Image Colorization using U-Net with Skip Connections and Fusion Layer on Landscape Images

TL;DR

This paper introduces a novel image colorization method combining U-Net with a Fusion Layer to enhance colorization quality by integrating local and global image information, validated through user studies and comparisons.

Contribution

The paper proposes a new colorization technique that merges local patch-based details with global image priors using a Fusion Layer in a U-Net architecture.

Findings

Improved colorization results over state-of-the-art methods.

Validated effectiveness through user study evaluations.

Demonstrated better visual quality in colorized images.

Abstract

We present a novel technique to automatically colorize grayscale images that combine the U-Net model and Fusion Layer features. This approach allows the model to learn the colorization of images from pre-trained U-Net. Moreover, the Fusion layer is applied to merge local information results dependent on small image patches with global priors of an entire image on each class, forming visually more compelling colorization results. Finally, we validate our approach with a user study evaluation and compare it against state-of-the-art, resulting in improvements.