Focus++

Inspiration

We were inspired by our personal experiences with virtual classes. We felt that it was very easy to get distracted while studying at home, and that we could definitely improve our efficiency. Thus, we came up with the idea to track if you are actually studying or not.

What it does

This application implements the Pomodoro technique for studying, but also monitors you during your work sessions to make sure you are focused and not distracted. It also provides a history of all previous sessions, where you can view your data graphically represented in a pie chart.

How we built it

• Convolutional neural network was built using Tensorflow
• Tensorflow.js used to load models and predict user focus state
• Express.js (Hosted in Node environment) used for server hosting data
• We used Ajax to send data to and from the server and pages (ReST)
• Pages were made using html, css (bootstrap)

Challenges we ran into

• Training neural network
• Acquiring data to train the network
• Manipulating data between server and local storage

Accomplishments that we're proud of

• Neural network accurate predicts user's focus state
• Intuitive user interface design
• Accessing previous data for comparisons

What we learned

• Implementing a reliable neural network
• Deploying neural network on browser with tensorflow.js
• Optimized methods to manipulate data

What's next for Focus++

• Mobile app
• More focus states
• Larger training dataset

Built With

• css3
• express.js
• html5
• javascript
• keras
• lowdb
• node.js
• python
• tensorflow