vieCap4H-VLSP 2021: Vietnamese Image Captioning for Healthcare Domain using Swin Transformer and Attention-based LSTM

TL;DR

This paper introduces a Vietnamese healthcare image captioning model using Swin Transformer and attention-based LSTM, achieving competitive BLEU4 scores and ranking third in a VLSP challenge.

Contribution

The study proposes a novel encoder-decoder architecture with Swin Transformer and attention-based LSTM for Vietnamese healthcare image captioning.

Findings

Achieved BLEU4 score of 0.293 on vietCap4H dataset

Ranked 3rd on the VLSP 2021 private leaderboard

Demonstrated effectiveness of Swin Transformer in Vietnamese image captioning

Abstract

This study presents our approach on the automatic Vietnamese image captioning for healthcare domain in text processing tasks of Vietnamese Language and Speech Processing (VLSP) Challenge 2021, as shown in Figure 1. In recent years, image captioning often employs a convolutional neural network-based architecture as an encoder and a long short-term memory (LSTM) as a decoder to generate sentences. These models perform remarkably well in different datasets. Our proposed model also has an encoder and a decoder, but we instead use a Swin Transformer in the encoder, and a LSTM combined with an attention module in the decoder. The study presents our training experiments and techniques used during the competition. Our model achieves a BLEU4 score of 0.293 on the vietCap4H dataset, and the score is ranked the 3$^{rd}$ place on the private leaderboard. Our code can be found at \url{https://git.io/JDdJm}.