Project Mirror

Slides

https://docs.google.com/presentation/d/1gj5gYn_p6PWtCXrmfrJtWsRWmWXATwOZ5Hneg7pLkvk/edit?usp=sharing

Inspiration

Unhealthy diet and lack of physical activity are leading global risks to health. Consuming low amounts of healthy foods, such as whole grains, and too much unhealthy foods, including sweetened beverages, account for one in every five deaths globally.

Just like a Mirror, our app aims to reflect the current You and help you achieve the best version of You.

What it does

Our app has 3 main functions: The food tracker, exercise tracker and personalised exercise+diet regime. The food tracker utilizes Computer Vision to recognize the food taken and outputs its calories. The exercise tracker tracks exercises and the muscles worked. The personalised regime (fitness and diet) employs an algorithm with the users’ fitness level and other information to give the user an optimised training routine and diet plan.

How is it different from current apps in the market

Majority fitness apps on the market focuses only on fitness. Such apps are designed more as a ‘Single Player” app to. Our app on the other hand focuses on not only fitness but also diet. Additionally, we leverage the power of competition as a source of motivation with habit formation as our end goal by deploying a leaderboard (and rewards). Such an approach greatly helps in giving less self-disciplined users a reason to exercise and eat healthily via social pressure when they see their friends climbing the ranks. Additionally, our algorithmically generated personalised plan give users an idea of where to start if they do not have their own routines.

How we built it

Currently, our prototype is made using Figma. However the actual app would be deployed using Flutter so that it is compatible with a wide range of devices. We used Google's TensorFlow library and the Convolutional Neural Network currently used is ResNet, and we deployed the model onto a Telegram bot for demonstration purposes.

Challenges we ran into

The team has little experience with Flutter, hence we decided to first prototype our app using Figma. Additionally, we needed to think of how to differentiate our app from the ones currently in the market while meeting the SDG goal.

Accomplishments that we're proud of

Finishing a working prototype of the app and coding a working Telegram bot to deploy our Computer Vision model.

What we learned

Although we did not use Flutter for our prototype (the actual app would be coded via Flutter), we still managed to learn the basics of Flutter. Additionally, we also learnt various Google Cloud technologies and how we could apply it into our app.

What's next for Project Mirror

We plan to deploy the app with Flutter. In the future, we could also store our data on the Google Cloud Database and use Google Cloud’s new unified ML platform Vertex AI to deploy our model.

Built With

• figma
• google-collab
• telegram-api
• tensorflow