OED: Towards One-stage End-to-End Dynamic Scene Graph Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Guan, Li, Zhimin, Chen, Qingchao, Liu, Yang
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910460261957632
author Wang, Guan
Li, Zhimin
Chen, Qingchao
Liu, Yang
author_facet Wang, Guan
Li, Zhimin
Chen, Qingchao
Liu, Yang
contents Dynamic Scene Graph Generation (DSGG) focuses on identifying visual relationships within the spatial-temporal domain of videos. Conventional approaches often employ multi-stage pipelines, which typically consist of object detection, temporal association, and multi-relation classification. However, these methods exhibit inherent limitations due to the separation of multiple stages, and independent optimization of these sub-problems may yield sub-optimal solutions. To remedy these limitations, we propose a one-stage end-to-end framework, termed OED, which streamlines the DSGG pipeline. This framework reformulates the task as a set prediction problem and leverages pair-wise features to represent each subject-object pair within the scene graph. Moreover, another challenge of DSGG is capturing temporal dependencies, we introduce a Progressively Refined Module (PRM) for aggregating temporal context without the constraints of additional trackers or handcrafted trajectories, enabling end-to-end optimization of the network. Extensive experiments conducted on the Action Genome benchmark demonstrate the effectiveness of our design. The code and models are available at \url{https://github.com/guanw-pku/OED}.
format Preprint
id arxiv_https___arxiv_org_abs_2405_16925
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle OED: Towards One-stage End-to-End Dynamic Scene Graph Generation
Wang, Guan
Li, Zhimin
Chen, Qingchao
Liu, Yang
Computer Vision and Pattern Recognition
Dynamic Scene Graph Generation (DSGG) focuses on identifying visual relationships within the spatial-temporal domain of videos. Conventional approaches often employ multi-stage pipelines, which typically consist of object detection, temporal association, and multi-relation classification. However, these methods exhibit inherent limitations due to the separation of multiple stages, and independent optimization of these sub-problems may yield sub-optimal solutions. To remedy these limitations, we propose a one-stage end-to-end framework, termed OED, which streamlines the DSGG pipeline. This framework reformulates the task as a set prediction problem and leverages pair-wise features to represent each subject-object pair within the scene graph. Moreover, another challenge of DSGG is capturing temporal dependencies, we introduce a Progressively Refined Module (PRM) for aggregating temporal context without the constraints of additional trackers or handcrafted trajectories, enabling end-to-end optimization of the network. Extensive experiments conducted on the Action Genome benchmark demonstrate the effectiveness of our design. The code and models are available at \url{https://github.com/guanw-pku/OED}.
title OED: Towards One-stage End-to-End Dynamic Scene Graph Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2405.16925