Wats'on the roof
Managing large scale rescue operations, reducing response times, saving lives
HackMIT 2018
Roof? Wats'on?
- Emotions are strong feelings that grow from different circumstances and situations.
- StacksFX harnesses those emotions and detects your mood in real-time.
- And from that, it recommends music to you. Getting better each time, by learning from you.
How it Works?
- StacksFX uses the camera installed in the laptops to click over 16 frames of the user in one go.
- These images get classified and help in detecting mood.
- On the basis of the mood, songs get recommended.
Browser (Client Side)
Server API
if classified as happy
then StacksFX recommends a song tagged "happy" using Spotify's API
If classified as sad:
then StacksFX
recommends a song tagged "sad" using Spotify's API
Cascade Classifier
How it Works?
- With every recommendation, there is a feedback loop; that learns preferences song by song, constantly building a custom profile of the user.
- Recommends better songs to the user, inching closer to that perfect song.
My emotion is my emotion, none of your emotion !
Feedback System
Under the hood
- StacksFX uses WebRTC to click images
- Cascade classifier of OpenCV for emotion detection
- Spotify API for song recommendations and parsing song metadata
- Django for the backend
- CSS + Bootstrap + Js for frontend
More Features to be added
- Wider range of emotions detected like anger, surprise.
- Improving the feedback system.
- More accessibility in the future - like chrome extensions, packages, web.
- Refactoring code - making it reusable for others.