Skip to content

2
2023-286
Commit 0c887a48 authored by Fernando W.H.K - IT20116606's avatar Fernando W.H.K - IT20116606 💻
Browse files
Options

Delete facial-expression-recognition-cnn.py

parent 28fb0974
Hide whitespace changes
Inline Side-by-side
Showing with 0 additions and 360 deletions
facial-expression-recognition-cnn.py deleted 100644 → 0
View file @ 28fb0974
#!/usr/bin/env python
# coding: utf-8
# In[1]:
#import necessary libraries
import numpy as np
import os
import matplotlib.pyplot as plt
import tensorflow as tf
import keras
import cv2
from sklearn.metrics import classification_report, confusion_matrix
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout
from keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau
# In[2]:
#load the FER-2013 dataset
data_path = './facial_dataset'
train_dir = os.path.join(data_path, '/train')
test_dir = os.path.join(data_path, '/test')
# In[3]:
img_shape = 48
batch_size = 64
train_data_path = './facial_dataset/train/'
test_data_path = './facial_dataset/test/'
# In[7]:
# Define data augmentation parameters
train_datagen = ImageDataGenerator(
rescale = 1 / 255.,
# Data Augmentation
rotation_range=10,
zoom_range=0.2,
width_shift_range=0.1,
height_shift_range=0.1,
horizontal_flip=True,
fill_mode='nearest',
)
test_datagen = ImageDataGenerator(
rescale = 1 / 255.,
)
train_data = train_datagen.flow_from_directory(
train_data_path,
class_mode="categorical",
target_size=(img_shape,img_shape),
color_mode='rgb',
shuffle=True,
batch_size=batch_size,
subset='training',
)
test_data = test_datagen.flow_from_directory(
test_data_path,
class_mode="categorical",
target_size=(img_shape,img_shape),
color_mode="rgb",
shuffle=False,
batch_size=batch_size,
)
# In[8]:
from keras.layers import BatchNormalization
def Create_CNN_Model():
model = Sequential()
#CNN1
model.add(Conv2D(32, (3,3), activation='relu', input_shape=(img_shape, img_shape, 3)))
model.add(BatchNormalization())
model.add(Conv2D(64,(3,3), activation='relu', padding='same'))
model.add(BatchNormalization())
model.add(MaxPooling2D(pool_size=(2,2), padding='same'))
model.add(Dropout(0.25))
#CNN2
model.add(Conv2D(64, (3,3), activation='relu', ))
model.add(BatchNormalization())
model.add(Conv2D(128,(3,3), activation='relu', padding='same'))
model.add(BatchNormalization())
model.add(MaxPooling2D(pool_size=(2,2), padding='same'))
model.add(Dropout(0.25))
#CNN3
model.add(Conv2D(128, (3,3), activation='relu'))
model.add(BatchNormalization())
model.add(Conv2D(256,(3,3), activation='relu', padding='same'))
model.add(BatchNormalization())
model.add(MaxPooling2D(pool_size=(2,2), padding='same'))
model.add(Dropout(0.25))
#Output
model.add(Flatten())
model.add(Dense(1024, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.25))
model.add(Dense(512, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.25))
model.add(Dense(256, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.25))
model.add(Dense(128, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.25))
model.add(Dense(64, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.25))
model.add(Dense(32, activation='relu'))
model.add(BatchNormalization())
model.add(Dropout(0.25))
model.add(Dense(7,activation='softmax'))
return model
# In[9]:
CNN_Model = Create_CNN_Model()
CNN_Model.summary()
CNN_Model.compile(optimizer="adam", loss='categorical_crossentropy', metrics=['accuracy'])
# In[10]:
# Create Callback Checkpoint
checkpoint_path = "CNN_Model_Checkpoint"
Checkpoint = ModelCheckpoint(checkpoint_path, monitor="val_accuracy", save_best_only=True)
# Create Early Stopping Callback to monitor the accuracy
Early_Stopping = EarlyStopping(monitor = 'val_accuracy', patience = 15, restore_best_weights = True, verbose=1)
# Create ReduceLROnPlateau Callback to reduce overfitting by decreasing learning rate
Reducing_LR = tf.keras.callbacks.ReduceLROnPlateau( monitor='val_loss',
factor=0.2,
patience=2,
# min_lr=0.000005,
verbose=1)
callbacks = [Early_Stopping, Reducing_LR]
steps_per_epoch = train_data.n // train_data.batch_size
validation_steps = test_data.n // test_data.batch_size
# In[12]:
CNN_history = CNN_Model.fit( train_data , validation_data= test_data , epochs=80, batch_size= batch_size,
callbacks=callbacks, steps_per_epoch= steps_per_epoch, validation_steps=validation_steps)
# In[13]:
CNN_Score = CNN_Model.evaluate(test_data)
print(" Test Loss: {:.5f}".format(CNN_Score[0]))
print("Test Accuracy: {:.2f}%".format(CNN_Score[1] * 100))
# In[14]:
CNN_Score = CNN_Model.evaluate(train_data)
print(" Train Loss: {:.5f}".format(CNN_Score[0]))
print("Train Accuracy: {:.2f}%".format(CNN_Score[1] * 100))
# In[15]:
def plot_curves(history):
loss = history.history["loss"]
val_loss = history.history["val_loss"]
accuracy = history.history["accuracy"]
val_accuracy = history.history["val_accuracy"]
epochs = range(len(history.history["loss"]))
plt.figure(figsize=(15,5))
#plot loss
plt.subplot(1, 2, 1)
plt.plot(epochs, loss, label = "training_loss")
plt.plot(epochs, val_loss, label = "val_loss")
plt.title("Loss")
plt.xlabel("epochs")
plt.legend()
#plot accuracy
plt.subplot(1, 2, 2)
plt.plot(epochs, accuracy, label = "training_accuracy")
plt.plot(epochs, val_accuracy, label = "val_accuracy")
plt.title("Accuracy")
plt.xlabel("epochs")
plt.legend()
#plt.tight_layout()
# In[16]:
plot_curves(CNN_history)
# In[17]:
CNN_Predictions = CNN_Model.predict(test_data)
# Choosing highest probalbilty class in every prediction
CNN_Predictions = np.argmax(CNN_Predictions, axis=1)
# In[18]:
test_data.class_indices
# In[19]:
import seaborn as sns
from sklearn.metrics import confusion_matrix
fig, ax= plt.subplots(figsize=(15,10))
cm=confusion_matrix(test_data.labels, CNN_Predictions)
sns.heatmap(cm, annot=True, fmt='g', ax=ax)
ax.set_xlabel('Predicted labels',fontsize=15, fontweight='bold')
ax.set_ylabel('True labels', fontsize=15, fontweight='bold')
ax.set_title('CNN Confusion Matrix', fontsize=20, fontweight='bold')
# In[20]:
# Print classification report and confusion matrix
print('Classification report:')
print(classification_report(test_data.labels, CNN_Predictions))
# In[21]:
Emotion_Classes = ['Angry',
'Disgust',
'Fear',
'Happy',
'Neutral',
'Sad',
'Surprise']
# In[22]:
# Shuffling Test Data to show diffrent classes
test_preprocessor = ImageDataGenerator(
rescale = 1 / 255.,
)
test_generator = test_preprocessor.flow_from_directory(
test_data_path,
class_mode="categorical",
target_size=(img_shape,img_shape),
color_mode="rgb",
shuffle=True,
batch_size=batch_size,
)
# In[23]:
# Display 10 random pictures from the dataset with their labels
Random_batch = np.random.randint(0, len(test_generator) - 1)
Random_Img_Index = np.random.randint(0, batch_size - 1 , 10)
fig, axes = plt.subplots(nrows=2, ncols=5, figsize=(25, 10),
subplot_kw={'xticks': [], 'yticks': []})
for i, ax in enumerate(axes.flat):
Random_Img = test_generator[Random_batch][0][Random_Img_Index[i]]
Random_Img_Label = np.argmax(test_generator[Random_batch][1][Random_Img_Index[i]])
Model_Prediction = np.argmax(CNN_Model.predict( tf.expand_dims(Random_Img, axis=0) , verbose=0))
ax.imshow(Random_Img)
if Emotion_Classes[Random_Img_Label] == Emotion_Classes[Model_Prediction]:
color = "green"
else:
color = "red"
ax.set_title(f"True: {Emotion_Classes[Random_Img_Label]}\nPredicted: {Emotion_Classes[Model_Prediction]}", color=color)
plt.show()
plt.tight_layout()
# In[24]:
CNN_Model.save("Facial_Expressions.h5")
# In[25]:
from IPython.display import FileLink
FileLink("Facial_Expressions.h5")
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment