This repository contains the automatic assessment system for Image Description by Language Learners.
We focus on image description and a corresponding assessment system for language learners. To achieve automatic assessment of image description, we construct a novel dataset, the Language Learner Image Description (LLID) dataset, which consists of images, their descriptions, and assessment annotations. Then, we propose a novel task of automatic error correction for image description, and we develop a baseline model that encodes multimodal information from a learner sentence with an image and accurately decodes a corrected sentence. Our experimental results show that the developed model can revise errors that cannot be revised without an image.
The easiest way to use LLID Feedback and its dependencies is using docker.
docker-compose build
docker-compose up
- Download pretrained model in 'llid/model' （以下のリンク先のllid_dataというフォルダの中身をllid/modelに置く） https://drive.google.com/drive/folders/1jXtTCGbbnoO8x-uEnwWjQHZe20fuyzjD?usp=sharing
- Download pretrained image features in 'llid/data' （以下のリンク先のepoch_16.thというファイルをllid/modelに置く） https://drive.google.com/file/d/1-8zKLeULvVmYLVM2qYqSReGk2dJU6kNq/view?usp=sharing