aug.py

import cv2
import numpy as np
import os.path
import errno
import color_aug as ca
import trans_aug as trans

dataset_dir="/home/hoangnt/workspace/Project/DL/License_Plate_Recognition/gray/"
save_dir="/home/hoangnt/workspace/Project/DL/License_Plate_Recognition/auged/"
extension=".jpeg"


class IdentityMetadata():
    def __init__(self, base, file):
        # dataset base directory
        self.base = base
        # image file name
        self.file = file
    def __repr__(self):
        return self.image_path()
    def image_path(self):
        return os.path.join(self.base, self.file) 
    
class IdentityLabels():
    def __init__(self, img_name):
        # image file name
        self.img_name = img_name
    def __repr__(self):
        return self.label()
    def label(self):
        return os.path.join(self.img_name) 
    
def load_metadata(path):
    metadata = []
    for i in os.listdir(path):
        metadata.append(IdentityMetadata(path, i))
    return np.array(metadata)

def load_label(path):
    label = []
    for i in os.listdir(path):
        label.append(IdentityLabels(i.split('.')[0]))
    return np.asarray(label, dtype=np.str)

def create_folder(path):
    # create folder with label
    #print(len(load_label(dataset_dir)))
    for i in range(len(load_label(dataset_dir))):
        filename = save_dir + load_label(dataset_dir)[i]
        #print(filename)
        if not os.path.exists(filename):
            try:
                os.mkdir(filename)
                #print(os.path.dirname(filename))
            except OSError as exc: # Guard against race condition
                if exc.errno != errno.EEXIST:
                    raise

#create_folder(save_dir)

'''
metadata = load_metadata('gray')


#rotate random 10 images
k = 0
while(k < 10):
    for i in range(0, metadata.size):
        dir_to_save_label = save_dir + load_label(dataset_dir)[i]
        img = cv2.imread(str(metadata[i]))
        img = trans.rotate_random_image(img)
        cv2.imwrite(os.path.join(dir_to_save_label , str(k) + extension),img)
    k+=1

'''
'''

for i in range(0, metadata.size):
        dir_to_save_label = save_dir + load_label(dataset_dir)[i]
        img_gaussian_blur = cv2.imread(str(metadata[i]))
        img_blur = cv2.imread(str(metadata[i]))
        img_sharpen = cv2.imread(str(metadata[i]))
        img_zoom = cv2.imread(str(metadata[i]))

        #gaussian blur
        img_gaussian_blur = ca.gausian_blur(img_gaussian_blur, 0.50)
        cv2.imwrite(os.path.join(dir_to_save_label , 'gaussian-' + load_label(dataset_dir)[i] + extension),img_gaussian_blur)

        #zoom
        img_zoom = trans.zoom(img_zoom)
        cv2.imwrite(os.path.join(dir_to_save_label , 'zoom-' + load_label(dataset_dir)[i] + extension),img_zoom)

'''

dir_label_in_auged = [name for name in os.listdir(save_dir)]
#print(dir_label_in_auged)
for i in range(0, len(dir_label_in_auged)):
    metadata = load_metadata(save_dir+dir_label_in_auged[i])
    #print(metadata)
    for j in range(metadata.size):
        #print(metadata[j])
        dir_to_save_label = save_dir + dir_label_in_auged[i]
        img_gaussian_blur = cv2.imread(str(metadata[j]))
        img_zoom = cv2.imread(str(metadata[j]))

        #gaussian blur
        img_gaussian_blur = ca.gausian_blur(img_gaussian_blur, 0.50)
        cv2.imwrite(os.path.join(dir_to_save_label , 'gaussian-' + str(j) + extension),img_gaussian_blur)

        #zoom
        img_zoom = trans.zoom(img_zoom)
        cv2.imwrite(os.path.join(dir_to_save_label , 'zoom-' + str(j) + extension),img_zoom)