VieCap4H-VLSP 2021: ObjectAoA-Enhancing performance of Object Relation Transformer with Attention on Attention for Vietnamese image captioning

TL;DR

This paper introduces an enhanced Object Relation Transformer with Attention on Attention for Vietnamese image captioning, significantly improving image understanding and captioning performance.

Contribution

It extends the Object Relation Transformer architecture with Attention on Attention, achieving better results on Vietnamese image captioning datasets.

Findings

Outperforms original Transformer on VieCap4H dataset

Significant improvement in captioning accuracy

Effective enhancement of visual understanding

Abstract

Image captioning is currently a challenging task that requires the ability to both understand visual information and use human language to describe this visual information in the image. In this paper, we propose an efficient way to improve the image understanding ability of transformer-based method by extending Object Relation Transformer architecture with Attention on Attention mechanism. Experiments on the VieCap4H dataset show that our proposed method significantly outperforms its original structure on both the public test and private test of the Image Captioning shared task held by VLSP.