FDSG: Forecasting Dynamic Scene Graphs
This addresses the limitation of existing video scene understanding methods that fail to fully extrapolate entity and relationship dynamics, enabling more comprehensive future scene prediction.
The paper tackles the problem of predicting future entity labels, bounding boxes, and relationships in videos, introducing a new task called Scene Graph Forecasting. The proposed FDSG framework outperforms state-of-the-art methods on Action Genome across dynamic scene graph generation, anticipation, and forecasting tasks.
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.