Project Overview
This project was inspired due to the strict health protocol on the COVID-19 Pandemic era. With the large availability in face mask data, I decided that it would be an interesting project to do, whilst learning the fastai API.
The model uses the ResNet34 architecture. The model is finetuned on face mask images from Kaggle Dataset titled "500 GB of images for Face Mask Detection. Part 1" . The model achieved satisfying training and validation accuracy with few epochs, due to the amount of data available.
After the training is done, the model is saved and deployed locally with the help of streamlit library.
Tools Used
Python
Pytorch
FastAI
Streamlit
Deep Learning
Transfer Learning