Spatial-Temporal Transformer for Dynamic Scene Graph Generation

TL;DR

This paper introduces STTran, a spatial-temporal Transformer model for dynamic scene graph generation from videos, capturing spatial relationships within frames and temporal dependencies across frames, validated on the Action Genome dataset.

Contribution

The paper proposes a novel flexible neural network architecture combining spatial and temporal modules for dynamic scene graph generation from videos.

Findings

Superior performance on Action Genome dataset

Effective modeling of spatial and temporal relationships

Ablative studies validate module contributions

Abstract

Dynamic scene graph generation aims at generating a scene graph of the given video. Compared to the task of scene graph generation from images, it is more challenging because of the dynamic relationships between objects and the temporal dependencies between frames allowing for a richer semantic interpretation. In this paper, we propose Spatial-temporal Transformer (STTran), a neural network that consists of two core modules: (1) a spatial encoder that takes an input frame to extract spatial context and reason about the visual relationships within a frame, and (2) a temporal decoder which takes the output of the spatial encoder as input in order to capture the temporal dependencies between frames and infer the dynamic relationships. Furthermore, STTran is flexible to take varying lengths of videos as input without clipping, which is especially important for long videos. Our method is validated on the benchmark dataset Action Genome (AG). The experimental results demonstrate the superior performance of our method in terms of dynamic scene graphs. Moreover, a set of ablative studies is conducted and the effect of each proposed module is justified. Code available at: https://github.com/yrcong/STTran.