,int(lista[i]))
            cord_roi = get_roi_rect(path,class_number,int(lista[i]))
            write_labels_defectA(class_number,lista[i],cord_defect_A,reason='new_experiment',new_num=contador,defect=defect)
            write_labels_defectB(class_number,lista[i],cord_defect_B['x1'],cord_defect_B['y1'],cord_defect_B['x2'],cord_defect_B['y2'],reason='new_experiment',new_num=contador,defect=defect)
            write_labels_expROI(class_number,lista[i],cord_roi['x1'],cord_roi['y1'],cord_roi['x2'],cord_roi['y2'],reason='new_experiment',new_num=contador,defect=defect)
        else:
            print ('No defect')
        contador+=1
    return 1
    
def rectangle_expanded_roi(gt):
    """
    Create a rectangle ROI based on ellypse coordinates.
        
    Input: 
        gt: Ground truth of Ellipse Coordinates.
    
    Output: 
        x1, y1, x2, y2, x, y Coordinates of ROI
    
    """
    c1 = gt['x_position_center']
    c2 = gt['y_position_center']
    major_axis = gt['semi_major_ax']
    minus_axis =gt['semi_minor_ax']
    angle = gt['rotation_angle']
    ratation_angle_rad = np.deg2rad(angle)
    x = np.sqrt(major_axis**2*pow(np.cos(ratation_angle_rad),2) + minus_axis**2*pow(np.sin(ratation_angle_rad),2))
    y = np.sqrt(major_axis**2*pow(np.sin(ratation_angle_rad),2) + minus_axis**2*pow(np.cos(ratation_angle_rad),2))
    x1 = c1 - int(x)*2
    y1 = c2 - int(y)*2
    x2 = c1 + int(x)*2
    y2 = c2 + int(y)*2
    return x1, y1, x2, y2, x, y
    
def RemodeNanInstances(bag,bagInstanceLabels):
    """
    Remove Instances with Nan columns
    """
    n = len(bag)
    for i in range(0,n):
        A = bag[i]
        B = bagInstanceLabels[i]
        nanrows = np.unique(np.where(np.isnan(A))[0])
        if len (nanrows)>0: 
            A = np.delete(A, nanrows, 0)
            B = np.delete(B, nanrows, 0)
        bag[i] = A
        bagInstanceLabels[i]=B
    return bag,bagInstanceLabels,
    
def save_image_defect(defect,num,cl_number,image):
    """
    Save Image with defect.
    
    Input:
        defect = Type of defect
        num = number of image
        cl_number = Class number
        image = image to be saved.
        
    Note: Paths are fixed.(Fix)
    """
    filename_with_defect = str(num) + '_'+defect+'.png'
    cv2.imwrite('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(cl_number)+"_def"+defect+"/"+filename_with_defect, image)
     
def save_list_selected_images(lista,defect,class_number,number_experimet):
    """
    Save lista for defect type.
        
    Note: Paths are fixed.(Fix)
    """
    if defect == '':
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_NO/'
    elif defect == 'A':
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_A/'
    else:
        dst = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_'+str(number_experimet)+'_DAGM/Class'+str(class_number)+'/Cl_'+str(class_number)+'_'+str(defect)+'/'
    if defect == '':
        list_name = 'list_no_defect.pckl'
    elif defect == 'A':
        list_name = 'list_defect_A.pckl'
    elif defect == 'AB':
        list_name = 'list_defect_AB.pckl'
    elif defect == 'B':
        list_name = 'list_defect_B.pckl'
    f = open(dst+list_name, 'wb')
    pickle.dump(lista, f)
    f.close()
    return 1
    
def select_images(defect,class_number):
    """
    Load images from class number to user select witch ones to use in the experiments.
    
    Input: 
        defect = Defect type
        class_number = class number.
    Output: 
        list_images = Array with number of images to select.
    """
    list_images = []
    file_type = '.png'
    if defect =='':
        src = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(class_number)+'/'
    elif defect == 'A':
        src = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(class_number)+'_def/'
    else:
        src = '/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(class_number)+'_def'+defect+'/'

    for i in range (1,151):
        if defect =='':
            filename = str(i)+file_type
        elif defect == 'A':
            filename = str(i)+file_type
        else:
            filename = str(i)+'_'+defect+file_type
        image = cv2.imread(src+filename) 
        cv2.imshow('Image',image)
        var = raw_input("Do you want to use this Image?: ")
        if var == '1': 
            print ('Image '+str(i)+' added')
            list_images.append(i)
        else:
            print ('Image '+str(i)+' discarded ')
        cv2.destroyAllWindows()
        if len(list_images)>=100:
            break
    return list_images
    
def sliding_window(image, image2 ,stepSize, windowSize):
    """
    Sliding windows implementation.    
    """
    # slide a window across the image
    for y in xrange(0, image2.shape[0], stepSize):
        for x in xrange(0, image2.shape[1], stepSize):
            # yield the current window
            yield (x, y, image[y:y + windowSize[1], x:x + windowSize[0]],image2[y:y + windowSize[1], x:x + windowSize[0]])


def write_labels_defectA(cl_number,num,gt,reason='',new_num='',defect=''):
    """
    Input:
        cl_number = Class number
        num = Number of Image
        gt = Ground truth Labels A
        reason= if is for experiment
        new_num = new number of Image
        defect = Defect Type.
    
    Output: 
        Return true
    
    Note: Paths are fixed.(Fix)
    """
    if reason == 'new_experiment':
        if defect=='A':
            target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_1_DAGM/Class'+str(cl_number)+'/Cl_'+str(cl_number)+'_A/'+'labels_defect_A.txt', 'a')
            line = str(new_num)+'\t'+ str(gt['semi_major_ax'])+'\t'+ str(gt['semi_minor_ax'])+'\t'+ str(int(gt['rotation_angle']))+'\t'+ str(gt['x_position_center'])+'\t'+ str(gt['y_position_center'])
        else:
            target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_1_DAGM/Class'+str(cl_number)+'/Cl_'+str(cl_number)+'_'+str(defect)+'/'+'labels_defect_A.txt', 'a')
            line = str(new_num)+'\t'+ str(gt['semi_major_ax'])+'\t'+ str(gt['semi_minor_ax'])+'\t'+ str(int(gt['rotation_angle']))+'\t'+ str(gt['x_position_center'])+'\t'+ str(gt['y_position_center'])
    else:
        target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(cl_number)+'_defAB/'+'labels_defect_A.txt', 'a')
        line = str(num)+'\t'+ str(gt['semi_major_ax'])+'\t'+ str(gt['semi_minor_ax'])+'\t'+ str(int(gt['rotation_angle']))+'\t'+ str(gt['x_position_center'])+'\t'+ str(gt['y_position_center'])
    target.write(line)
    target.write("\n")
    target.close()
    return True
    
def write_labels_defectB(cl_number,num,x1,y1,x2,y2,reason='',new_num='',defect=''):
    """
    Input:
        cl_number = Class number
        num = Number of Image
        x1,y1,x2,y2 = Coordinates defect B Roi
        reason = if it is for experiment
        new_num = new number of Image
        defect = defect Type
    
    Output: 
        Return true
    
    Note: Paths are fixed.(Fix)
    """
    if reason == 'new_experiment':
        target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_1_DAGM/Class'+str(cl_number)+'/Cl_'+str(cl_number)+'_'+str(defect)+'/'+'labels_defect_B.txt', 'a')
        line = str(new_num)+'\t'+str(x1)+'\t'+str(y1)+'\t'+str(x2)+'\t'+str(y2)
    else:
        target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(cl_number)+'_def'+defect+'/'+'labels_defect_B.txt', 'a')
        line = str(num)+'\t'+str(x1)+'\t'+str(y1)+'\t'+str(x2)+'\t'+str(y2)
    target.write(line)
    target.write("\n")
    target.close()
    return True
    
def write_labels_expROI(cl_number,num,x1,y1,x2,y2,reason='',new_num='',defect=''):
    """
    Input:
        cl_number = Class number
        num = Number of Image
        x1,y1,x2,y2 = Coordinates of ROI
        reason= if it is for experiment
        new_num= new number of Image
        defect= defect Type
    
    Output: 
        Return true
    
    Note: Paths are fixed.(Fix)
    """
    if reason == 'new_experiment':
        target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/Experiment_1_DAGM/Class'+str(cl_number)+'/Cl_'+str(cl_number)+'_'+str(defect)+'/'+'ROI.txt', 'a')
        line = str(new_num)+'\t'+str(x1)+'\t'+str(y1)+'\t'+str(x2)+'\t'+str(y2)
    else:
        target = open('/Users/josemiguelarrieta/Dropbox/11_Semestre/Jovenes_Investigadores/images/optical2/Class'+str(cl_number)+'_def'+defect+'/'+'ROI.txt', 'a')
        line = str(num)+'\t'+str(x1)+'\t'+str(y1)+'\t'+str(x2)+'\t'+str(y2)
    target.write(line)
    target.write("\n")
    target.close()
    return True
    
def WeakLabeling(path,num,class_number,defect = 'X',exp = True):
    """    
    This function was implemented for using WeekLabeling. 
    
    Input:
        path: Path Where images are located.
        num :Number of Image.
        class_number: Class number of image.
        defect: Type of defect.
        exp: if it is a experiment (default: True)
    
    Output: 
        roi: ROI cropped. 
        roi_mask: Mask of ROI cropped.
    """
    image = load_image_dagm(path,num,class_number,defect = defect,exp = True)
    length, width,_ = image.shape
    #Create Mask of Image Uploaded!.
    mask = np.zeros(image.shape[:2], dtype = "uint8") 
    #get roi
    roi_labels = get_roi_rect(path,class_number,num,exp=True,defect=defect)
    if defect == 'A': 
        #get labels defectA
        dfA_A = get_labels_defectA(path,class_number,num,exp=True,defect='A',degrees=True)    
        #Draw Defect on Mask
        cv2.ellipse(mask,(dfA_A['x_position_center'],dfA_A['y_position_center']),(dfA_A['semi_major_ax'],dfA_A['semi_minor_ax']),dfA_A['rotation_angle'],0,360,255,-1)  #Draw Ellipse [Ground Truth]
    elif defect == 'B':
        #get labels defectBs
        dfB_B = get_labels_defectB(path,class_number,num,exp=True,defect='B')
        #Draw Defect on Mask
        cv2.rectangle(mask,(dfB_B['x1'],dfB_B['y1']),(dfB_B['x2'],dfB_B['y2']),255,-1)
    elif defect == 'AB':
        dfAB_A = get_labels_defectA(path,class_number,num,exp=True,defect='AB',degrees=True)
        dfAB_B = get_labels_defectB(path,class_number,num,exp=True,defect='AB')
        maskA = np.zeros(image.shape[:2], dtype = "uint8")
        cv2.ellipse(maskA,(dfAB_A['x_position_center'],dfAB_A['y_position_center']),(dfAB_A['semi_major_ax'],dfAB_A['semi_minor_ax']),dfAB_A['rotation_angle'],0,360,255,-1)
        maskB = np.zeros(image.shape[:2], dtype = "uint8")
        cv2.rectangle(maskB,(dfAB_B['x1'],dfAB_B['y1']),(dfAB_B['x2'],dfAB_B['y2']),255,-1)
        #Week labeling [Cropping-ROI]        
        start_x, end_x, start_y, end_y = get_coordinates_crop(roi_labels['x1'],roi_labels['x2'],roi_labels['y1'],roi_labels['y2'],length,width)
        roi = image[start_y:end_y,start_x:end_x]      
        roi_maskA = maskA[start_y:end_y,start_x:end_x]
        roi_maskB = maskB[start_y:end_y,start_x:end_x]
        return roi,roi_maskA,roi_maskB
    else:
        return 0,0
    #Week labeling [Cropping-ROI]
    start_x, end_x, start_y, end_y = get_coordinates_crop(roi_labels['x1'],roi_labels['x2'],roi_labels['y1'],roi_labels['y2'],length,width)
    roi = image[start_y:end_y,start_x:end_x]      
    roi_mask = mask[start_y:end_y,start_x:end_x]
    return roi,roi_mask