課程資料
課程簡介
Face Generation with GANs, WGANs and ProGANs. Image super-resolution with SRGAN, Interior Design with Stable Diffusion
Image generation has come a long way, back in the early 2010s generating random 64x64 images was still very new. Today we are able to generate high quality 1024x1024 images not only at random, but also by inputting text to describe the kind of image we wish to obtain.
In this course, we shall take you through an amazing journey in which you'll master different concepts with a step by step approach. We shall code together a wide range of Generative adversarial Neural Networks and even the Diffusion Model using Tensorflow 2, while observing best practices.
You shall work on several projects like:
Digits generation with the Variational Autoencoder (VAE),
Face generation with DCGANs,
then we'll improve the training stability by using the WGANs and
finally we shall learn how to generate higher quality images with the ProGAN and the Diffusion Model.
From here, we shall see how to upscale images using the SrGAN
Final Project: AI Interior Designer
You will build an application that can take any photo of an empty room and breathe life into it. We will architect a pipeline that truly understands the space.
Step 1: Scene Understanding. First, we’ll use the Depth Anything model to generate a precise depth map, giving our AI an understanding of the room's 3D geometry.
Step 2: Intelligent Masking. Next, we'll use a powerful combination of Grounding DINO and Segment Anything (SAM) to automatically detect and create masks for key areas like the door, and windows.
Step 3: Controlled Generation. Finally, we will feed the original image, the depth map, and the segmentation masks into ControlNet with a Stable Diffusion Inpainting model. This allows us to tell the AI, "Generate a modern sofa here on the floor, respecting the room's depth and leaving the windows untouched." The result is a stunning, realistic, and context-aware interior design.
If you are willing to move a step further in your career, this course is destined for you and we are super excited to help achieve your goals!
This course is offered to you by Neuralearn. And just like every other course by Neuralearn, we lay much emphasis on feedback. Your reviews and questions in the forum, will help us better this course. Feel free to ask as many questions as possible on the forum. We do our very best to reply in the shortest possible time.
Enjoy!!!
課程章節
- 9 個章節
- 37 堂課
- 第 1 章 Introduction
- 第 2 章 Variational Autoencoder
- 第 3 章 Deep Convolutional Generative Adversarial Neural Network
- 第 4 章 Wasserstein GAN
- 第 5 章 High quality face generation with ProGan
- 第 6 章 Image super resolution with SRGan
- 第 7 章 Diffusion Models
- 第 8 章 Interior Designer (Empty House Filling) with Stable Diffusion (Img2Img)
- 第 9 章 Interior Designer with Stable Diffusion Inpainting Model
課程內容
- Understanding how variational autoencoders work
- Image generation with variational autoencoders
- Building DCGANs with Tensorflow 2
- More stable training with Wasserstein GANs in Tensorflow 2
- Generating high quality images with ProGANs
- Building mask remover with CycleGANs
- Image super-resolution with SRGANs
- Advanced Usage of Tensorflow 2
- Image generation with Diffusion models
- How to code generative A.I architectures from scratch using Python and Tensorflow
此課程所涵蓋的技能
評價
-
AAmitabh Joshi
Great explanations!
-
MMarcus Karoumi
Glad to have found this course after searching for a longtime for one which covers diffusion models math. highly recommended.
-
MMohd Rehan
Good
-
MMarco Caimi
Content-wise, this course is quite complete: even Diffusion Models are explained, which is a plus. However, there are major audio issues throughout the whole course and the presentation is not that good. (especially the drawn overlays: these mouse-written formulas are barely intellegible, are often in the way and hard to follow). Moreover, no links are shared about any of the papers mentioned in the course.