Models See Hallucinations: Evaluating the Factuality in Video Captioning
It addresses a critical issue for video captioning applications by highlighting and providing tools to measure factual inconsistencies, though it is incremental as it builds on prior work in text-to-text tasks.
The paper tackles the problem of factual errors in video captioning, finding that 57.0% of model-generated sentences contain such errors, and proposes a new metric, FactVC, that outperforms existing ones in evaluating factuality.
Video captioning aims to describe events in a video with natural language. In recent years, many works have focused on improving captioning models' performance. However, like other text generation tasks, it risks introducing factual errors not supported by the input video. These factual errors can seriously affect the quality of the generated text, sometimes making it completely unusable. Although factual consistency has received much research attention in text-to-text tasks (e.g., summarization), it is less studied in the context of vision-based text generation. In this work, we conduct a detailed human evaluation of the factuality in video captioning and collect two annotated factuality datasets. We find that 57.0% of the model-generated sentences have factual errors, indicating it is a severe problem in this field. However, existing evaluation metrics are mainly based on n-gram matching and show little correlation with human factuality annotation. We further propose a weakly-supervised, model-based factuality metric FactVC, which outperforms previous metrics on factuality evaluation of video captioning. The datasets and metrics will be released to promote future research for video captioning.