plot images

src/consts.py

@@ -2,6 +2,7 @@ import pandas as pd
 FEATURES ='../data/train_features'
 LABELS  = '../data/train_labels'
 JPG_IMAGES = '../images'
+RGB_DIR = "rgb"
 FC_DIR = "fc"
 MASK_DIR = "mask"
 METADATA = pd.read_csv('../data/train_metadata.csv')

src/generators.py

@@ -1,41 +1,41 @@
-train_datagen = ImageDataGenerator(rescale=1 / 255.0,
+from consts import JPG_IMAGES, RGB_DIR, MASK_DIR, FC_DIR, BATCH, IMG_SIZE
-                                    horizontal_flip=True,
+import os
-                                    vertical_flip=True,
+from tensorflow.keras.preprocessing.image import ImageDataGenerator
-                                  validation_split=0.2)
 
-# mask_datagen =  ImageDataGenerator(rescale=1/255,
+def create_generators(mode='train'):
-#                                   validation_split=0.2)
+  '''
+  mode can be train or validation.
+  '''
+  if(mode=='train'):
+    subset = 'training'
+  else:
+    subset = 'validation'
 
-#training data
+  train_datagen = ImageDataGenerator(rescale=1 / 255.0,
-#rgb images
+                                     horizontal_flip=True,
-rgb_dir = '../input/ai4earth-mask-the-clouds/RGB_images'
+                                     vertical_flip=True,
-rgb_gen = train_datagen.flow_from_directory(directory=rgb_dir,
+                                     validation_split=0.2)
-                                           target_size= img_size,
+
+  rgb_gen = train_datagen.flow_from_directory(directory=os.path.join(JPG_IMAGES, RGB_DIR),
+                                           target_size= IMG_SIZE,
                                            batch_size=BATCH,
                                            class_mode=None,
                                            classes=None,
                                            shuffle=False,
-                                            seed=seed,
+                                           subset=subset)
-                                           subset='training')
 
-#false color (nir,green,blue)
+  mask_gen = train_datagen.flow_from_directory(
-fc_dir = '../input/ai4earth-mask-the-clouds/False_color'
+                                           directory=os.path.join(JPG_IMAGES, MASK_DIR ),
-fc_gen = train_datagen.flow_from_directory(
+                                           target_size= IMG_SIZE,
-                                          directory=fc_dir,
-                                           target_size= img_size,
                                            batch_size=BATCH,
                                            class_mode=None,
                                            classes=None,
-                                           shuffle=False,seed=seed,
+                                           shuffle=False,
-                                            subset='training')
+                                           subset=subset)
+
+    
+  # train_genenerator = zip(rgb_gen,mask_gen)
+  # for (imgs, mask) in train_genenerator:
+  #     yield (imgs, mask)
+  return rgb_gen, mask_gen
         
-#training labels
-mask_dir ='../input/ai4earth-mask-the-clouds/Masks'
-mask_gen = train_datagen.flow_from_directory(
-                                          directory=mask_dir,
-                                           target_size= img_size,
-                                           batch_size=BATCH,
-                                           class_mode=None,
-                                           classes=None,
-                                           shuffle=False,seed=seed,
-                                           subset='training')

src/helpers.py

@@ -69,5 +69,6 @@ def convert_tif_to_jpg(features, labels,
     #konwersja maski
     mask= rio.open(labels).read().reshape(512,512,1)
     mask *= 255
+    
     cv2.imwrite(filename= mask_path + f'/{file_name}.jpeg',  
                 img=mask.astype(np.uint8))

src/setup.py

@@ -1,33 +1,28 @@
 import os
 import numpy as np
 import pandas as pd
-import random
 import matplotlib.pyplot as plt
 import time
-
+import sys
-#progess bar 
+import random
 from tqdm import tqdm
 
 import cv2 
-import warnings 
-#deep learning
 
-from consts import SEED, JPG_IMAGES, FC_DIR, FEATURES, MASK_DIR, LABELS
+from consts import SEED, JPG_IMAGES, FC_DIR, FEATURES, MASK_DIR, LABELS, RGB_DIR
 from helpers import  create_folder, convert_tif_to_jpg, progress_bar
-
+from utils import plot_image_grid
-warnings.filterwarnings('ignore')
-
-def transform_photo(tif_dir):
-    dp = create_folder(tif_dir, JPG_IMAGES)
-    fc = create_folder(FC_DIR, dp)
-    mk = create_folder(MASK_DIR, dp)
-    convert_tif_to_jpg(os.path.join(FEATURES, tif_dir), os.path.join(LABELS, tif_dir + ".tif"),  dp, fc, mk)
-    
    
 if __name__ == "__main__":
     dirs = os.listdir(FEATURES)
-    progress_bar(0, len(dirs), prefix = 'Converting TIF to JPG:', suffix = 'Complete', length = 50)
+    dp = create_folder(RGB_DIR, JPG_IMAGES)
-    for i, d in enumerate(dirs):
+    fc = create_folder(FC_DIR, JPG_IMAGES)
-        progress_bar(i, len(dirs), prefix = 'Converting TIF to JPG:', suffix = 'Complete', length = 50)
+    mk = create_folder(MASK_DIR, JPG_IMAGES)
-        if d != ".DS_Store":
+    if(len(sys.argv) <= 1):
-            transform_photo(d)
+        progress_bar(0, len(dirs), prefix = 'Converting TIF to JPG:', suffix = 'Complete', length = 50)
+        for i, d in enumerate(dirs):
+            progress_bar(i, len(dirs), prefix = 'Converting TIF to JPG:', suffix = 'Complete', length = 50)
+            convert_tif_to_jpg(os.path.join(FEATURES, d), os.path.join(LABELS, d + ".tif"),  dp, fc, mk)
+    elif(sys.argv[1] == '--show'):
+        img_names = [random.choice(os.listdir(dp)) for _ in range(3)]
+        plot_image_grid(img_names)

src/utils.py

@@ -1,47 +1,26 @@
-def plot_image_grid(image_list,
+import matplotlib.pyplot as plt
-                    label_list,
+import cv2
-                    sample_images=False,
+import os
-                    num_images=6,
+from consts import JPG_IMAGES, RGB_DIR, MASK_DIR
-                    pre_title='class',
-                    num_img_per_row=3,
-                    cmap=None,
-                    img_h_w=3):
-    '''viz images from a list of images and labels
-    INPUTS:
-    image_list: a list of images to be plotted,
-    label_list: a list of correspomding image labels'''
-    
 
+def plot_image_grid(image_names):
 
     #number of img rows
-    n_row= num_images//num_img_per_row
+    n_row= len(image_names)
 
-    plt.subplots(n_row,num_img_per_row,figsize=(img_h_w*num_img_per_row,img_h_w*n_row))
+    plt.subplots(n_row, 2, figsize=(6, 3*n_row))
 
-    if sample_images:
+    for i, img in enumerate(image_names):
-    #select_random images 
+        r_img = cv2.imread(os.path.join(JPG_IMAGES, RGB_DIR, img))
-        sampled_ids = random.choices(np.arange(0,len(image_list)),k=num_images)
+        plt.subplot(n_row, 2, i*2+1)
-
+        plt.title(f'RGB - {img}')
-        for i,idx in enumerate(sampled_ids):
+        plt.axis('off')
-
+        plt.imshow(r_img, cmap=None)
-            img = image_list[idx]
+        m_img = cv2.imread(os.path.join(JPG_IMAGES, MASK_DIR, img))
-            label = label_list[i]
+        plt.subplot(n_row, 2, i*2+2)
-            plt.subplot(n_row,num_img_per_row,i+1)
+        plt.title(f'MASK - {img}')
-            plt.title(f'{pre_title} - {label}')
+        plt.axis('off')
-            plt.axis('off')
+        plt.imshow(m_img, cmap=None)
-            plt.imshow(img,cmap=cmap)
-    else:
-        for i,img in enumerate(image_list):
-
-            label = label_list[i]
-            plt.subplot(n_row,num_img_per_row,i+1)
-            plt.title(f'{pre_title} - {label}')
-            plt.axis('off')
-            plt.imshow(img,cmap=cmap)
-
-            # break the loop 
-            if i==num_images-1 :
-                  break 
 
     #show
     plt.tight_layout()