add new generator

configs/bengali.yml

@@ -0,0 +1,56 @@
+MODEL:
+  input_shape :  [128, 128, 3]
+  encodings_len: 256
+  mode : 'triplet'
+  distance_type : 'l1'
+  backbone : 'resnet18'
+  backbone_weights : 'imagenet'
+  freeze_backbone : False
+  embeddings_normalization: True
+
+TRAIN:
+  margin: 0.5
+  batch_size : 8
+  augmentation_type : 'default'
+
+  # optimizer parameters
+  optimizer : 'radam'
+  learning_rate : 0.0001
+  decay_factor : 0.99
+  step_size : 1
+
+  # embeddings learning training parameters
+  n_epochs : 1000
+  n_batches : 200
+  val_batch_size : 8
+  val_steps : 200
+  negatives_selection_mode : 'semihard'
+  mining_n_classes: 5
+  mining_n_samples: 3
+
+  # softmax pretraining parameters
+  softmax_pretraining : True
+  softmax_batch_size : 8
+  softmax_val_steps : 200
+  softmax_steps_per_epoch : 500
+  softmax_epochs : 20
+
+  # plot training history
+  plot_history : True
+
+PATHS:
+  work_dir : 'work_dirs/road_signs_resnet18/'
+  dataset_path : '/home/rauf/datasets/bengali/pngs/train/'
+  csv_file : '/home/rauf/datasets/bengali/train.csv'
+  image_id_column : 'image_id'
+  label_column : 'label'
+  encodings_path : 'encodings/'
+  model_save_name : 'best_model_resnet18.h5'
+  encodings_save_name: 'encodings_resnet18.pkl'
+
+ENCODINGS:
+  # encodings parameters
+  save_encodings : True
+  centers_only: False
+  max_num_samples_of_each_class : 30
+  knn_k : 1

configs/road_signs_resnet18.yml

@@ -1,47 +1,53 @@
-input_shape :  [48, 48, 3]
-encodings_len: 256
-margin: 0.5
-mode : 'triplet'
-distance_type : 'l1'
-backbone : 'resnet18'
-backbone_weights : 'imagenet'
-project_name : 'road_signs/'
-freeze_backbone : False
-augmentation_type : 'default'
-embeddings_normalization: True
+MODEL:
+  encodings_len: 256
+  mode : 'triplet'
+  distance_type : 'l1'
+  backbone : 'resnet18'
+  backbone_weights : 'imagenet'
+  freeze_backbone : False
+  embeddings_normalization: True
 
-# optimizer parameters
-optimizer : 'radam'
-learning_rate : 0.0001
-decay_factor : 0.99
-step_size : 1
+TRAIN:
+  input_shape :  [48, 48, 3]
+  margin: 0.5
+  augmentation_type : 'default'
 
-# embeddings learning training parameters
-n_epochs : 1000
-n_steps_per_epoch : 200
-val_batch_size : 8
-val_steps : 200
-negatives_selection_mode : 'semihard'
-mining_n_classes: 5
-mining_n_samples: 3
+  # optimizer parameters
+  optimizer : 'radam'
+  learning_rate : 0.0001
+  decay_factor : 0.99
+  step_size : 1
 
-# softmax pretraining parameters
-softmax_pretraining : True
-softmax_batch_size : 8
-softmax_val_steps : 200
-softmax_steps_per_epoch : 500
-softmax_epochs : 20
+  # embeddings learning training parameters
+  n_epochs : 1000
+  n_steps_per_epoch : 200
+  val_batch_size : 8
+  val_steps : 200
+  negatives_selection_mode : 'semihard'
+  mining_n_classes: 5
+  mining_n_samples: 3
 
-#paths
-work_dir : 'work_dirs/road_signs_resnet18/'
-dataset_path : '/home/rauf/datasets/road_signs/road_signs_separated/'
-plot_history : True
-encodings_path : 'encodings/'
-model_save_name : 'best_model_resnet18.h5'
-encodings_save_name: 'encodings_resnet18.pkl'
+  # softmax pretraining parameters
+  softmax_pretraining : True
+  softmax_batch_size : 8
+  softmax_val_steps : 200
+  softmax_steps_per_epoch : 500
+  softmax_epochs : 20
 
-# encodings parameters
-save_encodings : True
-centers_only: False
-max_num_samples_of_each_class : 30
-knn_k : 1
+  # plot training history
+  plot_history : True
+
+PATHS:
+  work_dir : 'work_dirs/road_signs_resnet18/'
+  dataset_path : '/home/rauf/datasets/road_signs/road_signs_separated/'
+  csv_file : '/home/rauf/datasets/road_signs/road_signs_separated/'
+  encodings_path : 'encodings/'
+  model_save_name : 'best_model_resnet18.h5'
+  encodings_save_name: 'encodings_resnet18.pkl'
+
+ENCODINGS:
+  # encodings parameters
+  save_encodings : True
+  centers_only: False
+  max_num_samples_of_each_class : 30
+  knn_k : 1

embedding_net/backbones.py

@@ -1,17 +1,17 @@
-from keras.layers import Dense, Input, Lambda, Dropout, Flatten
-from keras.layers import Conv2D, MaxPool2D, BatchNormalization, concatenate
-from keras.models import Model
-from keras.regularizers import l2
-import keras.backend as K
+from tensorflow.keras.layers import Dense, Input, Lambda, Dropout, Flatten
+from tensorflow.keras.layers import Conv2D, MaxPool2D, BatchNormalization, concatenate
+from tensorflow.keras.models import Model
+from tensorflow.keras.regularizers import l2
+import tensorflow.keras.backend as K
 
 
 def get_backbone(input_shape,
                  encodings_len=4096,
-                 backbone_type='simple',
+                 backbone_name='simple',
                  embeddings_normalization=True,
                  backbone_weights='imagenet',
                  freeze_backbone=False):
-    if backbone_type == 'simple':
+    if backbone_name == 'simple':
         input_image = Input(input_shape)
         x = Conv2D(64, (10, 10), activation='relu',
                    kernel_regularizer=l2(2e-4))(input_image)
@@ -34,7 +34,7 @@ def get_backbone(input_shape,
                 x, axis=1), name='l2_norm')(encoded_output)
         base_model = Model(
             inputs=[input_image], outputs=[encoded_output])
-    elif backbone_type == 'simple2':
+    elif backbone_name == 'simple2':
         input_image = Input(input_shape)
         x = Conv2D(32, kernel_size=3, activation='relu',
                    kernel_regularizer=l2(2e-4))(input_image)
@@ -76,11 +76,28 @@ def get_backbone(input_shape,
         base_model = Model(
             inputs=[input_image], outputs=[encoded_output])
     else:
-        from classification_models.keras import Classifiers
-        classifier, preprocess_input = Classifiers.get(backbone_type)
-        backbone_model = classifier(input_shape=input_shape,
-                                    weights=backbone_weights,
-                                    include_top=False)
+        if backbone_name.startswith('efficientnet'):
+            import efficientnet.tfkeras as efn
+            efficientnet_models = {
+                'efficientnet-b0': efn.EfficientNetB0,
+                'efficientnet-b1': efn.EfficientNetB1,
+                'efficientnet-b2': efn.EfficientNetB2,
+                'efficientnet-b3': efn.EfficientNetB3,
+                'efficientnet-b4': efn.EfficientNetB4,
+                'efficientnet-b5': efn.EfficientNetB5,
+                'efficientnet-b6': efn.EfficientNetB6,
+                'efficientnet-b7': efn.EfficientNetB7,
+            }
+            Efficientnet_model = efficientnet_models[backbone_name]
+            backbone_model = Efficientnet_model(input_shape=input_shape, 
+                                            weights=backbone_weights, 
+                                            include_top=False)
+        else:
+            from classification_models.tfkeras import Classifiers
+            classifier, preprocess_input = Classifiers.get(backbone_name)
+            backbone_model = classifier(input_shape=input_shape,
+                                        weights=backbone_weights,
+                                        include_top=False)
 
         if freeze_backbone:
             for layer in backbone_model.layers[:-2]: