For those who doesn't know Fast.ai, It's an AI library built on top of PyTorch and it really changed the game. With a few lines of code you can build a running Machine Learning model that can produce a high accuracy.
You first need to install pytorch framework using pip install torch and then you need to install Fast.ai library using pip install fastai and you are ready to go.
You only need to import these two classes
from fastai.vision import *
from fastai.metrics import *
Fast.ai took binary classification problems into a whole new level with Transfer Learning using a pretrained model we can build a binary classifier with a few amount of data and produce a very decent results. In this notebook, I tried to implement a very basic binary classification problem "Cat or dog" and using a small dataset from Kaggle and ResNet34 model and only 10 epoches, I was able to get build a model with very good numbers.
Another kind of Classifiers in which we try to build a model that can distinguish diffrent kinds of Bears baised on a dataset we have collected from Google Images. Using The same model from the previous notebook (ResNet34) and again a very decent numbers with only couple of minutes of training.
P.S: This notebook is just the "Hello World" of the library