added data_download script

data_download.py

@@ -0,0 +1,14 @@
+import os
+import zipfile
+from google.cloud import storage
+
+key_file_path = "key_service_account.json"
+os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = key_file_path
+
+storage_client = storage.Client()
+bucket = storage_client.bucket("smoke_detection")
+blob = bucket.blob("smoke_artifacts.zip")
+blob.download_to_filename("smoke_artifacts.zip")
+unzip_path = ""
+with zipfile.ZipFile("smoke_artifacts.zip", 'r') as zip_ref:
+    zip_ref.extractall(unzip_path)

main.py

@@ -46,7 +46,6 @@
     model_evalution = EvaluationPipeline()
     model_evalution.main()
     logger.info(f">>>>>> stage {STAGE_NAME} completed <<<<<<\n\nx==========x")
-
 except Exception as e:
     logger.exception(e)
     raise e

params.yaml

@@ -2,7 +2,7 @@ IMAGE_SIZE: [224, 224]
 BATCH_SIZE: 32
 LEARNING_RATE: 0.001
 batch_size : 32   # set batch size for training
-epochs : 5   # number of all epochs in training
+epochs : 10   # number of all epochs in training
 patience : 1   #number of epochs to wait to adjust lr if monitored value does not improve
 stop_patience : 3   # number of epochs to wait before stopping training if monitored value does not improve
 threshold : 0.9   # if train accuracy is < threshold adjust monitor accuracy, else monitor validation loss

scores.json

@@ -1,4 +1,4 @@
 {
-    "loss": 0.12056513875722885,
-    "accuracy": 0.9553571343421936
+    "loss": 0.13449855148792267,
+    "accuracy": 0.96875
 }

src/cnnClassifier/components/training.py

@@ -125,7 +125,7 @@ def train(self, callback_list: list):
         self.model.compile(loss=tf.keras.losses.SparseCategoricalCrossentropy(),
                            optimizer=tf.keras.optimizers.Adam(learning_rate=self.config.learning_rate),
                            metrics=['accuracy'])
-        fine_tune_epochs = self.config.epochs + 5
+        fine_tune_epochs = self.config.epochs + 10
         self.history = self.model.fit(self.train_gen, epochs=fine_tune_epochs,
                                       steps_per_epoch=len(self.train_gen),
                                       initial_epoch=self.history.epoch[-1],