Generative Adversarial Networks (Gans): Complete Guide

Deep Learning and Computer Vision to implement projects using one of the most revolutionary technologies in the world!

What you'll learn

Understand the basic intuition about GANs

Generate images of digits (0 - 9) using DCGAN and WGAN

Transform satellite images into maps using Pix2Pix architecture

Transform zebras into horses using CycleGAN architecture

Transfer styles between images

Apply super resolution to improve image quality using ESRGAN architecture

Create new faces of people with high quality and definition using StyleGAN

Generate images through textual descriptions

Restore old photos using GFP-GAN

Complete missing parts of images using Boundless architecture

Generate deepfakes to swap faces with SimSwap

Requirements

Programming logic

Basic Python programming

Knowledge about neural networks is desirable, but not mandatory

Description

GANs (Generative Adversarial Networks) are considered one of the most modern and fascinating technologies within the field of Deep Learning and Computer Vision. They have gained a lot of attention because they can create fake content. One of the most classic examples is the creation of people who do not exist in the real world to be used to broadcast television programs. This technology is considered a revolution in the field of Artificial Intelligence for producing high quality results, remaining one of the most popular and relevant topics.In this course you will learn the basic intuition and mainly the practical implementation of the most modern architectures of Generative Adversarial Networks! This course is considered a complete guide because it presents everything from the most basic concepts to the most modern and advanced techniques, so that in the end you will have all the necessary tools to build your own projects! See below some of the projects that you are going to implement step by step:Creating of digits from 0 to 9Transforming satellite images into map images, like Google Maps styleConvert drawings into high-quality photosCreate zebras using horse imagesTransfer styles between images using paintings by famous artists such as Van Gogh, Cezanne and Ukiyo-eIncrease the resolution of low quality images (super resolution)Generate deepfakes (fake faces) with high qualityCreate images through textual descriptionsRestore old photosComplete missing parts of imagesSwap the faces of people who are in different environmentsTo implement the projects, you will learn several different architectures of GANs, such as: DCGAN (Deep Convolutional Generative Adversarial Network), WGAN (Wassertein GAN), WGAN-GP (Wassertein GAN-Gradient Penalty), cGAN (conditional GAN), Pix2Pix (Image-to-Image), CycleGAN (Cycle-Consistent Adversarial Network), SRGAN (Super Resolution GAN), ESRGAN (Enhanced Super Resolution GAN), StyleGAN (Style-Based Generator Architecture for GANs), VQ-GAN (Vector Quantized Generative Adversarial Network), CLIP (Contrastive Language–Image Pre-training), BigGAN, GFP-GAN (Generative Facial Prior GAN), Unlimited GAN (Boundless) and SimSwap (Simple Swap).During the course, we will use the Python programming language and Google Colab online, so you do not have to worry about installing and configuring libraries on your own machine!

Overview

Section 1: Introduction

Lecture 1 Course content

Lecture 2 Introduction to GANs

Lecture 3 How GANs work

Lecture 4 Course materials

Section 2: DCGAN and WGAN

Lecture 5 DCGAN - intuition

Lecture 6 MNIST dataset

Lecture 7 Building the generator

Lecture 8 Building the discriminator

Lecture 9 Loss (error) calculation

Lecture 10 Training

Lecture 11 Visualizing the results

Lecture 12 HOMEWORK and solution

Lecture 13 WGAN - intuition 1

Lecture 14 WGAN - intuition 2

Lecture 15 WGAN-GP - intuition

Lecture 16 Preparing the environment

Lecture 17 Wassertein loss

Lecture 18 Gradient penalty

Lecture 19 Training 1

Lecture 20 Training 2 and visualization

Lecture 21 HOMEWORK and solution

Section 3: cGAN - Pix2Pix and CycleGAN

Lecture 22 cGAN - intuition

Lecture 23 Pix2Pix - intuition

Lecture 24 Map dataset

Lecture 25 Preprocessing the images 1

Lecture 26 Preprocessing the images 2

Lecture 27 Loading the data

Lecture 28 Building the generator 1

Lecture 29 Building the generator 2

Lecture 30 Building the generator 3

Lecture 31 Building the discriminator 1

Lecture 32 Building the discriminator 2

Lecture 33 Generating the images

Lecture 34 Training 1

Lecture 35 Training 2 and results

Lecture 36 Pretrained Pix2Pix with PyTorch

Lecture 37 Facades dataset

Lecture 38 Visualizing the results

Lecture 39 Drawing to photo 1

Lecture 40 Drawing to photo 2

Lecture 41 Night to day

Lecture 42 HOMEWORK and solution

Section 4: Additional content 1: Artificial neural networks

Lecture 43 Biological fundamentals

Lecture 44 Single layer perceptron

Lecture 45 Multilayer perceptron – sum and activation functions

Lecture 46 Multilayer perceptron – error calculation

Lecture 47 Gradient descent

Lecture 48 Delta parameter

Lecture 49 Updating weights with backpropagation

Lecture 50 Bias, error, stochastic gradient descent, and more parameters

Section 5: Additional content 2: Convolution neural networks

Lecture 51 Introduction to convolutional neural networks

Lecture 52 Convolutional operator

Lecture 53 Pooling

Lecture 54 Flattening

Lecture 55 Dense neural network

Section 6: Final remarks

Lecture 56 Final remarks

People interested in creating complex applications using GANs,Undergraduate and graduate students who are taking courses on Computer Vision, Artificial Intelligence, Digital Image Processing or Computer Vision,People who want to implement their own projects using Computer Vision techniques,Data Scientists who want to increase their project portfolio