David Serrano-Lozano

Fig.1: David mimicking a T-Rex at the American Museum of Natural History.

Hola! I am David Serrano-Lozano (see Fig. 1).

I am a PhD Student at the Computer Vision Center, Universitat Autònoma de Barcelona (UAB), based in Barcelona, Spain. I am advised by Prof. Javier Vazquez-Corral and Dr. Luis Herranz.

My research interests lie at the intersection of Low-Level Computer Vision, Generative Models and Computational Photography. I am paticularly focused on making models more interpretable and understandable.

I earned my M.Sc. in Computer Vision from Universitat Autònoma de Barcelona (UAB) where I received honors for my master’s thesis on “Uncertainty as a Proxy of the Generalization Error for Marine Species Identification” advised by Dr. David Masip. Previously, I obtained a B.Sc. in Telecommunication Networks Engineering from Universitat Politècnica de Catalunya (UPC), where I completed my bachelor’s thesis on “Automatic Detection of Visible Events in Fusion Reactors” advised by Prof. J. Ramon Morros. During my M.Sc. I spent two years at Ctrl4Enviro, managing the research and development of efficient computer vision models deployed on NVIDIA edge-computing embedded systems.

selected publications

ECCV’24

NamedCurves: Learned Image Enhancement via Color Naming

David Serrano-Lozano, Luis Herranz, Michael S. Brown, and Javier Vazquez-Corral

In ECCV, 2024

Abs arXiv HTML Code

A popular method for enhancing images involves learning the style of a professional photo editor using pairs of training images comprised of the original input with the editor-enhanced version. When manipulating images, many editing tools offer a feature that allows the user to manipulate a limited selection of familiar colors. Editing by color name allows easy adjustment of elements like the "blue" of the sky or the "green" of trees. Inspired by this approach to color manipulation, we propose a learning-based image enhancement technique that separates the image into a small set of named colors. Our method learns to globally adjust the image for each specific named color via tone curves and then combines the images using an attention-based fusion mechanism to mimic spatial editing. We demonstrate the effectiveness of our method against several competing methods on the well-known Adobe 5K dataset and the PPR10K dataset, showing notable improvements.