What is Deep Learning and why it is important for this chapter?
Deep learning is a subfield of machine learning that involves neural networks with multiple layers (deep neural networks). It’s particularly powerful in handling complex tasks such as image processing, speech recognition, natural language processing, and more. Handling images is exactly what we will do here.
We’ll look at images – instead of tabular data like in the past chapters. This time we have images with clothes and want to predict a label what is on this image.The project what we’ll do is a classification project (here: multi-class classification). We want to predict if an image belongs to one of ten different clothing categories.
Use Case: We have a website and the user wants to create a listing in the fashion category, so he wants to sell a t-shirt for example. That means he uses this website and uploads a picture and there is a fashion classification service. This service will get this picture and it will reply with a suggested category (here: t-shirt).
This classification service will contain a neural network which will look at the image and predict a category for this image. The dataset is a clothing dataset (https://github.com/alexeygrigorev/clothing-dataset) with 5.000 images of 20 different classes. But we’ll use a subset of this dataset which contains 10 most popular classes (https://github.com/alexeygrigorev/clothing-dataset-small).
This time there already train, validate, and test folders with images of this 10 classes we’re interested in, so we don’t have to do train-test split by our own. We just need to clone the repo.
!git clone git@github.com:alexeygrigorev/clothing-dataset-small.git
You can find the theory behind that neural networks here: https://cs231n.github.io
knowMLedge.com 