The 43rd IPP Symposium

Learning to Adjust Photographs

Sylvain Paris, Adobe

I will first briefly present a few research projects that have been done in my group at Adobe. The rest of my talk will be about my on-going project about leveraging machine learning in the context of photographic adjustment. We collected 5000 photographs and hired 5 photographers to retouch them by hand. This gives us a unique source of information that shows what photographers actually do whey they adjust images. We exploit these data with a supervised learning approach to enable high-quality automatic photo adjustment. We also use transfer learning to adapt to users' preferences. This work is in collaboration with Vladimir Bychkovsky and Frédo Durand at MIT, and Eric Chan at Adobe.

Biography: Sylvain Paris obtained his PhD at INRIA in Grenoble, France in 2004 under the supervision of Dr. François Sillion. Then, he moved to Cambridge Massachusetts where he worked with Prof. Frédo Durand at MIT as a post-doctoral associate. Since 2007, he is a researcher at Adobe. His research interests are computational photography, hair modeling, and related topics.