DATA AUGMENTATION WITH PGGAN & MULTICLASS CLASSIFICATION WITH VGG16

Abstract

In recent years, deep learning has made significant strides in image classification tasks. However, the performance of these models is often constrained by the availability of large, annotated datasets. Data augmentation has emerged as a critical technique to address this limitation by artificially increasing the diversity of training data. This paper explores the use of Progressive Growing of GANs (PGGAN) for data augmentation in a multiclass image classification problem. PGGANs, known for their ability to generate high-fidelity images, are leveraged to augment a limited dataset, thus enhancing the performance of a pre-trained VGG16 model fine-tuned for multiclass classification. We demonstrate the effectiveness of our approach through extensive experiments on a publicly available dataset, showcasing that the augmented data significantly improves the model's accuracy and robustness. Our results indicate that the integration of PGGAN-generated images into the training process can mitigate overfitting and provide substantial performance gains, especially in scenarios with limited data. This study highlights the potential of GAN-based data augmentation in advancing the
capabilities of convolutional neural networks (CNNs) in image classification tasks.