Keep it Consistent: Topic-Aware Storytelling from an Image Stream via Iterative Multi-agent Communication

TL;DR

This paper introduces a topic-aware multi-agent framework for visual storytelling that improves narrative coherence by jointly generating a global topic description and a story from an image stream.

Contribution

It proposes a novel multi-agent communication approach that integrates topic detection and story generation, enhancing semantic consistency in visual storytelling.

Findings

Outperforms state-of-the-art methods on VIST dataset

Produces more coherent and semantically relevant stories

Validated by quantitative, ablation, and human evaluations

Abstract

Visual storytelling aims to generate a narrative paragraph from a sequence of images automatically. Existing approaches construct text description independently for each image and roughly concatenate them as a story, which leads to the problem of generating semantically incoherent content. In this paper, we propose a new way for visual storytelling by introducing a topic description task to detect the global semantic context of an image stream. A story is then constructed with the guidance of the topic description. In order to combine the two generation tasks, we propose a multi-agent communication framework that regards the topic description generator and the story generator as two agents and learn them simultaneously via iterative updating mechanism. We validate our approach on VIST dataset, where quantitative results, ablations, and human evaluation demonstrate our method's good ability in generating stories with higher quality compared to state-of-the-art methods.