From 77fac4295d07e979b3dc5ddeaba93cbfecd09154 Mon Sep 17 00:00:00 2001
From: Abdulrahman <abdulrahman2.ali@live.uwe.ac.uk>
Date: Mon, 13 May 2024 15:23:02 +0300
Subject: [PATCH] chore: Organize and improve car prediction CNN model code
---
.../FINAL_car_prediction_cnn_BRAND.ipynb | 67 ++++++++++++++-----
1 file changed, 52 insertions(+), 15 deletions(-)
diff --git a/mlmodel/Final/FINAL_car_prediction_cnn_BRAND.ipynb b/mlmodel/Final/FINAL_car_prediction_cnn_BRAND.ipynb
index 4efae18..b407119 100644
--- a/mlmodel/Final/FINAL_car_prediction_cnn_BRAND.ipynb
+++ b/mlmodel/Final/FINAL_car_prediction_cnn_BRAND.ipynb
@@ -79,23 +79,12 @@
},
{
"cell_type": "code",
- "execution_count": 33,
+ "execution_count": null,
"id": "f9ff0b13-86db-4de5-b2a8-599a66ccc27b",
"metadata": {
"tags": []
},
- "outputs": [
- {
- "data": {
- "text/plain": [
- " ...])"
- ]
- },
- "execution_count": 33,
- "metadata": {},
- "output_type": "execute_result"
- }
- ],
+ "outputs": [],
"source": [
"import os\n",
"\n",
@@ -123,10 +112,11 @@
"\n",
"# Create a DataFrame containing the paths to resized images and their labels\n",
"resized_data = pd.DataFrame({\n",
- " 'brand': data['brand']\n","
+ " 'src': [os.path.join(resized_dir, os.path.basename(src)) for src in data['src']],\n",
+ " 'brand': data['brand']\n",
"})\n",
"\n",
- "X, y\n"
+ "# X, y\n"
]
},
{
@@ -324,6 +314,30 @@
"predicted_class_index = np.argmax(predictions)\n",
"\n",
"class_to_brand = {\n",
+ " 0: 'Acura',\n",
+ " 1: 'Audi',\n",
+ " 2: 'Volkswagen',\n",
+ " 3: 'Toyota',\n",
+ " 4: 'Subaru',\n",
+ " 5: 'Porsche',\n",
+ " 6: 'Nissan',\n",
+ " 7: 'MINI',\n",
+ " 8: 'Mercedes-Benz',\n",
+ " 9: 'Mazda',\n",
+ " 10: 'Lincoln',\n",
+ " 11: 'Lexus',\n",
+ " 12: 'Kia',\n",
+ " 13: 'Jeep',\n",
+ " 14: 'Jaguar',\n",
+ " 15: 'Hyundai',\n",
+ " 16: 'Honda',\n",
+ " 17: 'GMC',\n",
+ " 18: 'Ford',\n",
+ " 19: 'Dodge',\n",
+ " 20: 'Chevrolet',\n",
+ " 21: 'Cadillac',\n",
+ " 22: 'BMW',\n",
+ " 23: 'Volvo',\n",
"}\n",
"\n",
"# Get the predicted brand name\n",
@@ -584,6 +598,13 @@
"from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\n",
"\n",
"datagen = ImageDataGenerator(\n",
+ " rotation_range=15,\n",
+ " width_shift_range=0.1,\n",
+ " height_shift_range=0.1,\n",
+ " shear_range=0.1,\n",
+ " zoom_range=0.1,\n",
+ " horizontal_flip=True,\n",
+ " vertical_flip=True,\n",
" fill_mode='nearest'\n",
")\n",
"\n",
@@ -591,14 +612,30 @@
"\n",
"input_shape = X_train.shape[1:]\n",
"model = models.Sequential([\n",
+ " layers.Conv2D(32, (3, 3), activation='relu', input_shape=input_shape),\n",
+ " layers.MaxPooling2D((2, 2)),\n",
+ " layers.Conv2D(64, (3, 3), activation='relu'),\n",
+ " layers.MaxPooling2D((2, 2)),\n",
+ " layers.Conv2D(128, (3, 3), activation='relu'),\n",
+ " layers.MaxPooling2D((2, 2)),\n",
+ " layers.Flatten(),\n",
+ " layers.Dropout(0.5),\n",
+ " layers.Dense(512, activation='relu'),\n",
" layers.Dense(24, activation='softmax')\n",
"])\n",
+ "model.compile(optimizer='adam',\n",
+ " loss='categorical_crossentropy',\n",
" metrics=['accuracy'])\n",
"\n",
"checkpoint_callback = ModelCheckpoint('model2.keras', monitor='val_loss', save_best_only=True)\n",
"early_stopping = EarlyStopping(monitor='val_loss', patience=10)\n",
"\n",
"history = model.fit(\n",
+ " datagen.flow(X_train, y_train, batch_size=batch_size),\n",
+ " epochs=100,\n",
+ " validation_data=(X_test, y_test),\n",
+ " verbose=1,\n",
+ " callbacks=[checkpoint_callback, early_stopping],\n",
")"
]
},
--
GitLab