add modulization

.gitignore

@@ -103,6 +103,8 @@ venv.bak/
 # mypy
 .mypy_cache/
 
+# vscode
+.vscode/
 
 *DS_Store*
 tf_log/

configs/road_signs.yml

@@ -10,7 +10,7 @@ project_name : 'road_signs/'
 freeze_backbone : True
 
 #paths
-dataset_path : '/home/rauf/plates_competition/dataset/road_signs/road_signs_separated/'
+dataset_path : '/home/rauf/dataset/road_signs/road_signs_separated/'
 tensorboard_log_path : 'tf_log/'
 weights_save_path : 'weights/'
 plots_path : 'plots/'

model.py

@@ -7,15 +7,13 @@
 import random
 from keras.models import Model, load_model
 from keras import optimizers
-from keras.utils import plot_model
 from keras.layers import Dense, Input, Lambda, Dropout, Flatten
 from keras.layers import Conv2D, MaxPool2D, BatchNormalization, concatenate
 from classification_models import Classifiers
-import utils
 import pickle
-import losses_and_accuracies as lac
-from utils import parse_net_params
-from backbones import get_backbone
+from . import losses_and_accuracies as lac
+from .utils import parse_net_params, load_encodings
+from .backbones import get_backbone
 import matplotlib.pyplot as plt
 
 
@@ -48,11 +46,13 @@ def __init__(self,  cfg_file):
         self.l_model = []
 
         self.encodings_path = params['encodings_path']
+        self.configs_path = params['configs_path']
         self.plots_path = params['plots_path']
         self.tensorboard_log_path = params['tensorboard_log_path']
         self.weights_save_path = params['weights_save_path']
 
         os.makedirs(self.encodings_path, exist_ok=True)
+        os.makedirs(self.configs_path, exist_ok=True)
         os.makedirs(self.plots_path, exist_ok=True)
         os.makedirs(self.tensorboard_log_path, exist_ok=True)
         os.makedirs(self.weights_save_path, exist_ok=True)
@@ -105,7 +105,6 @@ def _create_model_siamese(self):
         self.model = Model(
             inputs=[input_image_1, input_image_2], outputs=prediction)
 
-        plot_model(self.model, to_file='{}model.png'.format(self.plots_path))
         print('Base model summary')
         self.base_model.summary()
 
@@ -131,7 +130,6 @@ def _create_model_triplet(self):
         self.model  = Model(inputs=[input_image_a,input_image_p, input_image_n], 
                             outputs=merged_vector)
 
-        plot_model(self.model, to_file='{}model.png'.format(self.plots_path))
         print('Base model summary')
         self.base_model.summary()
 
@@ -215,7 +213,7 @@ def generate_encodings(self, save_file_name='encodings.pkl', max_num_samples_of_
         f.close()
 
     def load_encodings(self, path_to_encodings):
-        self.encoded_training_data = utils.load_encodings(path_to_encodings)
+        self.encoded_training_data = load_encodings(path_to_encodings)
 
     def load_model(self,file_path):
         self.model = load_model(file_path, 
@@ -224,16 +222,20 @@ def load_model(self,file_path):
                                                  'triplet_loss': lac.triplet_loss})
         self.base_model = Model(inputs=[self.model.layers[3].get_input_at(0)], 
                                 outputs=[self.model.layers[3].layers[-1].output])
+        self.base_model._make_predict_function()
 
     def calculate_distances(self, encoding):
         training_encodings = self.encoded_training_data['encodings']
         return np.sqrt(
             np.sum((training_encodings - np.array(encoding))**2, axis=1))
 
-    def predict(self, image_path):
-        img = cv2.imread(image_path)
+    def predict(self, image):
+        if type(image) is str:
+            img = cv2.imread(image)
+        else:
+            img = image
         img = cv2.resize(img, (self.input_shape[0], self.input_shape[1]))
-
+        print(img.shape)
         encoding = self.base_model.predict(np.expand_dims(img, axis=0))
         distances = self.calculate_distances(encoding)
         max_element = np.argmin(distances)

requirements.txt

@@ -0,0 +1,6 @@
+image-classifiers
+keras
+tensorflow-gpu
+matplotlib
+albumentations
+pydot

test.py

@@ -1,15 +1,6 @@
-import keras.backend as K
 from model import SiameseNet
-from data_loader import SiameseImageLoader
-import matplotlib.pyplot as plt
-from keras import optimizers
-from keras.models import Model, load_model
-import albumentations as A
-import cv2
-import time
 
-
-model = SiameseNet('configs/road_signs.yml')
+model = SiameseNet('SiameseNet/configs/road_signs.yml')
 model.load_model('{}best_model_4.h5'.format(model.weights_save_path))
 model.load_encodings('{}encodings.pkl'.format(model.encodings_path))
 

utils.py

@@ -5,8 +5,8 @@
 from matplotlib import pyplot as plt
 import yaml
 from keras import optimizers
-from augmentations import get_aug
-from data_loader import SiameseImageLoader
+from .augmentations import get_aug
+from .data_loader import SiameseImageLoader
 
 
 def load_encodings(path_to_encodings):
@@ -96,13 +96,20 @@ def parse_net_params(filename='configs/road_signs.yml'):
         augmentations = None
 
     params = {k: v for k, v in cfg.items() if k not in ['optimizer']}
-    params['encodings_path'] = os.path.join(cfg['encodings_path'],
+
+    current_directory = os.path.dirname(os.path.abspath(__file__))
+    print(current_directory)
+    params['encodings_path'] = os.path.join(current_directory,
+                                            cfg['encodings_path'],
                                             cfg['project_name'])
-    params['plots_path'] = os.path.join(cfg['plots_path'],
+    params['plots_path'] = os.path.join(current_directory,
+                                        cfg['plots_path'],
                                         cfg['project_name'])
-    params['tensorboard_log_path'] = os.path.join(cfg['tensorboard_log_path'],
+    params['tensorboard_log_path'] = os.path.join(current_directory,
+                                                  cfg['tensorboard_log_path'],
                                                   cfg['project_name'])
-    params['weights_save_path'] = os.path.join(cfg['weights_save_path'],
+    params['weights_save_path'] = os.path.join(current_directory,
+                                               cfg['weights_save_path'],
                                                cfg['project_name'])
 
     if 'dataset_path' in cfg: