User Tools

Site Tools


projects:year8:8a.024.sbu

8a.024.SBU - Movement/Mobility Visual Analytics from Sensor Data - Phase II

Project - Summary

Wearable biosensors, such as those embedded in smart phones, can provide data to assess neuro-motor control in mobile settings, at homes, schools, workplaces and clinics.

In this study, we proposed a method to estimate the lower human body pose (estimating the 3D joint locations) by sensing feet movement using a smart shoe device, without the involvement of any visual sensing device such as a camera. This project utilized Zeblok smart shoes equipped with 4 inertia measurement units (IMUs) on the toe, left ball, right ball and heel and the linear accelerometer inside the insole interfaced to the Bio-Informatics Cloud to track the limb movement. We trained an LSTM model to predict the skeletal pose of a subject on eight different actions. Experiments show that the model accurately predicts the 3D skeletal pose using sensor data as input for certain actions.

Project - Team

Team Member Role Email Phone Number Academic Site/IAB
Dimitris Samaras PI samaras@cs.stonybrook.edu (631) 632-8464 Stony Brook University
Souradeep Chakraborty Researcher souchakrabor@cs.stonybrook.edu (631) 320-6641 Stony Brook University
Mouli Narayanan Project Mentor mouli.narayanan@zeblok.com Not available Zeblok
Funded By: Zeblok

Project - Novelty of Approach

Several methods exist for predicting 3D human body pose from single or multiple captured images such as [3,4,5]. However, smart wearable sensors aim to track biometric records at home or workplace environments, without the need for an individual to visit a unit with dedicated visual pose tracking systems, which often use expensive multi-camera apparatus. Existing methods for estimation of 3D body pose requires at least one conventional camera or a depth camera that can capture and record the video of a subject’s motion. However, none of the existing systems predict human 3D body pose from sensing feet movement only. We present a novel method to track the lower body pose of a human subject only from the sensory data from the shoe insoles.

Project - Deliverables

Deliverables
1
2
3
4

Project - Benefits to IAB

The developed system would eliminate the need of cameras for tracking body pose. Also, the system would allow us to easily track a subject’s activities over a time period and thus provide better healthcare. We presented a demo of the work at the CVDI conference meet in 2019.

Project - Documents

FilenameFilesizeLast modified
8a.024.sbu_ip_disclosure.docx24.0 KiB2020/08/17 16:11
8a.024.sbu_final_report.docx925.5 KiB2020/08/17 16:11

Life Form Feedback

Year 8 Project Poster Session Feedback (Fall 2019)
Real name Great Progress On Course Needs Change Off Course Abstain
Kimmo Valtonen (kimmo.valtonen)     
Sumit Shah (sumit.shah)     
Count:02000
projects/year8/8a.024.sbu.txt · Last modified: 2021/06/02 17:26 by sally.johnson