This image displays a scientific poster presented at CVPR in Seattle, WA, titled "Zero-Reference Low-Light Enhancement." The researchers, Wenjing Wang, Huan Yang, and Jianlong Fu, focus on using normal light images for enhanced low-light processing without supervision. The poster outlines the introduction, aim, and methodology, which involves creating an illumination-invariant prior to bridge between varying light conditions. A diagrammatic depiction of the training framework and details of degradation solutions are included. The bottom of the poster shows comparative examples of enhancement effects.
Text transcribed from the image:
CVPR Zero-Reference Low-L
SEATTLE
Wenjing Wang Huan Yang Jianlong
Introduction
Zero-Perence Low-Light Enhancement leam
solely with nomalight images aducing the need
Method
Training framework
Pat a physca quadruple prom
Pecoraruct the prior back to imed
Our aim improve the bes
Outa usage during hig
unason specific hyper
Our methodology daganation
OH
Selon of detal degraddon bypass
Lowlight whees te pe