From 527daa62819fdb7d19759126d198a7db51811a9f Mon Sep 17 00:00:00 2001
From: mr2-alkhateeb <marwan2.alkhateeb@live.uwe.ac.uk>
Date: Sun, 12 May 2024 18:11:45 +0000
Subject: [PATCH] Added comments and organised
---
mlmodel/{ => Version1}/evaluation.py | 7 ++++++
mlmodel/{ => Version1}/main.py | 6 +++++
mlmodel/{ => Version1}/preprocessing.py | 2 ++
mlmodel/{ => Version1}/training.py | 0
mlmodel/Version2/categorization.py | 31 ------------------------
mlmodel/Version2/cnn_training.py | 2 ++
mlmodel/Version2/evaluation.py | 2 ++
mlmodel/Version2/preprocessing.py | 12 +--------
mlmodel/Version3/eval.py | 6 ++---
mlmodel/Version3/preprocess.py | 1 +
mlmodel/{ => Version3}/resnet_model5.h5 | Bin
11 files changed, 24 insertions(+), 45 deletions(-)
rename mlmodel/{ => Version1}/evaluation.py (85%)
rename mlmodel/{ => Version1}/main.py (77%)
rename mlmodel/{ => Version1}/preprocessing.py (96%)
rename mlmodel/{ => Version1}/training.py (100%)
rename mlmodel/{ => Version3}/resnet_model5.h5 (100%)
diff --git a/mlmodel/evaluation.py b/mlmodel/Version1/evaluation.py
similarity index 85%
rename from mlmodel/evaluation.py
rename to mlmodel/Version1/evaluation.py
index 8d309fc..55ef648 100644
--- a/mlmodel/evaluation.py
+++ b/mlmodel/Version1/evaluation.py
@@ -2,23 +2,30 @@ from joblib import load
from sklearn.metrics import accuracy_score, classification_report
import numpy as np
+# Define function to evaluate model
def evaluate_model(model_path, preprocessed_data_file):
+ # Load model
model = load(model_path)
+ # Load preprocessed data
preprocessed_data = np.load(preprocessed_data_file)
test_images = preprocessed_data['test_images']
test_labels = preprocessed_data['test_labels']
+ # Reshape images to 2D
if len(test_images.shape) > 2:
num_samples, height, width, channels = test_images.shape
test_images = test_images.reshape(num_samples, height * width * channels)
+ # Predict
predicted_labels = model.predict(test_images)
+ # Check accuracy
accuracy = accuracy_score(test_labels, predicted_labels)
print(f"Model Accuracy: {accuracy:.2f}")
+ Print Classification Report
report = classification_report(test_labels, predicted_labels)
print("Classification Report:")
print(report)
diff --git a/mlmodel/main.py b/mlmodel/Version1/main.py
similarity index 77%
rename from mlmodel/main.py
rename to mlmodel/Version1/main.py
index bd2e8ce..25e4f0c 100644
--- a/mlmodel/main.py
+++ b/mlmodel/Version1/main.py
@@ -1,13 +1,19 @@
from preprocessing import preprocess_images, split_data, save_preprocessed_data
from training import train_model
+# Set directory for images
dataset_dir = r'Cars Dataset\train'
+
+# Preprocess images
images, labels = preprocess_images(dataset_dir)
+# Split images into training and testing
train_images, test_images, train_labels, test_labels = split_data(images, labels)
+# Save the preprocessed data
output_file = 'preprocessed_data.npz'
save_preprocessed_data(train_images, test_images, train_labels, test_labels, output_file)
+# Train model
model = train_model(train_images, train_labels)
diff --git a/mlmodel/preprocessing.py b/mlmodel/Version1/preprocessing.py
similarity index 96%
rename from mlmodel/preprocessing.py
rename to mlmodel/Version1/preprocessing.py
index ce69547..4e427eb 100644
--- a/mlmodel/preprocessing.py
+++ b/mlmodel/Version1/preprocessing.py
@@ -4,6 +4,7 @@ import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
+# Function to preprocess images
def preprocess_images(dataset_dir, image_size=(224, 224)):
images = []
labels = []
@@ -28,6 +29,7 @@ def preprocess_images(dataset_dir, image_size=(224, 224)):
return images, labels
+# Function to reshape images
def preprocess_images_2d(images):
# Reshape the images into 2D arrays
num_samples, height, width, channels = images.shape
diff --git a/mlmodel/training.py b/mlmodel/Version1/training.py
similarity index 100%
rename from mlmodel/training.py
rename to mlmodel/Version1/training.py
diff --git a/mlmodel/Version2/categorization.py b/mlmodel/Version2/categorization.py
index 6edd44b..deb9711 100644
--- a/mlmodel/Version2/categorization.py
+++ b/mlmodel/Version2/categorization.py
@@ -1,34 +1,3 @@
-# import os
-# import shutil
-
-# # Set the directory where your images are stored
-# source_dir = 'Car Model Training'
-
-# # Set the directory where you want to organize the folders by category
-# target_dir = 'categories'
-
-# # Check if the target directory exists, if not, create it
-# if not os.path.exists(target_dir):
-# os.mkdir(target_dir)
-
-# # Loop through each file in the source directory
-# for filename in os.listdir(source_dir):
-# if filename.endswith((".png", ".jpg", ".jpeg")): # Check for image files
-# # Assuming the category is the first three parts of the filename (Make_Model_Year)
-# category_parts = filename.split('_')[:3]
-# category = '_'.join(category_parts) # This will join them as 'Make_Model_Year'
-
-# # Create a new folder path for the category if it doesn't exist
-# category_path = os.path.join(target_dir, category)
-# if not os.path.exists(category_path):
-# os.mkdir(category_path)
-
-# # Move the file
-# shutil.move(os.path.join(source_dir, filename), os.path.join(category_path, filename))
-
-# print("Images have been organized into category folders.")
-
-
import os
import shutil
diff --git a/mlmodel/Version2/cnn_training.py b/mlmodel/Version2/cnn_training.py
index 9f2d5cf..9c68d16 100644
--- a/mlmodel/Version2/cnn_training.py
+++ b/mlmodel/Version2/cnn_training.py
@@ -4,6 +4,7 @@ from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout
from tensorflow.keras.preprocessing.image import ImageDataGenerator
+# Define cnn model
def create_cnn_model(input_shape, num_classes):
model = Sequential([
Conv2D(32, (3, 3), activation='relu', input_shape=input_shape),
@@ -19,6 +20,7 @@ def create_cnn_model(input_shape, num_classes):
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
return model
+# Define function to train the model
def train_cnn_model(train_images, train_labels):
# Assume train_images and train_labels are already provided and preprocessed
diff --git a/mlmodel/Version2/evaluation.py b/mlmodel/Version2/evaluation.py
index 8f758da..5d724b0 100644
--- a/mlmodel/Version2/evaluation.py
+++ b/mlmodel/Version2/evaluation.py
@@ -2,12 +2,14 @@ from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import numpy as np
+# Define function to evaluate model
def evaluate_model(model_path, preprocessed_data_file):
model = load_model(model_path)
preprocessed_data = np.load(preprocessed_data_file)
test_images = preprocessed_data['test_images'] / 255.0
test_labels = preprocessed_data['test_labels']
+ # Evaluate model using the test images and labels
results = model.evaluate(test_images, test_labels)
print(f"Test Loss, Test Accuracy: {results}")
diff --git a/mlmodel/Version2/preprocessing.py b/mlmodel/Version2/preprocessing.py
index 0eb1ad6..e5eefa5 100644
--- a/mlmodel/Version2/preprocessing.py
+++ b/mlmodel/Version2/preprocessing.py
@@ -4,6 +4,7 @@ import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
+# Define function to preprocess images
def preprocess_images(dataset_dir, image_size=(224, 224)):
images = []
labels = []
@@ -32,17 +33,6 @@ def preprocess_images(dataset_dir, image_size=(224, 224)):
return images, labels
-# def preprocess_images_2d(images):
-# # Reshape the images into 2D arrays
-# num_samples, height, width, channels = images.shape
-# images_2d = images.reshape(num_samples, height * width * channels)
-
-# # Standardize the pixel values
-# scaler = StandardScaler()
-# images_2d_scaled = scaler.fit_transform(images_2d)
-
-# return images_2d_scaled
-
# Split the dataset for training
def split_data(images, labels, test_size=0.2):
return train_test_split(images, labels, test_size=test_size)
diff --git a/mlmodel/Version3/eval.py b/mlmodel/Version3/eval.py
index e547659..e0bebe1 100644
--- a/mlmodel/Version3/eval.py
+++ b/mlmodel/Version3/eval.py
@@ -1,11 +1,11 @@
import tensorflow as tf
-from tensorflow.keras.models import Model
+import numpy as np
import cv2
-
+from tensorflow.keras.models import Model
from tensorflow.keras.preprocessing import image
from tensorflow.keras.preprocessing.image import load_img, img_to_array
-import numpy as np
+# Car brands
categories = ['Acura', 'Alfa Romeo', 'Aston Martin', 'Audi', 'Bentley']
# Load the model
diff --git a/mlmodel/Version3/preprocess.py b/mlmodel/Version3/preprocess.py
index a793d38..3c3d2bf 100644
--- a/mlmodel/Version3/preprocess.py
+++ b/mlmodel/Version3/preprocess.py
@@ -2,6 +2,7 @@ import os
import shutil
from sklearn.model_selection import train_test_split
+# Function to split data
def split_data(source_folder, train_size=0.7, val_size=0.15, test_size=0.15):
# Specify the car brands to include
included_brands = ['Audi', 'Bentley', 'Acura', 'Aston Martin', 'Alfa Romeo']
diff --git a/mlmodel/resnet_model5.h5 b/mlmodel/Version3/resnet_model5.h5
similarity index 100%
rename from mlmodel/resnet_model5.h5
rename to mlmodel/Version3/resnet_model5.h5
--
GitLab