Create and deploy your first Image Classification model in < 10 mins
We are living in strange times. The era of COVID has forced us into our homes and while it's a good time to catch up with our loved ones, it also opens an avenue to work on our long-pending learning goals and try out new stuff (you know, what interests us). I recently completed some of the courses on AI and have decided to share a series of posts related to it. In today’s post, we will talk about training an Image Classification model and deploy it as well.
What is FastAI?
FastAI is an open-source library for Deep Learning, built on top of PyTorch, and provides a simplified way of training Neural Networks. The library is a one-stop solution for most training needs: computer vision, text, tabular and collaborative filtering.
Let’s get started!!
We will build an Image Classification model to detect whether an image is of a Grizzly Bear, Black Bear, or a Teddy Bear 🧸 :P
We will be using Transfer Learning to get a pretty amazing pre-trained model: ImageNet, and fine-tune it to our use-case of classifying bears. To get the training & validation dataset (i.e. images of the 3 types of bears), you can do a simple Google search and get a plethora of options to choose from. Note that the images need to be separated into different…
