Fine-Grained Classification via Mixture of Deep Convolutional Neural Networks

TL;DR

This paper introduces MixDCNN, a novel deep learning system that partitions images into subsets, learns specialized CNNs for each, and combines their outputs for improved fine-grained classification accuracy.

Contribution

The paper proposes a joint end-to-end training framework for multiple CNN experts tailored for fine-grained image classification, outperforming previous methods.

Findings

Achieves 12.7% relative improvement over existing methods

Consistently outperforms prior techniques on three datasets

State-of-the-art results on two datasets

Abstract

We present a novel deep convolutional neural network (DCNN) system for fine-grained image classification, called a mixture of DCNNs (MixDCNN). The fine-grained image classification problem is characterised by large intra-class variations and small inter-class variations. To overcome these problems our proposed MixDCNN system partitions images into K subsets of similar images and learns an expert DCNN for each subset. The output from each of the K DCNNs is combined to form a single classification decision. In contrast to previous techniques, we provide a formulation to perform joint end-to-end training of the K DCNNs simultaneously. Extensive experiments, on three datasets using two network structures (AlexNet and GoogLeNet), show that the proposed MixDCNN system consistently outperforms other methods. It provides a relative improvement of 12.7% and achieves state-of-the-art results on two datasets.