Sober-Sense is a solution that combines both facial recognition and audio recognition technology to determine an individual's level of sobriety. This allows us to check if they are fit to drive a car. If they are not fit to drive it will redirect the user to Uber app or Ola app.
Sober-Sense is a fully responsive web-app with two tests to detect sobriety:
- First is a face test, we trained this fully from scratch on a dataset of both drunk and sober people.
- Second we analyzed spectrograms & used an algorithm called levenshtein distance to be able to detect whether a user is slurring.
- It used flask which is a a web application framework written in python, Flask helps in implementing a Machine Learning app in python that can be easily plugged, extended and deployed as a web application.
- It uses several modules like
os,Torchvision,numpy,pydub,torch.nn.module,PIL>IMAGE,Shutil,torch.optim
- Python
- Pytorch
- JavaScript
- Bootstrap
- HTML
- CSS
Note : On phone it's taking picture and video of user in realtime, I am showing it in laptop so here we uploaded the pic and video