acm-header
Sign In

Communications of the ACM

ACM TechNews

Cornell Cis and Adobe Collaboration Creates AI Photo Tool


View as: Print Mobile App Share:
The result of the style of a reference image transferred onto an input image, while preserving photorealism.

Cornell University and Adobe researchers have developed software that can transfer the look of one photo onto another, while using neural networks to preserve the details of the original image.

Credit: Cornell University/Adobe

Researchers at Cornell University and Adobe have developed Deep Photo Style Transfer, software that can transpose the look of one photo onto another using neural networks to ensure the details of the original image are preserved.

"What motivated us is the idea that style could be imprinted on a photograph but it is still intrinsically the same photo," says Cornell professor Kavita Bala.

The researchers say the major breakthrough involved preserving boundaries and edges while still transferring the style. They used deep machine learning to add a neural network layer that pays close attention to edges within the image, such as the border between a tree and a lake.

"The method we came up with is surprisingly very effective," Bala says. "It has definitely captured people's imaginations as a way to stylize photos in a more dramatic way."

From Cornell Chronicle
View Full Article

 

Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA


 

No entries found