Revisiting Scene Graph Generation from the Perspective of Detector-Conditioned Reachability
For researchers in scene graph generation, this work provides a principled way to combine two dominant paradigms, leading to consistent improvements across multiple benchmarks.
The paper identifies complementary predictive behaviors between detector-based and query-based scene graph generation methods, and proposes a Dual-SGG method that combines both via a dual-query design, achieving state-of-the-art results on Visual Genome, Open Images v6, and GQA-200.
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.