Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

OpenPSG: Open-set Panoptic Scene Graph Generation via Large Multimodal Models

Zijian Zhou, Zheng Zhu, Holger Caesar, Miaojing Shi

TL;DR

This paper focuses on the task of open-set relation prediction integrated with a pretrained open-set panoptic segmentation model to achieve true open-set panoptic scene graph generation (OpenPSG), and is the first to propose the open-set PSG task.

Abstract

Panoptic Scene Graph Generation (PSG) aims to segment objects and recognize their relations, enabling the structured understanding of an image. Previous methods focus on predicting predefined object and relation categories, hence limiting their applications in the open world scenarios. With the rapid development of large multimodal models (LMMs), significant progress has been made in open-set object detection and segmentation, yet open-set relation prediction in PSG remains unexplored. In this paper, we focus on the task of open-set relation prediction integrated with a pretrained open-set panoptic segmentation model to achieve true open-set panoptic scene graph generation (OpenPSG). Our OpenPSG leverages LMMs to achieve open-set relation prediction in an autoregressive manner. We introduce a relation query transformer to efficiently extract visual features of object pairs and estimate the existence of relations between them. The latter can enhance the prediction efficiency by filtering irrelevant pairs. Finally, we design the generation and judgement instructions to perform open-set relation prediction in PSG autoregressively. To our knowledge, we are the first to propose the open-set PSG task. Extensive experiments demonstrate that our method achieves state-of-the-art performance in open-set relation prediction and panoptic scene graph generation. Code is available at \url{https://github.com/franciszzj/OpenPSG}.

OpenPSG: Open-set Panoptic Scene Graph Generation via Large Multimodal Models

TL;DR

This paper focuses on the task of open-set relation prediction integrated with a pretrained open-set panoptic segmentation model to achieve true open-set panoptic scene graph generation (OpenPSG), and is the first to propose the open-set PSG task.

Abstract

Panoptic Scene Graph Generation (PSG) aims to segment objects and recognize their relations, enabling the structured understanding of an image. Previous methods focus on predicting predefined object and relation categories, hence limiting their applications in the open world scenarios. With the rapid development of large multimodal models (LMMs), significant progress has been made in open-set object detection and segmentation, yet open-set relation prediction in PSG remains unexplored. In this paper, we focus on the task of open-set relation prediction integrated with a pretrained open-set panoptic segmentation model to achieve true open-set panoptic scene graph generation (OpenPSG). Our OpenPSG leverages LMMs to achieve open-set relation prediction in an autoregressive manner. We introduce a relation query transformer to efficiently extract visual features of object pairs and estimate the existence of relations between them. The latter can enhance the prediction efficiency by filtering irrelevant pairs. Finally, we design the generation and judgement instructions to perform open-set relation prediction in PSG autoregressively. To our knowledge, we are the first to propose the open-set PSG task. Extensive experiments demonstrate that our method achieves state-of-the-art performance in open-set relation prediction and panoptic scene graph generation. Code is available at \url{https://github.com/franciszzj/OpenPSG}.
Paper Structure (34 sections, 5 equations, 4 figures, 8 tables)

This paper contains 34 sections, 5 equations, 4 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Panoptic Scene Graph Generation
  4. Open-set Scene Graph Generation
  5. Large Multimodal Models
  6. Task Definition
  7. Method
  8. Object Segmenter
  9. Patchify Module.
  10. Pairwise Module.
  11. Relation Query Transformer
  12. Pair Feature Extraction Query.
  13. Relation Existence Estimation Query.
  14. Multimodal Relation Decoder
  15. Generation Instruction.
  16. ...and 19 more sections

Figures (4)

  • Figure 1: The left image is the input to our OpenPSG, the middle one displays the panoptic segmentation result, and the right one shows the predicted relations between objects. Our method can predict both known (close-set) relations, e.g., (0_person, playing, 1_skateboard), (2_person, looking at, 0_person), and unknown (open-set) relations, e.g., (0_person, pop shove-it, 1_skateboard), (2_person, recording, 0_person).
  • Figure 2: The overall framework of our OpenPSG, which comprises three components: object segmenter, relation query transformer and multimodal relation decoder.
  • Figure 3: Visualization results produced by our OpenPSG. The left image is the input to our OpenPSG, the middle one displays the panoptic segmentation result, and the right one shows the predicted relations between objects.
  • Figure 4: $\theta$ in the selector of RelQ-Former.