Colorization Transformer

TL;DR

The Colorization Transformer introduces a self-attention based method for diverse, high-fidelity image colorization, outperforming previous state-of-the-art techniques in both quantitative and human evaluations.

Contribution

It proposes a novel three-step transformer-based architecture for image colorization that produces diverse and high-quality color outputs, advancing the state-of-the-art.

Findings

Outperforms previous methods in FID scores

Achieves higher human preference rates over ground truth in many cases

Produces diverse colorizations from grayscale images

Abstract

We present the Colorization Transformer, a novel approach for diverse high fidelity image colorization based on self-attention. Given a grayscale image, the colorization proceeds in three steps. We first use a conditional autoregressive transformer to produce a low resolution coarse coloring of the grayscale image. Our architecture adopts conditional transformer layers to effectively condition grayscale input. Two subsequent fully parallel networks upsample the coarse colored low resolution image into a finely colored high resolution image. Sampling from the Colorization Transformer produces diverse colorings whose fidelity outperforms the previous state-of-the-art on colorising ImageNet based on FID results and based on a human evaluation in a Mechanical Turk test. Remarkably, in more than 60% of cases human evaluators prefer the highest rated among three generated colorings over the ground truth. The code and pre-trained checkpoints for Colorization Transformer are publicly available at https://github.com/google-research/google-research/tree/master/coltran