Skip to content

Latest commit

 

History

History
92 lines (77 loc) · 3.29 KB

README.md

File metadata and controls

92 lines (77 loc) · 3.29 KB
Raw

Tri-NIT-Hack

Problem Statement : ML01

Covine is an android app with 2 major features :

AI based Chatbot for covid 19 assistance

Covid Alert System using Graph Colouring Algorithm

Technocrats.ML01.mp4

TechStack :

  • Flutter & Firebase
  • Tensorflow for Natural Language Processing
  • Django for backend and REST API for communication

Project Structure :

  ├── README.md
├── backend
│   ├── api
│   │   ├── __init__.py
│   │   ├── admin.py
│   │   ├── apps.py
│   │   ├── data.py
│   │   ├── migrations
│   │   ├── models.py
│   │   ├── nlp.py
│   │   ├── serializers.py
│   │   ├── tests.py
│   │   └── views.py
│   ├── backend
│   │   ├── __init__.py
│   │   ├── asgi.py
│   │   ├── settings.py
│   │   ├── urls.py
│   │   └── wsgi.py
│   ├── db.sqlite3
│   ├── manage.py
│   └── requirements.txt
└── covine
    ├── README.md
    ├── android
    ├── fonts
    ├── images
    ├── ios
    ├── lib
    │   ├── components
    │   │   ├── bottom_sheet_text.dart
    │   │   ├── contact_card.dart
    │   │   └── rounded_button.dart
    │   ├── constants.dart
    │   ├── main.dart
    │   └── screens
    │       ├── about.dart
    │       ├── chatscreen.dart
    │       ├── login.dart
    │       ├── registration.dart
    │       ├── screen1.dart
    │       ├── screen2.dart
    │       ├── screen3.dart
    │       ├── tracing.dart
    │       └── welcome_screen.dart
    ├── pubspec.lock
    ├── pubspec.yaml

Chatbot Info :

We have used PreTrained Hugging face RoBERTa model and fine tuned on the covid-19 data. 
Then we served it using REST API using Django framework , on localhost .
Our android phone is connected to same network (we could not deploy because of storage limit in free hosting services like pythonanywhere , heroku ...)
User sends a question string from the app interface , the backend recieves it , forward propagates it through the model , returns the response to the request...
...which is then displayed in the mobile

App Info :

This app uses graph theory to build connection (edges) with persons(node) you came in close vicinity to. Incase of covid infection , alert is send to all immediate nodes(people) in the graph .

Feature #1 : Contact Tracing

- Uses BlueTooth Beacon to trace nearby people (offline mode)
- Displays all contact traces with time of contact , no of times contacted & location

Feature #2 : Report Covid & Alert Contacts

- An infected person declares Covid19 positive in the report portal
- All people who were in contact(share an edge with this node) are alerted as "Potential Threat"

Feature #3 : Crowd Map

- Maps people onto location and displays the headcount at the location
-Gives insight into crowd intensity