A Novel Heteromorphic Ensemble Algorithm for Hand Pose Recognition

Imagining recognition of behaviors from video sequences for a machine is full challenges but meaningful. This work aims to predict students’ behavior in an experimental class, which relies on the symmetry idea reality annotated centered feature space. A heteromorphic ensemble algorithm proposed make obtained features more aggregated and reduce computational burden. Namely, deep learning models are improved obtain vectors representing gestures frames classification optimized recognition. So, symmetric realized by decomposing task into three schemas including hand detection cropping, joints extraction, gesture classification. Firstly, new detector method named YOLOv4-specific tiny (STD) reconstituting YOLOv4-tiny model, could produce two outputs with some attention mechanism leveraging context information. Secondly, efficient pyramid squeeze (EPSA) net integrated EvoNorm-S0 spatial pool (SPP) layer joint position Lastly, D–S theory used fuse classifiers, support vector (SVM) random forest (RF), mixed classifier S–R. Eventually, synergetic effects our shown experiments self-created datasets high average accuracy 89.6%.