Unsupervised Diverse Colorization Via Generative Adversarial Networks

ARXIV  ·  2017  ·  Impact 5

Yun Cao; Zhiming Zhou; Weinan Zhang; Yong Yu;

HIGHLIGHT

In this paper, we design a novel solution for unsupervised diverse colorization.

RELATED

Papers      Patents     Grants     Venues     Experts     Code    Cite    Share    Save

ABSTRACT

Colorization of grayscale images has been a hot topic in computer vision. Previous research mainly focuses on producing a colored image to match the original one. However, since many colors share the same gray value, an input grayscale image could be diversely colored while maintaining its reality. In this paper, we design a novel solution for unsupervised diverse colorization. Specifically, we leverage conditional generative adversarial networks to model the distribution of real-world item colors, in which we develop a fully convolutional generator with multi-layer noise to enhance diversity, with multi-layer condition concatenation to maintain reality, and with stride 1 to keep spatial information. With such a novel network architecture, the model yields highly competitive performance on the open LSUN bedroom dataset. The Turing test of 80 humans further indicates our generated color schemes are highly convincible.

 Paraphrase     Questionfy     Wikify     Simplify     Summarize

Download