Our inspiration stems from the global pandemic. Due to COVID-19, the world is in lockdown mode, meaning people are to minimize outdoor and social activities. Many people are unable to do what they enjoy and the only form of entertainment becomes technology. Additionally, workers need to work from home, meaning even more time spent on the computer. With such prolonged sessions on computers, health concerns arise. Constant monitoring of one's posture would be helpful for one to maintain better posture throughout time.
What it does
PostureML uses machine learning to determine the type of posture that the user is in. It sends reminders to the user whenever it detects that the user is in a malicious posture and reminds them of their posture, and how to fix it. The user will be able to choose the type of reminder and their interface. Finally, there will be another section to the website, where the user can view their statistics on a graph, knowing which malicious postures they have done the most.
How we built it
Challenges we ran into
Due to our unfamiliarity with Azure, we had trouble figuring out the values that Azure needs for an API call. It wasted a lot of time, resulting in most of the time of our project going towards figuring out the API call for Azure.
Accomplishments that we're proud of
Most of us did not have experience in everything we have done. This includes working with Azure, routing with React, and working with CustomVision. We are proud to incorporate all of these ideas into this project and have it working the way we want it to. We are proud to present an aesthetic and functional web-app for the users who would like for their posture to improve while being at a computer for long periods of time.
What we learned
We obtained a more thorough understanding of API calls and backend integration of API calling. We also gained more knowledge on frontend development with React and learning how to connect the information from the back end with the information to be displayed.
What's next for PostureML
The future of PostureML is bright.
PostureML has a smooth and easy-to-comprehend user interface. However, it still has room for improvement. The future for the PostureML user interface would be for the user to be able to hand-select the colours for their colour scheme. Next, the user should be able to select the ringtone they would like when receiving a notification. Other additional customizations would also be added into PostureML for the best user experience, such as a muted function, a overview for quantifying different posture types per unit time, and achievements and goals for the user.
Next, PostureML is also highly expandable. Nowadays, most laptop and phone comes with an integrated camera, and most desktop users buy a camera for work or personal purposes. Thus, expanding PostureML into phones, and desktop applications would be a good idea. These applications will be connected by an account that a user has. The data stored for the graphs will be universal between the platforms, allowing for the user to view their overall statistics on any platform they wish.
Finally, PostureML also has room for technical improvements. This includes additional classifications for different kinds of malicious posture, and even pre-detecting medical conditions such as scoliosis and lumbar lordosis. Additional classifications for the user to fix their posture, as well as additional functionalities such as recommending to see a physician would be integrated into future versions of PostureML.