Pre-Trained Deep Neural Network Model of VGG-16 For Flower Image Classification

Abstract

Flowers are vital to our ecosystem, feeding insects, birds, animals, and humans, and serving as medicinal
resources. Understanding flowers helps identify new or rare species, benefiting the medicinal industry. Flower
classification is crucial and has been extensively researched. Current methods fall into two categories: manual
feature extraction and deep learning. Manual methods extract color, texture, and shape features, combining
them with machine learning for classification. However, traditional methods have low accuracy and deep neural
networks require large datasets. Our work proposes a fine-tuned VGG16 deep learning model. Experiments on
the Oxford flower-102 dataset show that data enhancement improves classification accuracy and robustness,
outperforming traditional models