Our paper of Virtual Environment (VE) Lab is accepted to IEEE International Conference on Systems, Man, and Cybernetics (SMC) [LINK]

Title: 

A Hybrid Approach of Neural Networks for Age and Gender Classification through Decision Fusion

Authors:

Adithya Balasubramanyam, Ashok Kumar Patil, Bharatesh Chakravarthi, Jaeyeong Ryu, Young Ho Chai

Abstract:

Sensor acquired data from inertial measurement unit based motion capture systems often do not conform to human body kinematics. Although the data may appear visually correct when reconstructed on a stick model or 3D avatar, it does not represent natural movement when visualized on 3D data analytical tools, such as Motion-sphere. This work extends the Motion-Sphere human motion data visualization and analysis tool to edit sensor acquired human motion data (SHMD) to strictly adhere to human body kinematics. We use knee and elbow joint movements from the range of motion data in the TotalCapture dataset to compare SHMD against edited human motion data. This work proposes a method to edit SHMD using Motion Sphere for kinematic correctness. We verify that edited human motion data is admissible to human body kinematics and improves virtual reconstruction on 3D avatar and joint position estimation accuracy. Our results highlight the requirement for data visualization and editing tools, such as Motion-sphere to edit and clean SHMD.