Human Activity Analysis with Visual/Inertial Sensors
Human activity analysis is the process of analyzing/categorizing different activities (exercise, dancing etc.) performed by humans through raw data obtained from different types of sensors. We are working on novel human activity analysis techniques, where data from visual sensors (e.g. Kinect) and inertial sensors (e.g. smartphone sensors) are combined through sensor fusion, and analyzed through state-of-the-art machine learning algorithms. Currently, we have a working system where only visual data is used through Microsoft Kinect (Paper: N.M. Khan et al. A Visual Evaluation Framework for In-Home Physical Rehabilitation. IEEE International Symposium on Multimedia (ISM 2014): http://ieeexplore.ieee.org/document/7033026/). The student will evaluate alternatives to Kinect and avenues to fuse inertial sensor data obtained through smartphones.This will entail product research on latest developments on visual and inertial sensors. The student will also perform literature review on latest sensor fusion techniques particularly suited for human activity analysis. Once literature review is performed, the student will work on implementing avenues to collect raw data from suitable sensors and integrating the data with our current algorithm/ improving the current algorithm through novel fusion techniques. The student will be working with graduate students working on the same project.
Perform product research on visual/inertial sensors, literature review on sensor fusion techniques, implementation and integration of sensor data collection/sensor fusion algorithm with the existing codebase. The student is expected to perform considerable amount of software development.
Student should have strong programming experience (solid understanding of materials covered in: COE428, COE328, COE318, additional programming courses preferred). 3rd year (or higher) students will get preference.
Naimul Mefraz Khan : Human Activity Analysis with Visual/Inertial Sensors | Tuesday March 21st 2017 12:59 PM