BERTHA: Video Captioning Evaluation Via Transfer-Learned Human Assessment
This work addresses the problem of automated evaluation for video captioning, which is important for researchers and developers in computer vision and NLP, but it is incremental as it builds on existing BERT models and datasets.
The paper tackles the challenge of evaluating video captioning systems by introducing BERTHA, a deep learning model based on BERT that learns to assess captions similarly to humans, using a dataset of human judgments from TRECVid tasks, and it outperforms common metrics in some setups.
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.