refactoring 🔨

data_extractor/extract_data.py → DataExtractor/extract_data.py

@@ -1,4 +1,4 @@
-from data_extractor.extractor import Extractor
+from DataExtractor.extractor import Extractor
 
 
 def main():

data_extractor/extractor.py → DataExtractor/extractor.py

@@ -8,8 +8,8 @@
 
 from natsort import natsorted
 from tqdm import tqdm
-from isolator.isolator import Isolator
-from trainer.models.model_gnet_light import ModelGNetLight
+from Isolator.isolator import Isolator
+from Trainer.Models.model_gnet_light import ModelGNetLight
 from tensorflow.python.keras.preprocessing.image import ImageDataGenerator
 from tensorflow.python.keras.preprocessing.image import array_to_img, img_to_array, load_img
 
@@ -131,7 +131,7 @@ def create_random_images(self, category, count):
         category_dir = os.path.join(self.current_working_dir, constants.OUTPUT_DATA_DIR, category)
 
         for index in tqdm(range(count)):
-            if constants.USE_GRAYSCALE:
+            if constants.USE_GRAY_SCALE:
                 image_rand = np.random.randint(0, 255, size=(constants.IMG_SIZE, constants.IMG_SIZE), dtype=np.uint8)
             else:
                 image_rand = np.random.randint(0, 255,
@@ -157,7 +157,7 @@ def create_training_data(self):
                     img_array = cv2.imread(os.path.join(category_dir, img))
                     if img_array is not None:
                         # convert image to grayscale if parameter is set in constants file
-                        if constants.USE_GRAYSCALE:
+                        if constants.USE_GRAY_SCALE:
                             img_array = cv2.cvtColor(img_array, cv2.COLOR_BGR2GRAY)
                         # resize to normalize data size
                         new_array = cv2.resize(img_array, (constants.IMG_SIZE, constants.IMG_SIZE))
@@ -210,7 +210,7 @@ def categorize_with_trained_model(self):
                 image_path = os.path.join(extracted_data_dir, img)
                 img_array = cv2.imread(image_path)
                 if img_array is not None:
-                    if constants.USE_GRAYSCALE:
+                    if constants.USE_GRAY_SCALE:
                         image_array = cv2.cvtColor(img_array, cv2.COLOR_RGB2GRAY)
                     resized_image_array = cv2.resize(image_array, (constants.IMG_SIZE, constants.IMG_SIZE))
                     reshaped_image = resized_image_array.reshape(-1, constants.IMG_SIZE, constants.IMG_SIZE,

isolator/isolator.py → Isolator/isolator.py

@@ -1,9 +1,9 @@
 import cv2
 import constants
 import numpy as np
-from isolator.isolator_constants import IsolatorConstants
-from isolator.isolator_constants_320_240 import IsolatorConstants320240
-from isolator.isolator_constants_640_480 import IsolatorConstants640480
+from Isolator.isolator_constants import IsolatorConstants
+from Isolator.isolator_constants_320_240 import IsolatorConstants320240
+from Isolator.isolator_constants_640_480 import IsolatorConstants640480
 
 
 class Isolator:
@@ -16,7 +16,7 @@ def get_regions_of_interest(self, image):
         # 0: roi, 1: type
         regions_of_interest = []
 
-        if constants.USE_GRAYSCALE:
+        if constants.USE_GRAY_SCALE:
             image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 
         if self.CONSTANTS is None:
@@ -40,7 +40,7 @@ def get_contours_and_rois(self, image):
         # 0: roi, 1: type
         contours_signal_type = []
 
-        if constants.USE_GRAYSCALE:
+        if constants.USE_GRAY_SCALE:
             image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 
         if self.CLASS_CONSTANTS is not None:
@@ -64,7 +64,7 @@ def get_contours_and_rois(self, image):
         return contours_signal_type
 
     def __set_constants(self, image):
-        if constants.USE_GRAYSCALE:
+        if constants.USE_GRAY_SCALE:
             image_height, image_width = image.shape
         else:
             image_height, image_width, _ = image.shape
@@ -84,7 +84,7 @@ def __crop(self, image):
         width_start = self.CONSTANTS.CROP_WIDTH_START
         width_end = self.CONSTANTS.CROP_WIDTH_END
 
-        if constants.USE_GRAYSCALE:
+        if constants.USE_GRAY_SCALE:
             image_info = image[info_start:info_end, width_start:width_end]
             image_stop = image[stop_start:stop_end, width_start:width_end]
         else:
@@ -102,7 +102,7 @@ def __detect_edges(self, channel):
         return sobel
 
     def __preprocess(self, image):
-        if constants.USE_GRAYSCALE:
+        if constants.USE_GRAY_SCALE:
             # create gradient image from gray scale image
             # for better performance (only 1 channel)
             image = self.__detect_edges(image)

isolator/isolator_constants_320_240.py → Isolator/isolator_constants_320_240.py

@@ -1,4 +1,4 @@
-from isolator.isolator_constants import IsolatorConstants
+from Isolator.isolator_constants import IsolatorConstants
 
 
 class IsolatorConstants320240(IsolatorConstants):

isolator/isolator_constants_640_480.py → Isolator/isolator_constants_640_480.py

@@ -1,4 +1,4 @@
-from isolator.isolator_constants import IsolatorConstants
+from Isolator.isolator_constants import IsolatorConstants
 
 
 class IsolatorConstants640480(IsolatorConstants):

ReadMe.md

@@ -12,9 +12,9 @@ and a category for 'not numbers'.
 ### Includes
 
 - DataExtractor (to extract data for the dataset)
+- Isolator (to find the numbers inside an image)
 - Trainer (to train the model)
 - Tester (to test the model on custom images)
-- Utils (test, simulate the isolator on images)
 
 ## Install
 

tester/test_model.py → Tester/test_model.py

@@ -1,20 +1,22 @@
 import numpy as np
-from tester import Tester
+from Tester.tester import Tester
+from Trainer.Models.model_gnet_light import ModelGNetLight
+from Trainer.Models.model_gnet_deep import ModelGNetDeep
 
 
 def main():
-    tester = Tester()
+    tester = Tester(ModelGNetDeep('GNet'))
 
     # test the model with a given image
-    # tester.test_model_with_image('frame_overlaying_cnt_2.jpg')
+    # Tester.test_model_with_image('frame_overlaying_cnt_2.jpg')
 
     # test the model with a random color array
     # image_random_color = np.random.randint(0, 255, size=(28, 28, 3), dtype=np.uint8)
-    # tester.test_model_with_array(image_random_color)
+    # Tester.test_model_with_array(image_random_color)
 
     # test the model with a random gray scale array
     # image_random_gray = np.random.randint(0, 255, size=(28, 28), dtype=np.uint8)
-    # tester.test_model_with_array(image_random_gray)
+    # Tester.test_model_with_array(image_random_gray)
 
     # test the model with a folder of images
     tester.test_model_with_folder('continuous')

tester/tester.py → Tester/tester.py

@@ -4,15 +4,13 @@
 import shutil
 import numpy as np
 from tqdm import tqdm
-from isolator.isolator import Isolator
-from trainer.models.model_gnet_light import ModelGNetLight
+from Isolator.isolator import Isolator
 
 
 class Tester:
 
-    def __init__(self):
+    def __init__(self, model_obj):
         model_path = "{}.h5".format(constants.MODEL_DIR)
-        model_obj = ModelGNetLight('GNet')
         model_obj.create_model(weights_path=model_path)
         self.model = model_obj.model
         self.model.summary()

trainer/models/model.py → Trainer/Models/model.py

@@ -10,7 +10,7 @@
 from tensorflow.python.keras.optimizers import Adam
 from tensorflow.python.framework.graph_util import convert_variables_to_constants
 from tensorflow.python.keras.callbacks import TensorBoard
-from trainer.utils.tensorboard_filter_visualisation import TensorBoardFilterVisualisation
+from Trainer.Utils.tensorboard_filter_visualisation import TensorBoardFilterVisualisation
 
 # only show tensorflow errors
 tf.logging.set_verbosity(tf.logging.ERROR)

trainer/models/model_gnet_deep.py → Trainer/Models/model_gnet_deep.py

@@ -1,5 +1,5 @@
 import constants
-from trainer.models.model import Model
+from Trainer.Models.model import Model
 from tensorflow.python.keras.models import Sequential
 from tensorflow.python.keras.layers import Dense, Dropout, Activation, Flatten
 from tensorflow.python.keras.layers import Conv2D, MaxPooling2D

trainer/models/model_gnet_light.py → Trainer/Models/model_gnet_light.py

@@ -1,5 +1,5 @@
 import constants
-from trainer.models.model import Model
+from Trainer.Models.model import Model
 from tensorflow.python.keras.models import Sequential
 from tensorflow.python.keras.layers import Dense, Dropout, Activation, Flatten
 from tensorflow.python.keras.layers import Conv2D, MaxPooling2D

trainer/models/model_gnet_light_v2.py → Trainer/Models/model_gnet_light_v2.py

@@ -1,5 +1,5 @@
 import constants
-from trainer.models.model import Model
+from Trainer.Models.model import Model
 from tensorflow.python.keras.models import Sequential
 from tensorflow.python.keras.layers import Dense, Dropout, Activation, Flatten
 from tensorflow.python.keras.layers import Conv2D

trainer/train_model.py → Trainer/train_model.py

@@ -1,14 +1,15 @@
 import time
 import constants
-from trainer.models.model_gnet_light import ModelGNetLight
-from trainer.models.model import Model
+from Trainer.Models.model_gnet_light import ModelGNetLight
+from Trainer.Models.model_gnet_deep import ModelGNetDeep
+from Trainer.Models.model import Model
 from tensorflow.python.keras.models import Sequential
 from tensorflow.python.keras.layers import Dense, Activation, Flatten
 from tensorflow.python.keras.layers import Conv2D, MaxPooling2D
 
 
 def main():
-    model = ModelGNetLight('new-without-duplicates-aug-batch-16-epoch-10-test')
+    model = ModelGNetDeep('new-without-duplicates-aug-batch-16-epoch-10-test')
     model.create_model()
     model.train_model()
     model.save_model()

constants.py

@@ -5,7 +5,7 @@
 MODEL_DIR = "../number_detection_model"
 
 IMG_SIZE = 28
-USE_GRAYSCALE = True
+USE_GRAY_SCALE = True
 DIMENSION = 1
 
 SIGNAL_TYPES = ["info", "stop"]