We wanted to solve a problem that has become more prevalent in today’s world. With an increased dependency on technology for both work and entertainment recently, there is also an increased amount of time in people’s days sitting and working at desks while not focusing on their health. Sitting for long periods of time have adverse effects on your health and can even increase the chance of getting life threatening diseases. Not only the action of sitting, but also how you sit, can affect your health. Bad posture, such as slouching or leaning too close to the monitor, can lead to pain and eye strain.
What it does
In order to help prevent these adverse effects to people’s health, we created HealthDESK, a website that uses AI technologies and webcam data to determine and notify you when you have poor posture, and remind you to maintain healthy habits while sitting at your desk.
How we built it
Using TensorFlow and OpenCV, we created an AI algorithm that can detect unhealthy habits that occur while sitting at your desk. The AI implements a variety of Computer Vision algorithms, such as Convolutional Neural Networks called ResNet-50 and MobileNet V2, along with a Cascade Classifier. After receiving bounding boxes and landmarks from these Computer Vision algorithms, we used mathematical computations to determine if a user is drinking water, stood up, touched their face, leaned in, and has bad posture.
Challenges we ran into
One challenge we ran into early was designing the user interface so that anyone would be able to use it. We also struggled to successfully display webcam data and add push notifications. Implementing the AI to detect unhealthy habits was also a complicated process.
Accomplishments that we're proud of
We were successfully able to use webcam data and AI in order to identify and notify users of unhealthy habits at their desk. We also were able to use a variety of Computer Vision algorithms and create a multi-detection system using them. Finally, we're proud of the simplistic yet powerful UI/UX design.
What we learned
We learned how to combine Computer Vision models and create a more advanced model. Also, we learned how to read webcam data as well as communicate with the back end. By the end, we were able to figure out how to combine both the front end and the back end into one comprehensive project.
What's next for HealthDESK
We hope to expand the usage of the AI in order to provide the user with one platform that can be used to benefit their physical health while working. We also hope to work on the scalability of the project by trying to expand the project to thousands of daily users.