Revisiting Scene Graph Generation from the Perspective of Detector-Conditioned Reachability

2026-07-07

AI Summary

Existing scene graph generation methods lack a systematic analysis of the behavioral differences between detector-based and query-based models, which hinders further performance improvements. This work addresses this gap by first revealing their complementarity from the perspective of detector-conditioned reachability and proposes Dual-SGG, a dual-query framework that effectively integrates both paradigms. Through controlled experiments, the study validates the efficacy of the proposed fusion mechanism. The method achieves significant performance gains over current state-of-the-art approaches on the Visual Genome, Open Images v6, and GQA-200 benchmarks, establishing a new architectural paradigm for scene graph generation.

Abstract

Scene graph generation (SGG) approaches can be broadly classified into detector-based and query-based methods according to their underlying reasoning mechanisms. However, the discrepancy in their predictive behaviors, induced by these distinct mechanisms, has not been systematically analyzed. In this work, we design a controlled experimental setup to examine prediction discrepancies from the perspective of detector-conditioned reachability. The results suggest clear complementary clues. Motivated by this observation, we introduce a Dual-SGG method that consolidates both reasoning mechanisms via a dual-query design, thereby leveraging the complementary predictive behaviors of both detector-based and query-based methods. Extensive experiments on the Visual Genome, Open Images v6, and GQA-200 datasets demonstrate the effectiveness of the proposed method.

Problem

• Scene Graph Generation
• detector-based methods
• query-based methods
• prediction discrepancy
• reasoning mechanisms

Innovation

• scene graph generation
• detector-conditioned reachability
• dual-query design
• complementary reasoning
• object detection