Team Name
SimpleStance
Timeline
Fall 2021 – Spring 2022
Students
- Rithik Kapoor
- Dhruva Malik
- Sri Subhash Pathuri
Abstract
During the pandemic, majority of the working population spent most of their time doing sedentary desk jobs. As a result, they have developed inadequate spine posture, which can cause profound implications such as back pain if not treated at an early stage. Implementation of machine learning and computer vision has allowed us to capture and analyze the video input from the user’s phone to remind him whenever he is not in the correct form based on the degree of severity.
Background
The spine is one of the most essential parts of the body. Without it, we cannot keep ourselves upright or stand up. It gives us body posture and support. It allows us to move freely and is designed to protect the spinal cord. Without the spinal cord, we cannot move even an inch; hence keeping the spine healthy is vital if you want to live a happy life. This is why we want to make a profound impact on everyone’s lives by developing an innovative product that will be immensely beneficial. Although there are other posture correction applications on the Google play store, almost all of them involve using some kind of expensive tracker that is placed on the body and don’t seem to use machine learning and computer vision, without a third object.
Project Requirements
- Users must be able to securely authenticate and create an account in the application.
- Users should be able to reset their password and modify their account details.
- Once. the user successfully logins, they should be alerted when their posture gets into a bad form.
- The users should be able to see an analysis of their performance over all the sessions in a week in the useranalytics page
System Overview
The structure of the posture detection application is comprised of three primary layers which are the
Application graphical user interface, server back-end, machine learning model, and the database. The
first layer which is the application graphical interface layer mainly comprised of the user interaction
and user experience part of the application. This layer will consist of multiple screens in this application facilitating the smooth transition between a variety of functionalities this application has to offer. It will
have the login, registration, posture detection and user statistics screens.
Secondly, the machine learning model is connected to the server back-end and the app GUI as seen
in the diagram above. The role of this model is to give better coordinates for the real time picture
captured and sent to this system through the application GUI.This model is responsible for collection
of the user statistics which show the amount of time spent by the user in bad and good posture. It is
also essential for analyzing the real time footage captured and is integrated into the application. The
image recognition forms as main component of the model. Further, the server back-end layer implements the back-end architecture that consists of a query manager and a database controller. It also acts
as an interface between the application GUI and the database. Further, the database layer servers as
the central storage for the application. It consists of two major storage buckets, the user table and the
user statistics table.
Results
Demo Video:
Future Work
- We would like to provide premium services with a slight cost to provide expert advice from health care professional as an all in one platform.
Project Files
Project Charter: Project Charter: https://drive.google.com/file/d/13iGlrXSY70E-f7P_N8VGjLQ0HI8iGtLZ/view?usp=sharing
System Requirements Specification : https://drive.google.com/file/d/1LHd9h7K7dejtASaXNM0qKf8xqvVGX59Y/view?usp=sharing
Architectural Design Specification (link): https://drive.google.com/file/d/1aLVWoYfYWY4AYePT5iz3y8V74FZfahcb/view?usp=sharing
Detailed Design Specification: https://drive.google.com/file/d/1npMyWZfmHfu2GSe_c2rGNWsvPXl2qQO6/view?usp=sharing
References
- Valentin Bazarevsky, Ivan Grishchenko, Karthik Raveendran, Tyler Zhu, Fan Zhang, and Matthias Grundmann. 2020. Blazepose: On-device real-time body pose tracking. arXiv preprint arXiv:2006.10204 (2020).
- Reem S AlOmar, Nouf A AlShamlan, Saad Alawashiz, Yaser Badawood, Badr A Ghwoidi, and Hassan Abugad. 2021. Musculoskeletal symptoms and their associated risk factors among Saudi office workers: a cross-sectional study. BMC Musculoskeletal Disorders 22, 1 (2021), 1–9
- European Foundation for the Improvement of Living, Working Conditions, et al. 2020. Living, Working and COVID-19—First Findings—April 2020. (2020).
