FDSG: Forecasting Dynamic Scene Graphs

TL;DR

FDSG is a novel framework that predicts future scene graphs in videos by modeling entity and relationship dynamics, outperforming existing methods on dynamic scene graph tasks.

Contribution

The paper introduces FDSG, a new approach that forecasts future scene graphs in videos using query decomposition and neural stochastic differential equations.

Findings

FDSG outperforms state-of-the-art methods on scene graph forecasting tasks.

Introduces Scene Graph Forecasting, a new benchmark for future scene graph prediction.

Effectively models entity and relationship dynamics over time.

Abstract

Dynamic scene graph generation extends scene graph generation from images to videos by modeling entity relationships and their temporal evolution. However, existing methods either generate scene graphs from observed frames without explicitly modeling temporal dynamics, or predict only relationships while assuming static entity labels and locations. These limitations hinder effective extrapolation of both entity and relationship dynamics, restricting video scene understanding. We propose Forecasting Dynamic Scene Graphs (FDSG), a novel framework that predicts future entity labels, bounding boxes, and relationships, for unobserved frames, while also generating scene graphs for observed frames. Our scene graph forecast module leverages query decomposition and neural stochastic differential equations to model entity and relationship dynamics. A temporal aggregation module further refines predictions by integrating forecasted and observed information via cross-attention. To benchmark FDSG, we introduce Scene Graph Forecasting, a new task for full future scene graph prediction. Experiments on Action Genome show that FDSG outperforms state-of-the-art methods on dynamic scene graph generation, scene graph anticipation, and scene graph forecasting. Codes will be released upon publication.