Image and Encoded Text Fusion for Multi-Modal Classification

TL;DR

This paper introduces a novel multi-modal classification method that fuses images and encoded text to enhance performance, demonstrating its effectiveness through experiments on large-scale datasets and comparisons with existing strategies.

Contribution

The paper proposes a new fusion technique that embeds encoded text onto images for improved multi-modal classification using CNNs, outperforming traditional fusion methods.

Findings

The approach achieves better accuracy than individual sources.

It outperforms early and late fusion strategies.

Experimental results on large datasets validate its effectiveness.

Abstract

Multi-modal approaches employ data from multiple input streams such as textual and visual domains. Deep neural networks have been successfully employed for these approaches. In this paper, we present a novel multi-modal approach that fuses images and text descriptions to improve multi-modal classification performance in real-world scenarios. The proposed approach embeds an encoded text onto an image to obtain an information-enriched image. To learn feature representations of resulting images, standard Convolutional Neural Networks (CNNs) are employed for the classification task. We demonstrate how a CNN based pipeline can be used to learn representations of the novel fusion approach. We compare our approach with individual sources on two large-scale multi-modal classification datasets while obtaining encouraging results. Furthermore, we evaluate our approach against two famous multi-modal strategies namely early fusion and late fusion.