Towards Image Captioning for the Portuguese Language: Evaluation on a Translated Dataset

João Gondim, Daniela Claro, Marlo Souza

2022

Abstract

Automatic describing an image comprehends the representation from the scene elements to generate a concise natural language description. Few resources, particularly annotated datasets for the Portuguese language, discourage the development of new methods in languages other than English. Thus, we propose a new image captioning method for the Portuguese language. We provide an analysis empowered by an encoder-decoder model with an attention mechanism when employing a multimodal dataset translated into Portuguese. Our findings suggest that: 1) the original and translated datasets are pretty similar considering the measure achievements; 2) the translation approach includes some dirty sentence formulations that disturb our model for the Portuguese language.

Paper Citation

in Harvard Style

Gondim J., Claro D. and Souza M. (2022). Towards Image Captioning for the Portuguese Language: Evaluation on a Translated Dataset. In Proceedings of the 24th International Conference on Enterprise Information Systems - Volume 2: ICEIS, ISBN 978-989-758-569-2, pages 384-393. DOI: 10.5220/0011080000003179

in Bibtex Style

@conference{iceis22,
author={João Gondim and Daniela Claro and Marlo Souza},
title={Towards Image Captioning for the Portuguese Language: Evaluation on a Translated Dataset},
booktitle={Proceedings of the 24th International Conference on Enterprise Information Systems - Volume 2: ICEIS,},
year={2022},
pages={384-393},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011080000003179},
isbn={978-989-758-569-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 24th International Conference on Enterprise Information Systems - Volume 2: ICEIS,
TI - Towards Image Captioning for the Portuguese Language: Evaluation on a Translated Dataset
SN - 978-989-758-569-2
AU - Gondim J.
AU - Claro D.
AU - Souza M.
PY - 2022
SP - 384
EP - 393
DO - 10.5220/0011080000003179