Ambiguous MS COCO Japanese--English Multimodal Task

[HOME]

INTRODUCTION

We present the MS COCO multimodal translation task for English <--> Japanese translation.

TASK DESCRIPTION

This is a Japanese--English multimodal machine translation task. We provide the Japanese--English Ambiguous MS COCO dataset for validation and testing, which contains ambiguous verbs that may require visual information in images for disambiguation. Participants can use the standard and additional training data (see below for details) to train their multimodal machine translation system.

Corpora

• The evaluation data and links to training data are all listed here. This page contains several helpful instructions that will simplify your data preparation and experimentation.

Submission Details

• Constrained submission: Please use the Flickr30kEntities Japanese (F30kEnt-Jp) dataset from here.
• Unconstrained submission: As additional training data, you can use the MS COCO English data from here and STAIR Japanese image captions from here. If you use either of these two or any other data sources then you must claim this in your submission.

CONTACT

For general questions, comments, etc. please email to "wat-organizer -at- googlegroups -dot- com". For questions related to this task contact "prajdabre -at- gmail -dot- com" or "chu -at- i -dot- kyoto-u -dot- ac -dot- jp".

NICT (National Institute of Information and Communications Technology)
Kyoto University
Last Modified: 2021-01-04