-
-
Data tracking
-
Capture what you read
-
Azure ML
-
Azure ML
Inspiration
While studying, almost always, I tend to drift away, or spend more time on some topics than I ideally should. While revising, sometimes I want to see stuff that I did not get right the first time, while sometimes I just want to go through stuff I already know. But no, I have to flip through each page/slide in either of the cases. This is when, we came up with the idea of, what if we could track how much we concentrated and how much time we spent, without doing anything except wearing some electrodes, or a Muse band
What it does
1)It tracks brain waves of the human body, and passes these values to a predictive model that tells you if you were concentrating or not.
2)If you were not concentrating, depending on the type of setting, there can be multiple responses from the software.
-Students : It goes through the text part of what you were studying where your concentration drops below a certain level, finds the most important keywords and suggests YouTube videos to study the similar material in more interesting sources.
-Teachers : If a teacher sees the classroom in general has dropped its concentration level, they can do something more interesting like a class activity or re-teach the last taught topic in a different way that engages students better.
-Special Students : Special students sometimes have a very low attention span, and it is difficult to keep track and keep them attentive for longer time spans. If a teacher gets to track brain waves for the students, she can immediately know what's wrong and help the student get back to the class.
-In general : The idea is that it the application learns about you over time. It sees how you react at the brain-wave level to different kinds of scenarios that you are put in. The next time it sees you losing concentration, it can easily suggest you a video you would love to watch, just to take a break and then help you get back to your work. These waves are being also worked upon to see if emotions can be captured from them. If that does happen sometime, the app would know what you would love to watch when you are feeling low, or your favorite sports team when you need the motivation.
How we built it
We used the Muse Headband to get the brain waves. We connected the device to an android app that we made, which records the brain waves and saves it on the Azure database. The software also tracks the slide number you are on using HTML/javaScript and exports another database and saves it on the Azure database. On the 2 Azure database, we then run a script to easily merge the 2 tables on the time stamp values. Once we had this database ready, we had our teammate sit for a while generating the test data. We then used Azure ML Studio to train our algorithm. We used a Decision Tree model to train our data on. We got an accuracy of 82% on our system, which we believe is pretty good as moving forward we will have a better/bigger data set developed, probably on a strong algorithm, with more features to get a better accuracy.
Challenges we ran into
1) Connecting the Muse Headbands to any of our systems was not easy. We faced problems with MacOS, Windows and iOS compatibility, although we followed all steps. Eventually we used Android and got it to work.
2) Bluetooth on the Android devices has a lot of issues when trying to connect the Museband because of interference from other bluetooth devices. We had no work around this except to work in isolated spaces
Accomplishments that we're proud of
1)We are super proud of the fact that we could get a 82% accuracy on the model as we expected it to be around 77%, the same as what a researcher who we followed for this project got a year back.
2) We are also proud about how we were able to get an amazing sync when it came to page number being viewed and time stamp being generated, as they were better matched than we expected them to be.
3) The scalability of this app - We can have a huge impact on a very large population, and nothing matters to us more than this fact.
What we learned
1) A lot of new technologies - Android, Muse API, Azure API and the list goes on - Credits - stackoverflow!
2) Science behind the brainwaves and how they can be used to calculate concentration levels. We also learnt a lot about the possibility of deciphering emotions using these apps.
What's next for Brain Tracker
1) We would love to integrate the recommendations idea. We actually have the plan ready, and we know Azure is going to make our lives easier on this. We just need to get a bigger dataset, and we probably will need to do this ourselves.
2) Making the band more personalized would be a really cool feature that we are currently eyeing too.
3) As stated before, research is on and some researchers claim it is possible to get emotions from the same brain waves too. I would love to find out if that is true, because that would add a whole new dimension to Brain Tracker
Log in or sign up for Devpost to join the conversation.