The problem StreamConnect solves

With Streamconnect, we aim to build an all in one video calling experience for blind people so that they can communicate in an effective way via the internet.
Streamconnect is a one to one video calling platform with a distinct feature of dynamically playing background music based on the expression of the other participant so that it can help the visually impaired person to detect the emotion of the other person during online conversation. There is also real time chat for the normal person and the messages will be converted to speech for the visually impaired person.

Challenges we ran into

We implemented WebRTC with Machine learning(FaceAPI) for the first time with emotion detection. Due to lack of resource we had to use free STUN server for deployed version which I found after many hours of Google search. It Had to be deployed in heroku cause Azure app service (Our favourite) does not support websockets.

The Website only runs in chrome due to restriction of WebRTC support in other browers, due to that we had to make it desktop only for now(we'll work on cross platform support after the hackathon). We faced problems with the responsiveness of WebRTC video due to different aspect ratios of in diffrenet systems.