Our Website

• BAD DATA HEALTH Site is currently not operating

---

Built With

• FastAI - Deep Learning library for Pytorch
• create-react-app - Used to build our front end
• Flask - Used to serve as REST API server to communicate our model and front-end
• Gunicorn - Used to handle our HTTP server
• NGINX - Used to handle out HTTP proxy server

---

Train your own model (Tutorial)

FastChest - Tutorial using Jupyter Notebook to train using on Google's Colab. No need to install on your computer. You will need a Kaggle account to download the data.

---

Deploy on your computer

Instalation

Install anaconda 3 in your system https://www.anaconda.com/products/individual

Then create new environment

conda create --name fast python=3.6

Activate the environment

conda activate fast

Then install packages

conda install pytorch=1.4 torchvision -c pytorch
conda install -c fastai fastai
conda install flask gunicorn

---

Test run

python model.py sample/NORMAL2-IM-1427-0001.jpeg | python model.py sample/person1947_bacteria_4876.jpeg

---

Changing the paths

You can change the paths of where the model is located And where the predictions will be sent by changing the following in model.py

model_path = 'models/'
output_path = 'predictions/'

make sure that the path of the models always contains the file "dense"

---

Deployment

gunicorn --bind 0.0.0.0:5000 wsgi:app

---

Contributors

• Luis Chumpitaz - Deep learning and Cloud Deployment
• Eduardo Ramos - Back and Front End Framework with React Development
• Jose Pascual - Front End Development
• Brooke Porter - Deep Learning (YOLOv3) Segmentation Post-Processing