PR-MCS: Perturbation Robust Metric for MultiLingual Image Captioning

Yongil Kim, Yerin Hwang, Hyeongu Yun, Seunghyun Yoon, Trung Bui, Kyomin Jung

Submission Type: Regular Long Paper

Submission Track: Resources and Evaluation

Submission Track 2: Language Grounding to Vision, Robotics and Beyond

Keywords: Image Captioning metric, Perturbation Robustness

TL;DR: Perturbation Robust Multi-Lingual CLIPScore

Abstract: Vulnerability to lexical perturbation is a critical weakness of automatic evaluation metrics for image captioning. This paper proposes Perturbation Robust Multi-Lingual CLIPScore(PR-MCS), which exhibits robustness to such perturbations, as a novel reference-free image captioning metric applicable to multiple languages. To achieve perturbation robustness, we fine-tune the text encoder of CLIP with our language-agnostic method to distinguish the perturbed text from the original text. To verify the robustness of PR-MCS, we introduce a new fine-grained evaluation dataset consisting of detailed captions, critical objects, and the relationships between the objects for 3,000 images in five languages. In our experiments, PR-MCS significantly outperforms baseline metrics in capturing lexical noise of all various perturbation types in all five languages, while maintaining a strong correlation with human judgments.

Submission Number: 1088