RelateAnything: Real-Time Open-Vocabulary Relation Prediction From Any Inputs

A new system called RelateAnything performs real-time open-vocabulary relation prediction from any inputs, according to the research announcement. The work targets a gap in scene-graph modeling, where models remain trained and evaluated on the fixed 50 or 56 predicates of a single annotation set even as open-vocabulary detection and promptable segmentation have moved the taxonomy out of the model and into the input. The system extends the open-vocabulary paradigm from detection and segmentation to relation prediction.

Open-vocabulary detection accepts any class list at inference, and promptable segmentation returns regions without class names: the taxonomy has left the model and become an input. Relation prediction has not. Scene-graph models are still trained and evaluated on the 50 or 56 predicates of one annot