Stop Pre-Training: Adapt Visual-Language Models to Unseen Languages

TL;DR

This paper introduces a simple method to adapt vision-language models to unseen languages using multilingual pre-trained language models and machine translation, improving performance without needing target language data.

Contribution

It proposes a cross-lingual token embedding alignment approach that enables VLP models to work effectively in unseen languages without large parallel corpora.

Findings

Outperforms state-of-the-art multilingual vision-language models

Effective across image-text retrieval, visual entailment, and visual reasoning tasks

Does not require target language data or large parallel corpora

Abstract

Vision-Language Pre-training (VLP) has advanced the performance of many vision-language tasks, such as image-text retrieval, visual entailment, and visual reasoning. The pre-training mostly utilizes lexical databases and image queries in English. Previous work has demonstrated that the pre-training in English does not transfer well to other languages in a zero-shot setting. However, multilingual pre-trained language models (MPLM) have excelled at a variety of single-modal language tasks. In this paper, we propose a simple yet efficient approach to adapt VLP to unseen languages using MPLM. We utilize a cross-lingual contextualized token embeddings alignment approach to train text encoders for non-English languages. Our approach does not require image input and primarily uses machine translation, eliminating the need for target language data. Our evaluation across three distinct tasks (image-text retrieval, visual entailment, and natural language visual reasoning) demonstrates that this approach outperforms the state-of-the-art multilingual vision-language models without requiring large parallel corpora. Our code is available at https://github.com/Yasminekaroui/CliCoTea.