Deep Learning Based Fine Grained Image Classification

Abstract

Image classification, specifically object classification is the focused research area in the computer vision and machine learning field in the past decade. In image classification a label or category is assigned to an input image based on its content. With breakthroughs in deep learning-based approaches, performance of image classification models' has improved significantly, particularly fine-grained image classification, which includes discriminating between items of the same category with slight changes. The object classification can be categorised as coarse grained object classification, which identifies highly diverse object categories, such as an elephant and a bus. One example of this type of object classification is a bus and an elephant. On the other hand, fine-grained image categorization seeks to recognise photos as belonging to distinct species of animals, birds, or plants, as well as distinct models of automobiles, versions of aircraft, and so on. The purpose of this study is to evaluate previously published research that investigates deep learning techniques for the classification of fine-grained images and to compare the effectiveness of these techniques using datasets that are open to the public.