Thinking Hallucination for Video Captioning

TL;DR

This paper investigates the root causes of hallucination in video captioning models, proposing solutions like auxiliary heads and context gates, and introduces a new metric COAHA to better evaluate hallucination, achieving state-of-the-art results.

Contribution

It identifies key factors causing hallucination in video captioning and proposes novel methods to mitigate them, along with a new evaluation metric for hallucination.

Findings

Proposed auxiliary heads improve visual feature robustness.

Context gates enhance feature fusion during captioning.

Achieved state-of-the-art results on MSR-VTT and MSVD datasets.

Abstract

With the advent of rich visual representations and pre-trained language models, video captioning has seen continuous improvement over time. Despite the performance improvement, video captioning models are prone to hallucination. Hallucination refers to the generation of highly pathological descriptions that are detached from the source material. In video captioning, there are two kinds of hallucination: object and action hallucination. Instead of endeavoring to learn better representations of a video, in this work, we investigate the fundamental sources of the hallucination problem. We identify three main factors: (i) inadequate visual features extracted from pre-trained models, (ii) improper influences of source and target contexts during multi-modal fusion, and (iii) exposure bias in the training strategy. To alleviate these problems, we propose two robust solutions: (a) the introduction of auxiliary heads trained in multi-label settings on top of the extracted visual features and (b) the addition of context gates, which dynamically select the features during fusion. The standard evaluation metrics for video captioning measures similarity with ground truth captions and do not adequately capture object and action relevance. To this end, we propose a new metric, COAHA (caption object and action hallucination assessment), which assesses the degree of hallucination. Our method achieves state-of-the-art performance on the MSR-Video to Text (MSR-VTT) and the Microsoft Research Video Description Corpus (MSVD) datasets, especially by a massive margin in CIDEr score.