YSU Krisp AI Lab Seminars: Narek Tumanyan
Summer continues to be full of useful workshops at YSU AI laboratory. This time, the topic will be of interest not only to specialists, but also to just interested ones.
Title: Splicing ViT Features for Semantic Appearance Transfer
Speaker: Narek Tumanyan, Graduate Student at The Weizmann Institute of Science
Date and time: August 19,12:00
Short introduction: The seminar will cover “Splice” – a method for semantically transferring the visual appearance of one natural image to another. Specifically, the goal is to generate an image in which objects in a source structure image are “painted” with the visual appearance of their semantically related objects in a target appearance image. The key idea of “Splice” is to leverage a pre-trained and fixed Vision Transformer (ViT) model by deriving novel representations of structure and appearance from deep ViT features. “Splice” does not involve adversarial training, nor does it require any additional input information such as semantic segmentation or correspondences, and can generate high-resolution results on a variety of in-the-wild image pairs, under significant variations in the number of objects, their pose and appearance.
To get a sense of the method and the results, feel free to check out the project page before the seminar: splice-vit.github.io