A Graph-Based Approach to Recognizing Complex Human Object Interactions in Sequential Data

The critical task of recognizing human–object interactions (HOI) finds its application in the domains surveillance, security, healthcare, assisted living, rehabilitation, sports, and online learning. This has led to development various HOI recognition systems recent past. Thus, purpose this study is develop a novel graph-based solution for purpose. In particular, proposed system takes sequential data as input recognizes interaction being performed it. That is, first all, pre-processes by adjusting contrast smoothing incoming image frames. Then, it locates human object through segmentation. Based on this, 12 key body parts are identified from extracted silhouette skeletonization technique called foresting transform (IFT). three types features extracted: full-body feature, point-based features, scene features. next step involves optimizing different using isometric mapping (ISOMAP). Lastly, optimized feature vector fed graph convolution network (GCN) which performs classification. performance was validated benchmark datasets, namely, Olympic Sports, MSR Daily Activity 3D, D3D-HOI. results showed that model outperforms existing state-of-the-art models achieving mean accuracy 94.1% with 93.2% 89.6% D3D-HOI datasets.