Git it Right

Computer Vision Recognized Pone Net
Pone Net with real person
User Interface

Motivation

Thanks to You-Know-What, it’s been forever since you’d hit the gym. Working out at home seems great until you decipher the identity of that slouched and sagging worm-figure in the mirror. You know something’s off with your elbow when you push up, but your trainer isn’t here, and you’d rather not see her in your current shape. What if some App could draw the line at your elbows flaring during a push-up?

What it does

GitItRight is a mobile application that detects and analyzes the user’s movements during workouts and provides feedback to aid improvement.

How we built it & accomplishments

Built with JavaScript within the React Native framework, GitItRight thrives in both the Android and the IOS environment. Utilizing Computer Vision’s power with TensorFlow, we construct a backbone to provide detection, analysis, and feedback for workouts like squats.

We design the GUI using BuilderX in-browser before converting it to ReactNative code. The design stems from multiple revisions of a paper prototype, which we drew using Procreate.

We are very proud of our progress because none of us had a shred of experience with Computer Vision or React Native before Hack the North. It took us less than 36 hours to go from clueless novices to semi-experts equipped to bring GitItRight to life. Blessed with this experience, we are now beyond excited to learn more about Computer Vision.

We are also delighted to say that the GUI is of professional standards. Nielsen’s 10 Usability Heuristics shine through our GUI design. We successfully emphasize the system’s close relationship with the real world, prioritize flexibility and efficiency, keep the style consistent and minimalist and the graphics aesthetically pleasing.

Challenges we face & what we learned

We chose React Native in the first place because it supports a variety of packages known for enhancing the aesthetics of the User Interface. However, these packages proved to be of inconsistent quality. Some were missing documentation; some gave rise to bugs so insidious that we found ourselves wishing for the sweet release of death. Also, while React Native is excellent for deployment across multiple platforms, its power does not come without a steep learning curve, especially when coupled with the dependency problems that arise with TensorFlow. We will be more careful in choosing which tech stacks to use in the future.

Next step

We will add more to the current collection of analyzable workouts; sit-ups and deadlifts are some of our very next steps. We also plan to explore the marvellous Computer Vision realm to enhance further the speed, sensitivity, and reliability of GitItRight’s detection and analysis of workouts. We hope to improve the user’s health and wellbeing by eliminating barriers to an active lifestyle, and we are determined to succeed.

Thank you so much for your time!