0xnomy / image-classification-ensemble-model-vgg16-inceptionv3

Image classification using InceptionV3 and VGG16

Tom and Jerry Image Classification

This project aims to classify images of Tom and Jerry characters using deep learning models. We utilize pre-trained models, InceptionV3 and VGG16, to enhance our image classification tasks. My Kaggle Profile

Table of Contents

• Introduction
• Data Preprocessing
• Model Training
• Results

Introduction

This project involves the classification of images into four categories: Jerry, Tom, None, and Both. We use transfer learning with InceptionV3 and VGG16 models for this purpose.

Data Preprocessing

• Setting constraints and loading dataset.
• Splitting the dataset into training and validation sets.
• Displaying sample images and class distribution.

Model Training

We trained two models: InceptionV3 and VGG16. Both models were fine-tuned on our custom dataset.

Results

• The models achieved the following accuracies:
  • InceptionV3: Training Accuracy - 85.40%, Validation Accuracy - 71.14%
  • VGG16: Training Accuracy - 99.79%, Validation Accuracy - 83.84%

Conclusion

This project demonstrates the effectiveness of transfer learning in image classification tasks. By leveraging pre-trained models like InceptionV3 and VGG16, we achieved high accuracy in classifying images of Tom and Jerry characters.

References

• TensorFlow Documentation
• Keras Documentation