BERTHA: Video Captioning Evaluation Via Transfer-Learned Human Assessment

TL;DR

BERTHA is a deep learning-based evaluation method for video captioning that leverages transfer learning from BERT to mimic human judgment, outperforming traditional metrics in certain scenarios.

Contribution

The paper introduces BERTHA, a novel BERT-based model trained on human evaluation data to assess video captioning quality more accurately.

Findings

BERTHA outperforms traditional metrics in some evaluation setups.

The model effectively mimics human judgments of caption quality.

The dataset of human evaluations will be publicly available.

Abstract

Evaluating video captioning systems is a challenging task as there are multiple factors to consider; for instance: the fluency of the caption, multiple actions happening in a single scene, and the human bias of what is considered important. Most metrics try to measure how similar the system generated captions are to a single or a set of human-annotated captions. This paper presents a new method based on a deep learning model to evaluate these systems. The model is based on BERT, which is a language model that has been shown to work well in multiple NLP tasks. The aim is for the model to learn to perform an evaluation similar to that of a human. To do so, we use a dataset that contains human evaluations of system generated captions. The dataset consists of the human judgments of the captions produce by the system participating in various years of the TRECVid video to text task. These annotations will be made publicly available. BERTHA obtain favourable results, outperforming the commonly used metrics in some setups.