import os
import time
import shutil
import pathlib
import itertools
from PIL import Image
import random
# import data handling tools
import cv2
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix, classification_report
# import Deep learning Libraries
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import Adam, Adamax
from tensorflow.keras.metrics import categorical_crossentropy
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.preprocessing.image import load_img,
img_to_array
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten,
Dense, Activation, Dropout, BatchNormalization
from tensorflow.keras import regularizers
# Ignore Warnings
import warnings
warnings.filterwarnings("ignore")
print ('modules loaded')
modules loaded
# Define the directories
train_dir = '/kaggle/input/wildfire-prediction-dataset/train' #
Replace with your train directory
valid_dir = '/kaggle/input/wildfire-prediction-dataset/valid' #
Replace with your validation directory
test_dir = '/kaggle/input/wildfire-prediction-dataset/test' #
Replace with your test directory
# Function to display one random sample from each class
def display_random_sample(dataset_dir):
classes = os.listdir(dataset_dir) # List of class directories
plt.figure(figsize=(15, 10)) # Adjust the figure size if needed
for i, class_name in enumerate(classes):
class_dir = os.path.join(dataset_dir, class_name)
image_files = os.listdir(class_dir) # Get all images in the
class folder
random_image = random.choice(image_files) # Select a random
image
img_path = os.path.join(class_dir, random_image)
# Load the image with its original size
img = load_img(img_path)
# Display the image
plt.subplot(1, len(classes), i + 1) # Display images in a
single row
plt.imshow(img)
plt.axis('off') # Turn off axis for better visualization
plt.title(f"{class_name} ({img.size[0]}x{img.size[1]})") #
Show class name and image size
plt.show()
# Display one random sample from train, validation, and test sets
print("Training set samples:")
display_random_sample(train_dir)
print("Validation set samples:")
display_random_sample(valid_dir)
print("Test set samples:")
display_random_sample(test_dir)
Training set samples:
Validation set samples:
Test set samples:
dir = '/kaggle/input/wildfire-prediction-dataset/train'
x_train = []
y_train = []
for direct in os.listdir(dir):
print("Loading dataset training {}".format(direct))
for filename in os.listdir(os.path.join(dir,direct)):
img_path = os.path.join(dir,direct,filename)
img = cv2.imread(img_path)
img = cv2.resize(img, (32,32))
img = np.array(img)
img = img/255
x_train.append(img)
y_train.append(direct)
Loading dataset training wildfire
Loading dataset training nowildfire
Premature end of JPEG file
dir_val = '/kaggle/input/wildfire-prediction-dataset/valid'
x_val=[]
y_val=[]
for direct in os.listdir(dir_val):
print("Loading dataset validation {}".format(direct))
for filename in os.listdir(os.path.join(dir_val,direct)):
img_path = os.path.join(dir_val,direct,filename)
image = cv2.imread(img_path)
image = cv2.resize(image,(32,32))
image = np.array(image)
image = image/255
x_val.append(image)
y_val.append(direct)
Loading dataset validation wildfire
Loading dataset validation nowildfire
dir_test = '/kaggle/input/wildfire-prediction-dataset/test'
x_test=[]
y_test=[]
for direct in os.listdir(dir_test):
print("Loading dataset test {}".format(direct))
for filename in os.listdir(os.path.join(dir_test,direct)):
img_path = os.path.join(dir_test,direct,filename)
image = cv2.imread(img_path)
image = cv2.resize(image,(32,32))
image = np.array(image)
image = image/255
x_test.append(image)
y_test.append(direct)
Loading dataset test wildfire
Premature end of JPEG file
Loading dataset test nowildfire
x_train = np.array(x_train)
x_val = np.array(x_val)
x_test = np.array(x_test)
y_train[30000]
'nowildfire'
# Replace "wildfire" with 1 and "nowildfire" with 0
y_train = [1 if label == 'wildfire' else 0 for label in y_train]
y_val = [1 if label == 'wildfire' else 0 for label in y_val]
y_test = [1 if label == 'wildfire' else 0 for label in y_test]
y_train = np.array(y_train)
y_val = np.array(y_val)
y_test = np.array(y_test)
len(x_train[4][4])
32
# Now check the shape of your datasets
print("x_train shape:", x_train.shape) # Should be (num_samples,
height, width, num_channels)
print("x_valid shape:", x_val.shape)
print("x_test shape:", x_test.shape)
print("y_train shape:", y_train.shape)
print("y_valid shape:", y_val.shape)
print("y_test shape:", y_test.shape)
x_train shape: (30250, 32, 32, 3)
x_valid shape: (6300, 32, 32, 3)
x_test shape: (6300, 32, 32, 3)
y_train shape: (30250,)
y_valid shape: (6300,)
y_test shape: (6300,)
# Step 1: Build the CNN model
model = Sequential([
Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)),
Conv2D(64, (3, 3), activation='relu', padding='same'),
Conv2D(64, (3, 3), activation='relu', padding='same'),
BatchNormalization(),
MaxPooling2D((2, 2)),
Conv2D(128, (3, 3), activation='relu', padding='same'),
Conv2D(128, (3, 3), activation='relu', padding='same'),
BatchNormalization(),
MaxPooling2D((2, 2)),
Conv2D(256, (3, 3), activation='relu', padding='same'),
Conv2D(256, (3, 3), activation='relu', padding='same'),
BatchNormalization(),
MaxPooling2D((2, 2)),
Conv2D(128, (3, 3), activation='relu', padding='same'),
Conv2D(128, (3, 3), activation='relu', padding='same'),
Flatten(),
Dense(128, activation='relu'),
Dense(128, activation='relu'),
BatchNormalization(),
Dropout(0.5),
Dense(64, activation='relu'),
Dense(32, activation='relu'),
BatchNormalization(),
Dense(1, activation='sigmoid') # Output layer with number of
classes
])
# Step 2: Compile the model
model.compile(optimizer='adam',
loss='binary_crossentropy', # Loss function for multi-
class classification
metrics=['accuracy'])
model.summary()
Model: "sequential_20"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━
━━━━━┓
┃ Layer (type) ┃ Output Shape ┃
Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━
━━━━━┩
│ conv2d_133 (Conv2D) │ (None, 30, 30, 32) │
896 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_134 (Conv2D) │ (None, 30, 30, 64) │
18,496 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_135 (Conv2D) │ (None, 30, 30, 64) │
36,928 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ batch_normalization_59 │ (None, 30, 30, 64) │
256 │
│ (BatchNormalization) │ │
│
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ max_pooling2d_42 (MaxPooling2D) │ (None, 15, 15, 64) │
0 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_136 (Conv2D) │ (None, 15, 15, 128) │
73,856 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_137 (Conv2D) │ (None, 15, 15, 128) │
147,584 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ batch_normalization_60 │ (None, 15, 15, 128) │
512 │
│ (BatchNormalization) │ │
│
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ max_pooling2d_43 (MaxPooling2D) │ (None, 7, 7, 128) │
0 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_138 (Conv2D) │ (None, 7, 7, 256) │
295,168 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_139 (Conv2D) │ (None, 7, 7, 256) │
590,080 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ batch_normalization_61 │ (None, 7, 7, 256) │
1,024 │
│ (BatchNormalization) │ │
│
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ max_pooling2d_44 (MaxPooling2D) │ (None, 3, 3, 256) │
0 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_140 (Conv2D) │ (None, 3, 3, 128) │
295,040 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ conv2d_141 (Conv2D) │ (None, 3, 3, 128) │
147,584 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ flatten_20 (Flatten) │ (None, 1152) │
0 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ dense_78 (Dense) │ (None, 128) │
147,584 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ dense_79 (Dense) │ (None, 128) │
16,512 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ batch_normalization_62 │ (None, 128) │
512 │
│ (BatchNormalization) │ │
│
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ dropout (Dropout) │ (None, 128) │
0 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ dense_80 (Dense) │ (None, 64) │
8,256 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ dense_81 (Dense) │ (None, 32) │
2,080 │
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ batch_normalization_63 │ (None, 32) │
128 │
│ (BatchNormalization) │ │
│
├─────────────────────────────────┼────────────────────────┼──────────
─────┤
│ dense_82 (Dense) │ (None, 1) │
33 │
└─────────────────────────────────┴────────────────────────┴──────────
─────┘
Total params: 1,782,529 (6.80 MB)
Trainable params: 1,781,313 (6.80 MB)
Non-trainable params: 1,216 (4.75 KB)
from tensorflow.keras.callbacks import EarlyStopping
# Step 3: Train the model
history = model.fit(
x_train, y_train,
validation_data=(x_val, y_val),
epochs=20,
batch_size=64,
callbacks=[EarlyStopping(monitor='val_loss', patience=4)]
)
Epoch 1/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 20s 23ms/step - accuracy: 0.8795 - loss:
0.2910 - val_accuracy: 0.6083 - val_loss: 1.1794
Epoch 2/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 13ms/step - accuracy: 0.9252 - loss:
0.1913 - val_accuracy: 0.9024 - val_loss: 0.2299
Epoch 3/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 12ms/step - accuracy: 0.9321 - loss:
0.1734 - val_accuracy: 0.5735 - val_loss: 0.9520
Epoch 4/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 12ms/step - accuracy: 0.9377 - loss:
0.1574 - val_accuracy: 0.8211 - val_loss: 0.3225
Epoch 5/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 12ms/step - accuracy: 0.9442 - loss:
0.1464 - val_accuracy: 0.9543 - val_loss: 0.1255
Epoch 6/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 12ms/step - accuracy: 0.9477 - loss:
0.1359 - val_accuracy: 0.9540 - val_loss: 0.1189
Epoch 7/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 13ms/step - accuracy: 0.9456 - loss:
0.1413 - val_accuracy: 0.9365 - val_loss: 0.1692
Epoch 8/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 12ms/step - accuracy: 0.9511 - loss:
0.1278 - val_accuracy: 0.9525 - val_loss: 0.1287
Epoch 9/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 13ms/step - accuracy: 0.9564 - loss:
0.1186 - val_accuracy: 0.8160 - val_loss: 0.3986
Epoch 10/20
473/473 ━━━━━━━━━━━━━━━━━━━━ 6s 12ms/step - accuracy: 0.9560 - loss:
0.1141 - val_accuracy: 0.9478 - val_loss: 0.1521
# Evaluate the model on test data
test_loss, test_accuracy = model.evaluate(x_test, y_test)
print(f'Test Accuracy: {test_accuracy:.4f}')
197/197 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9470 - loss:
0.1800
Test Accuracy: 0.9541
# Plot training & validation accuracy values
plt.plot(history.history['accuracy'], label='Train Accuracy')
plt.plot(history.history['val_accuracy'], label='Test Accuracy')
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epoch')
plt.legend(loc='upper left')
plt.show()
# Plot training & validation loss values
plt.plot(history.history['loss'], label='Train Loss')
plt.plot(history.history['val_loss'], label='Test Loss')
plt.title('Model Loss')
plt.ylabel('Loss')
plt.xlabel('Epoch')
plt.legend(loc='upper left')
plt.show()
# Predict on the datasets
y_train_pred = model.predict(x_train)
y_val_pred = model.predict(x_val)
y_test_pred = model.predict(x_test)
# Convert probabilities to binary classes if using a classification
model
y_train_pred_classes = (y_train_pred > 0.5).astype("int32")
y_val_pred_classes = (y_val_pred > 0.5).astype("int32")
y_test_pred_classes = (y_test_pred > 0.5).astype("int32")
946/946 ━━━━━━━━━━━━━━━━━━━━ 3s 3ms/step
197/197 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step
197/197 ━━━━━━━━━━━━━━━━━━━━ 0s 2ms/step
print("Confusion matrix for train: \n")
cm = confusion_matrix(y_train, y_train_pred_classes)
sns.heatmap(cm, annot=True)
plt.title(f"Confusion Matrix")
plt.xlabel("Predicted")
plt.ylabel("True")
plt.show()
Confusion matrix for train:
from sklearn.metrics import classification_report
print(f"classification report for train : \
n{classification_report(y_train, y_train_pred_classes)}")
classification report for train :
precision recall f1-score support
0 0.95 0.96 0.96 14500
1 0.97 0.95 0.96 15750
accuracy 0.96 30250
macro avg 0.96 0.96 0.96 30250
weighted avg 0.96 0.96 0.96 30250
print("Confusion matrix for valid: \n")
cm = confusion_matrix(y_val, y_val_pred_classes)
sns.heatmap(cm, annot=True)
plt.title(f"Confusion Matrix")
plt.xlabel("Predicted")
plt.ylabel("True")
plt.show()
Confusion matrix for valid:
print(f"classification report for valid : \
n{classification_report(y_val, y_val_pred_classes)}")
classification report for valid :
precision recall f1-score support
0 0.93 0.96 0.94 2820
1 0.97 0.94 0.95 3480
accuracy 0.95 6300
macro avg 0.95 0.95 0.95 6300
weighted avg 0.95 0.95 0.95 6300
print("Confusion matrix for test: \n")
cm = confusion_matrix(y_test, y_test_pred_classes)
sns.heatmap(cm, annot=True)
plt.title(f"Confusion Matrix")
plt.xlabel("Predicted")
plt.ylabel("True")
plt.show()
Confusion matrix for test:
print(f"classification report for test : \
n{classification_report(y_test, y_test_pred_classes)}")
classification report for test :
precision recall f1-score support
0 0.93 0.97 0.95 2820
1 0.97 0.94 0.96 3480
accuracy 0.95 6300
macro avg 0.95 0.96 0.95 6300
weighted avg 0.95 0.95 0.95 6300
