SRR-LGR: Local–Global Information-Reasoned Social Relation Recognition for Human-Oriented Observation

People’s interactions with each other form the social relations in society. Understanding human public space is of great importance for supporting administrations. Recognizing through visual data captured by remote sensing cameras one most efficient ways to observe a space. Generally speaking, persons same scene tend know other, and between person pairs are strongly correlated. The information which people interact also important cues relation recognition (SRR). existing works have not explored correlations people’s interactions. has only been extracted on simple level high semantic features support understanding lacking. To address this issue, we propose structure-aware local–global model SRR exploit high-level global where structure explored. In our proposed model, graph neural networks (GNNs) employed reason (local information) contextual contained constructed scene-relation graph. Experiments demonstrate that information-reasoned (SRR-LGR) can information. Further, results final show method outperforms state-of-the-art methods. addition, further discussed whether contributes equally different scene, exploiting an attention mechanism model. Further applications human-observation exploited.