GPT4SGG: Synthesizing Scene Graphs from Holistic and Region-specific Narratives

TL;DR

GPT4SGG introduces a novel framework that leverages both holistic and region-specific narratives with large language models to improve scene graph generation from caption data, addressing ambiguity and bias issues.

Contribution

The paper proposes GPT4SGG, a divide-and-conquer approach using large language models to synthesize accurate scene graphs from complex caption data by decomposing scenes into regions.

Findings

Significant performance improvement in scene graph accuracy.

Effective handling of ambiguity and long-tail bias.

Enhanced comprehensiveness of generated scene graphs.

Abstract

Training Scene Graph Generation (SGG) models with natural language captions has become increasingly popular due to the abundant, cost-effective, and open-world generalization supervision signals that natural language offers. However, such unstructured caption data and its processing pose significant challenges in learning accurate and comprehensive scene graphs. The challenges can be summarized as three aspects: 1) traditional scene graph parsers based on linguistic representation often fail to extract meaningful relationship triplets from caption data. 2) grounding unlocalized objects of parsed triplets will meet ambiguity issues in visual-language alignment. 3) caption data typically are sparse and exhibit bias to partial observations of image content. Aiming to address these problems, we propose a divide-and-conquer strategy with a novel framework named \textit{GPT4SGG}, to obtain more accurate and comprehensive scene graph signals. This framework decomposes a complex scene into a bunch of simple regions, resulting in a set of region-specific narratives. With these region-specific narratives (partial observations) and a holistic narrative (global observation) for an image, a large language model (LLM) performs the relationship reasoning to synthesize an accurate and comprehensive scene graph. Experimental results demonstrate \textit{GPT4SGG} significantly improves the performance of SGG models trained on image-caption data, in which the ambiguity issue and long-tail bias have been well-handled with more accurate and comprehensive scene graphs.