Co-attentional Transformers for Story-Based Video Understanding

TL;DR

This paper introduces a co-attentional transformer model for story-based video understanding, effectively capturing long-term dependencies and interactions in visual narratives, and demonstrates superior performance on the DramaQA dataset.

Contribution

The paper presents a novel co-attentional transformer architecture tailored for visual story understanding, improving long-term dependency modeling in video question answering tasks.

Findings

Outperforms baseline by 8 percentage points overall

Achieves at least 4.95 and up to 12.8 percentage points improvement across difficulty levels

Beats the winner of the DramaQA challenge

Abstract

Inspired by recent trends in vision and language learning, we explore applications of attention mechanisms for visio-lingual fusion within an application to story-based video understanding. Like other video-based QA tasks, video story understanding requires agents to grasp complex temporal dependencies. However, as it focuses on the narrative aspect of video it also requires understanding of the interactions between different characters, as well as their actions and their motivations. We propose a novel co-attentional transformer model to better capture long-term dependencies seen in visual stories such as dramas and measure its performance on the video question answering task. We evaluate our approach on the recently introduced DramaQA dataset which features character-centered video story understanding questions. Our model outperforms the baseline model by 8 percentage points overall, at least 4.95 and up to 12.8 percentage points on all difficulty levels and manages to beat the winner of the DramaQA challenge.