Hello trend,
This is Satya Mallick from LearnOpenCV.com.
I hope you are having fun with our series on Generative Adversarial Networks (GANs). In the past two weeks, we went over
- The basic ideas behind GAN and also learned to train a plain vanilla GAN in both PyTorch and Tensorflow.
- Deep Convolutional GAN (DCGAN) GAN (DCGAN) and implemented one in PyTorch and Tensorflow
If you have been following this series, you are probably also familiar with Dr. Pawel Adamicz (left) and Dr. Kavita Sundarajan (right); the fake researchers who did not invent GANs and whose pictures were created using a GAN (https://thispersondoesnotexist.com/).
While creating these personalities, we had no control over the kind of images the GAN would output. For example, while looking for an image that fit the female persona of Dr. Kavita Sundarajan, we kept generating images until we found a person who looked female, Indian, and scholarly. At least half of the faces generated during our search looked male, and much larger percentage looked non-Indian.
Wouldn't it be cool if we had more control over the kind of image we wanted the GAN to produce? It would have saved us a lot of time if we could tell the GAN to generate only female looking faces, or faces with darker skin tones.
Conditional GAN (cGAN) solves this problem by taking an additional piece of information (eg. class label) as input to guide the output image generation. We will learn all about cGAN in today's post
https://learnopencv.com/conditional-gan-cgan-in-pytorch-and-tensorflow/
And, we will also learn how to train a cGAN on the Rock Paper Scissors dataset in both PyTorch and Tensorflow.
https://github.com/spmallick/learnopencv/tree/master/Conditional-GAN-PyTorch-TensorFlow
INDIEGOGO CAMPAIGN ENDS SOON
Our Indiegogo campaign for OpenCV for Beginners ends on July 23, 2021.
All existing courses are also selling at a massive discount during this campaign. If you are interested, this is your chance to start your AI journey.
| AI Courses : From Basics to Mastery |
Bringing Autonomy Off-Road-driving
Autonomy is not just for self driving cars. There are a many application areas where autonomous robots are making our lives easier. On this week's OpenCV Weekly Webinar we will talk to Jack Morrison, CEO of Scythe Robotics - a company building autonomous lawn mowers!
- Topic : Bringing Autonomy Off-Road-driving
- Date and Time : July 15, 9 AM Pacific Time
- Registration : Free Registration Link.
Last week's webinar with Minsoo Lee (CEO, Bluebox Labs) is now live. Check it out at
Also check out their Kickstarter campaign which has raised more than $75k so far!
Satya
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.