An Efficient Hindi Image Captioning with Transformer Model

DOI

10.2991/978-94-6463-738-0_4How to use a DOI?

Keywords

Image Captioning; Transformer; DenseNet121; Data Augmentation; BLEU; pre-trained model

Abstract

Automatic image captioning in Indian languages like Hindi presents unique challenges due to the complex grammar and lack of large-scale datasets compared to English. This research introduces a new way of generating Hindi image captions using a pre-trained DenseNet121 as an encoder and a transformer as a decoder. The model incorporates data augmentation to improve the generalization of captions. We evaluated our model on a benchmark Hindi image captioning dataset. According to the experimental results, the proposed model has outperformed the state-of-the-art models in Hindi paragraph captioning. The model has achieved a BLEU-1 score of 76.54, a BLEU-2 score of 57.41, a BLEU-3 score of 36.64, and a BLEU-4 score of 20.71. Additionally, we have demonstrated competitive performance when compared to other existing methods.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Ashutosh Chandra Bhensle
AU  - Jyoti Prakash Patra
AU  - Sumitra Samal
PY  - 2025
DA  - 2025/06/22
TI  - An Efficient Hindi Image Captioning with Transformer Model
BT  - Proceedings of the International Conference on Advances and Applications in Artificial Intelligence (ICAAAI 2025)
PB  - Atlantis Press
SP  - 32
EP  - 43
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-738-0_4
DO  - 10.2991/978-94-6463-738-0_4
ID  - Bhensle2025
ER  -