Human object interaction detection: Design and survey

Abstract

Human-Object Interaction (HOI) detection is the process of estimating the interaction between a human and an object in an image. The first attempts at HOI involved two-stage detection methods, then improved to one-stage, and finally end-to-end methods in the last year with the development of transformers. The objective of this paper is to first provide a guideline to researchers seeking to design a human object interaction detection model in addition to describing the benchmark datasets and evaluation metrics used in this task. Second, it presents a survey about the different existing methods for HOI detection and critically analyzes their design decisions through every step. Finally, we report the characteristics of future research directions and present some open issues on human-object interaction detection by presenting the limitations facing HOI detection. © 2022 Elsevier B.V.