Rectif.ai | Devpost

Posture detection and notification

Introduction

Our app helps develop good posture habits by making people mindful about their postures when they are in front of the screen.

Inspiration

In today’s world, our work requires us to be sitting for a lot of the time. Failing to maintain good posture can thus, cause a variety of health problems related to our lungs, digestion, bones, and so on and so forth. Furthermore, bad posture habits are pretty common and it can be difficult breaking away from them without much deliberate effort.

What it does

The app helps develop good posture habits by making people mindful about their postures when they are in front of the screen.

How We built it

We applied Pytorch implementation of PoseNet model to detect key-points of the user. The pose keypoints are then passed onto another network that classifies it into good or bad posture. Because of the unavailability of the posture(good/bad) dataset, We had to create our own dataset for posture recognition. With the pose estimation model, We collected 150,000 different pose coordinates with labels by sitting in both good and bad postures.

Challenges we ran into

There was no proper bad posture dataset so chose a different domain problem and tweaked into our use case.
The infrastructure needed for best results. We didn't have good machines so used Mobilenet based Posenet and Linear Neural Networks for detection because of its speed and accuracy compared to the size of models and infrastructure required.

Accomplishments that we are proud of

To solve a problem which apparently had no dedicated dataset, we were able to propose a novel approach and the solution by creating our own custom dataset. Building the system that works despite having few data and resources is the accomplishment that we are very proud of.