This project develops improved color palette extraction from images, 2017.
About
To improve the other programs that extract the color palette from the image, our program serve clearly representative color sets with dominant/assort/accent color presented. Furthermore, our system serves extraction from the video. Our algorithm is based on the clustering and distance based merging system which uses domain knowledge of color-study. Furthermore our algorithm is combined with body inferring technique for obtaining representation color with more saliency.
The poster of the movie ‘LaLa-Land’
Extracted color palette (Dominant / Accent / Assort)
Awarded
Creative IT Convergence Korea: Grand prize - Ministry of Science and ICT, Korea (2018)
Grand prize - Pohang University of Science and Technology(2017)
Presentation Video
presentation from 4:13
awarded from 1:38:00
Poster PDF
Git
https://github.com/PulseKim/A_Palette