Recent Advances in Unsupervised Image-to-image Tranformation

 Today's talk is by Xun Huang from Cornell University (and Nvidia) on 'Recent Advances in Unsupervised Image to Image Translation' and was given at Microsoft Research in Sept 2019. 

Note that for the title of the post we're using HTC preferred terminology of Transformation for this subject area, since Translation is typically thought of as an affine transformation.


The project page for the talk is available here.  Here's a link to some of Xun's publications.  He also has a personal page with a lot of info on it.

Abstract for this talk:
Unsupervised image-to-image translation aims to map an image drawn from one distribution to an analogous image in a different distribution, without seeing any example pairs of analogous images. For example, given an image of a landscape taken in the summer, one may want to know what it would look like in the winter. There is not just a single answer. One could imagine many possibilities due to differences in weather, timing, lighting, etc.

Observations:
1:  What he is really covering in this talk is why they are interested in the Unit GAN architecture in stead of the CycleGAN architecture.  He is not the world's greatest speaker, and the small Microsoft audience keep interrupting him with particularly stupid questions, but what he has to say when he gets around to it is interesting.

2: Here are some associated references for the talk.

Here's his paper on MultiModal Unsupervised Image-to-Image Translation (MUNIT).

Original Unsupervised Image-to-Image Translation Networks (UNIT) Nvidia paper is here.

Proposes coupled GANs architecture.

Here's a comparison of CycleGan and Unit github project.

From the following paper.


3:  On a totally different topic, his PointFlow paper on 3D Point Cloud generation is pretty interesting.

Comments

Post a Comment

Popular posts from this blog

CycleGAN: a GAN architecture for learning unpaired image to image transformations

Image
 Like the Doublemint Twins touting the joys of Doublemint gum, 2 GANs are surely better than 1 GAN.  Especially if we package them together inside of one meta GAN module.  And this is exactly what the CycleGAN architecture does. Have you ever harbored dark secrets of turning a horse into a zebra?  The CycleGAN was developed to do just that. Learn how to turn a horse into a zebra.  And more. Now right away you can notice a difference between the image to image transformation GAN architectures we've been discussing over the last few posts.  Those last few posts described systems that learn from a database of matched input-output image pairs. And if your goal is to turn an edge representation into a nicely filled in continuous tone image, it's easy to build your database of matched input-output image pairs that your GAN system can then learn off of.  Take a continuous tone image (which will be the output of the database pair entry), then run it through an edge detector algorithm. 
Read more

Pix2Pix: a GAN architecture for image to image transformation

Image
 I thought following up yesterday's TraVelGAN post with a Pix2Pix GAN post would be useful to compare what is going on in the 2 architectures.  Two different approaches to the same problem. I stole this Pix2Pix Overview slide below from an excellent deeplearning.ai GAN course (note that they borrowed it from the original paper) because it gives you a good feel for what is going on inside of the Pix2Pix architecture.    Note how the Generator part is very much like an auto-encoder architecture, but rebuilt using the U-Net architecture features (based on skip-connections) that fastai has been discussing in their courses for several years before it became more widely known to the deep learning community at large  (and which originally came from an obscure medical image segmentation paper) . So the Generator in this Pix2Pix GAN is really pretty sophisticated, consisting of a whole image to image auto-encoder network with U-Net skip connections to generate better image quality at highe
Read more

Smart Fabrics

Image
"What makes smart fabrics revolutionary is that they have the ability to do many things that traditional fabrics cannot, including communicate, transform, conduct energy and even grow."  This is a quote from Rebeccah Pailes-Friedman, who is a professor at Pratt with a research focus on developing wearable technology and smart textiles. Enhancing textiles can be viewed through different lens.  Enhancements can be aesthetic, or performance boosting, or durability boosting. 
Read more