ARYS | Devpost

Frontend

Inspiration

Seeing the next generation of people suffer from something that I would never wish even on my worst enemy, is a very bad feeling. Seeing one of my close friends succumb to depression, my friends and I decided to do something about it and try to bring awareness and do our best to help people who are reserved with their feelings.

How we built this

Our initial attempt is by keeping track of their emotional state at a point of time by leveraging emotion recognition by Facial and Conversational cues. If our algorithm feels that the person is sad/in distress, we initially intimate people who are closest to them and then we suggest options/activities which we believe would best help the person at that situation or even help them open up.

How we implemented it

Our project is based on powerful models of complex neural networks which help us predict the emotional quotient of a person with precise certainty (Mostly PyTorch implemented) . We then use make use of the data and project it onto the fronted for the person to interact with.

Major challenges faced

Major challenges we faced were the poor quality of data we had for training and the long training times to get a good enough accuracy. As far as UX is concerned, we had difficulties choosing how to target the person without bothering them or agitating them.

Thing that we are proud of

We are proud of achieving a near State of the Art Accuracy for Facial model and for doing this in such a short time. We are also proud of the frontend we managed to conjure.

What we learnt from this

We learnt a lot about time management and presenting our idea clearly and concisely.

What's next for ASRY

We plan on extending our work even after this hackathon, as our goal is still something we would like to accomplish. We have some future works/features in our pipeline.