Human Activity Recognition Using Significant Skeletal Joints

Review Of Literature

The existing approaches based on conventional Local Spatio-Temporal (LST) are suffered from various challenges like dynamic background and illumination. To enhance this problem, the work of Zhang and Parker (2016), provides a multi-dimensional colour-depth LST feature-based feature detector technique to represent various features such as shape, pose variation, texture with local maxima as a region of interest. The authors have also used a support vector machine (SVM) with feature representation to build an effective action detection system. The study uses different standard datasets to demonstrate the effectiveness of the presented technique.