VSGNet: Spatial Attention Network for Detecting Human Object Interactions Using Graph Convolutions

TLDR

Effective visual understanding requires detection systems that learn and exploit object interactions while analyzing objects individually. The study seeks to advance Human‑Object Interaction detection. The authors propose VSGNet, a Visual‑Spatial‑Graph Network that extracts visual features from human‑object pairs, refines them with spatial configurations, and models structural connections via graph convolutions, evaluated on the V‑COCO dataset. VSGNet achieves an 8 % (4 mAP) improvement over state‑of‑the‑art methods.

Abstract

Comprehensive visual understanding requires detection frameworks that can effectively learn and utilize object interactions while analyzing objects individually. This is the main objective in Human-Object Interaction (HOI) detection task. In particular, relative spatial reasoning and structural connections between objects are essential cues for analyzing interactions, which is addressed by the proposed Visual-Spatial-Graph Network (VSGNet) architecture. VSGNet extracts visual features from the human-object pairs, refines the features with spatial configurations of the pair, and utilizes the structural connections between the pair via graph convolutions. The performance of VSGNet is thoroughly evaluated using the Verbs in COCO (V-COCO) dataset. Experimental results indicate that VSGNet outperforms state-of-the-art solutions by 8% or 4 mAP.