Mindlyst.ai

A collaborative effort to address a mental health issue which is a rising concern in rapidly developing world

Details about folders

• frontend:- contains frontend code
• backend:- contains backend code
• text and video api:- .rar file contains the api to the distilBERT model, .zip files contains the api to video model

---

Starting the Project (Here it contains the flow to deploy the text model on huggingface and video model on github codespaces)

1. List of Apps to download - Nodejs, ffmpeg, Docker for redis
2. Start the docker and then start the redis server (as redis required by django channels)
3. Setting up frontend => to install frontend dependenceis and start vite server

• cd frontend
• npm i
• 'npm run dev`

4. Setting up backend
   • cd backend
   • create a .env file and insert these variables into it
     • HUGGINGFACE_REPO_NAME= insert the hugging face repo name here
     • HUGGINGFACE_API_KEY= insert the read access token to the above repo here
     • GITHUB_CODESPACE_VIDEO_API= insert the api url of video running on github codespace here

• pip install -r requirements.txt => to install all dependencies
• daphne -b 0.0.0.0 -p 8000 backend.asgi:application => to run the django asgi server

---

Text API

• if using it on the hugging face then ensure your codespace is actively running
• create requirements.txt file, add modules to it, and add commit as init! indicating run this file before running any other code

---

Video API

• if using it on github codespaces then can do following
• Commands
  • docker-compose build => to build images. not start container
  • docker-compose up => to build images + start container.
  • docker-compose down => stop and remove container, volume, network
  • docker-compose stop => stop running container. but container, volume, network intact
  • docker-compose start => to start previously stop containers without rebuilding.
• Now make the port public for the api to access requests

Datasets Used

1. https://huggingface.co/datasets/Kanakmi/mental-disorders
2. emotion dataset - kaggle
3. Reddit mental health dataset