Detecting Concrete Visual Tokens for Multimodal Machine Translation

TL;DR

This paper proposes new methods for detecting and selecting visually-grounded tokens in multimodal machine translation, improving the use of visual context and translation performance.

Contribution

It introduces novel detection and selection techniques for concrete tokens, enhancing multimodal translation models with better visual grounding.

Findings

Performance improvements over baseline models

Enhanced visual context utilization during translation

Effective detection methods for visually-grounded tokens

Abstract

The challenge of visual grounding and masking in multimodal machine translation (MMT) systems has encouraged varying approaches to the detection and selection of visually-grounded text tokens for masking. We introduce new methods for detection of visually and contextually relevant (concrete) tokens from source sentences, including detection with natural language processing (NLP), detection with object detection, and a joint detection-verification technique. We also introduce new methods for selection of detected tokens, including shortest $n$ tokens, longest $n$ tokens, and all detected concrete tokens. We utilize the GRAM MMT architecture to train models against synthetically collated multimodal datasets of source images with masked sentences, showing performance improvements and improved usage of visual context during translation tasks over the baseline model.