Human Object Interaction Anticipation

Human-Object Interaction (HOI) Anticipation is a subtask in the field of computer vision that aims to predict a series of potential human-object interactions that may occur in the future within videos. The core objectives of this task include the precise localization of the interaction subjects (i.e., humans) and target objects (i.e., objects), as well as the accurate classification of interaction types. HOI Anticipation holds significant value in applications such as intelligent surveillance, human-computer interaction, and autonomous driving, enhancing the predictive capabilities and response speed of systems.