PROJECT TITLE :

Exemplar-Based Recognition of Human–Object Interactions

ABSTRACT:

Human action can be recognized from one still image by modeling human-object interactions (HOIs), which infers the mutual spatial structure info between human and the manipulated object as well as their appearance. Existing approaches rely heavily on accurate detection of human and object and estimation of human cause; they're thus sensitive to giant variations of human poses, occlusion, and unsatisfactory detection of small size objects. To overcome this limitation, a completely unique exemplar-based mostly approach is proposed during this paper. Our approach learns a set of spatial cause-object interaction exemplars, that are probabilistic density functions describing spatially how a person is interacting with a manipulated object for various activities. Specifically, a new framework consisting of an exemplar-based mostly HOI descriptor and an associated matching model is formulated for robust human action recognition in still pictures. In addition, the framework is extended to perform HOI recognition in videos, where the proposed exemplar illustration is employed for implicit frame choice to negate irrelevant or noisy frames by temporal structured HOI modeling. In depth experiments are dispensed on two image action datasets and two video action datasets. The results demonstrate the effectiveness of our proposed strategies and show that our approach is ready to achieve state-of-the-art performance, compared with many recently proposed competitors.