Introduction to Generative Adversarial Networks with PyTorch

This course is horrible, Nothing is really explained. They market in the intro video about doing vid2vid but then when you go to the section it doesnt havent any useful information in there at all. Im going to ask for a refund. But DO NOT BUY THIS COURSE. Or accept that you are throwing money away for poorly explained jupyter notebooks that anyone could find

Would love a bit more detail on what things are, like stride, Max Pool, etc. Maybe that's on another course.

Not at all good course for learning theory of GANs. Speaker rushes through the content, and have not covered much. Speaker's focus is on code mostly.

Great course! I have been working through another of Mustafa's courses and enjoy his teaching style and clear examples.

No he was very quick with the course content and the code, also the part of the hidden layer graph did not even work

I am a new PhD student in computer science with concentration on image and video coding, one sub-field of computer vision. I am impressed by the depth of this Udemy course. The course is not an introduction course but a nearly-complete course for GAN using Pytorch. If you read the research papers in GAN and their source codes in Github, you tends to be lost. However, this instructor gave you a very comprehensive review for the most important GAN papers and much clear codes that can boost your academic skills in GAN. After this course, I feel I would be able to grasp some latest papers in GAN and write sophisticate GAN codes. It would be much helpful for me to do research in GAN and hope I can publish a top conference paper in GAN in the near future. Thank you!

Instructor has good knowledge. Course is structured properly. Main issue which I found out to be was that this course is rushed too much. A lot of the important parts are just skipped over. Even someone having basic knowledge would struggle because cells of jupyter notebooks are just skipped at times. There was no description of the methods being used from torch.

Thank Mustafa for the really nice course. I liked the fact that you've included various visualization technics and explained many utilities for preprocessing! I think that Jupyter notebooks were very useful for understanding the concepts and it definitely helped me to learn more than by simply looking at the papers. However, I would like to see more visuals and text descriptions in notebooks to better understand the code. Also, I think it would be great if you could leave some code blocks empty for students to implement themselves and gave us some additional tests below those blocks to verify that our implementation is working properly. Additionally, I would consider adding some lectures about math in GANs. Coming with no background in GANs and having only a little experience in coding neural nets, it was a little hard for me to understand some math concepts like ones in Wasserstein GANs. Overall, it was a great course, and I thank you for your time answering my questions in comments section! It was a pleasure taking this course

It would be better writing the code during the class, not only saying what it does following a Jupyter Notebook - even because there are no in-depth explanations on the theory behind. The notebooks present a well structured code (so it gives a hint at how to make yours), but some of them are not actually working (I'm gonna post the problems in the forum later). In short: good introduction because it leads you to the right paths, but it could include some more content.

Mustafa is a great lecturer, he explains very smoothly and clearly all the concepts. I wasn't familar with PyTorch but his explanation helps me understand PyTorch better. Moreover the ressources are great and very intuitive and help keep track with the course.