Image Classification and Calorie Estimation Report
Python Code for Image Classification and Calorie Estimation
import os
import zipfile
import json
import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint
import matplotlib.pyplot as plt
from PIL import Image
Step 1: Unzip the datasets
def unzip_data(zip_path, extract_to):
with zipfile.ZipFile(zip_path, 'r') as zip_ref:
zip_ref.extractall(extract_to)
# Paths to the zipped data and extraction directory
data_paths = [
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F1.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F2.zip',
Image Classification and Calorie Estimation Report
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F3.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F4.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F5.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F6.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F9.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F10.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/images/F11.zip',
'/Users/ronitvyas/Downloads/archive/food-101/food-101/meta.zip'
]
extract_paths = '/Users/ronitvyas/Downloads/archive/food-101/food-101/extracted/'
# Unzip all files
for zip_path in data_paths:
unzip_data(zip_path, extract_paths)
# Custom function to filter valid image files
def is_valid_image(file_path):
try:
Image.open(file_path)
return True
except:
return False
# Remove invalid files
Image Classification and Calorie Estimation Report
for root, dirs, files in os.walk(extract_paths):
for file in files:
file_path = os.path.join(root, file)
if not is_valid_image(file_path):
os.remove(file_path)
Step 2: Load and preprocess the data
data_dir = '/Users/ronitvyas/Downloads/archive/food-101/food-101/extracted'
image_size = (128, 128)
batch_size = 32
datagen = ImageDataGenerator(rescale=1./255, validation_split=0.2)
train_generator = datagen.flow_from_directory(
data_dir,
target_size=image_size,
batch_size=batch_size,
class_mode='categorical',
subset='training'
)
validation_generator = datagen.flow_from_directory(
data_dir,
target_size=image_size,
Image Classification and Calorie Estimation Report
batch_size=batch_size,
class_mode='categorical',
subset='validation'
)
Step 3: Build the CNN model
model = Sequential([
Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 3)),
MaxPooling2D((2, 2)),
Conv2D(64, (3, 3), activation='relu'),
MaxPooling2D((2, 2)),
Conv2D(128, (3, 3), activation='relu'),
MaxPooling2D((2, 2)),
Flatten(),
Dense(128, activation='relu'),
Dropout(0.5),
Dense(len(train_generator.class_indices), activation='softmax')
])
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
Step 4: Train the model
early_stopping = EarlyStopping(monitor='val_loss', patience=5, restore_best_weights=True)
model_checkpoint = ModelCheckpoint('food_recognition_model.h5', save_best_only=True)
Image Classification and Calorie Estimation Report
history = model.fit(
train_generator,
epochs=50,
validation_data=validation_generator,
callbacks=[early_stopping, model_checkpoint]
)
Step 5: Evaluate the model
plt.plot(history.history['accuracy'], label='accuracy')
plt.plot(history.history['val_accuracy'], label='val_accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.legend(loc='lower right')
plt.title('Model Accuracy')
plt.savefig('/Users/ronitvyas/Downloads/archive/food-101/food-101/extracted/accuracy_plot.png')
plt.show()
plt.plot(history.history['loss'], label='loss')
plt.plot(history.history['val_loss'], label='val_loss')
plt.xlabel('Epoch')
plt.ylabel('Loss')
plt.legend(loc='upper right')
plt.title('Model Loss')
Image Classification and Calorie Estimation Report
plt.savefig('/Users/ronitvyas/Downloads/archive/food-101/food-101/extracted/loss_plot.png')
plt.show()
Step 6: Map food items to their calorie content
calories_file = '/Users/ronitvyas/Downloads/archive/food-101/food-101/extracted/meta/calories.txt'
calories_map = {}
with open(calories_file, 'r') as file:
for line in file:
class_name, calorie = line.strip().split()
calories_map[class_name] = float(calorie)
# Function to predict food item and estimate calories
def predict_and_estimate_calories(image_path):
img = Image.open(image_path)
img = img.resize(image_size)
img_array = np.array(img) / 255.0
img_array = np.expand_dims(img_array, axis=0)
predictions = model.predict(img_array)
predicted_class = list(train_generator.class_indices.keys())[np.argmax(predictions)]
estimated_calories = calories_map.get(predicted_class, "Calorie information not available")
return predicted_class, estimated_calories
Image Classification and Calorie Estimation Report
# Example usage
image_path =
'/Users/ronitvyas/Downloads/archive/food-101/food-101/extracted/F1/apple_pie/134.jpg'
predicted_class, estimated_calories = predict_and_estimate_calories(image_path)
print(f'Predicted Class: {predicted_class}, Estimated Calories: {estimated_calories}')
Model Accuracy and Loss Visualization
Image Classification and Calorie Estimation Report
Interpretation of Results
The graphs above illustrate the accuracy and loss of the model over 6 epochs. The accuracy graph
shows that the training
accuracy improves steadily, reaching around 26% by the end of the sixth epoch. However, the
validation accuracy remains
consistently low, indicating that the model is not generalizing well to unseen data.
The loss graph also indicates a divergence between the training and validation losses. While the
training loss decreases
consistently, the validation loss increases after the first epoch, suggesting that the model is
overfitting to the training data.
Overall, the model's performance on the validation set indicates that it is not yet able to generalize
well to new data.
Further steps such as data augmentation, regularization, or fine-tuning the model architecture and
hyperparameters might be
required to improve the model's generalization capability.
.png)
0 Comments