# PixColor: Pixel Recursive Colorization

Source: [PixColor: Pixel Recursive Colorization](<https://www.sota2.com/research/paper/68fa63631e209131ce9fe513>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-12-14T16:48:08.975Z

Authors: Sergio Guadarrama, Ryan Dahl, David Bieber, Mohammad Norouzi, Jonathon Shlens, Kevin Murphy

Published: 2017-05-19T22:10:51.000Z

## Abstract

We propose a novel approach to automatically produce multiple colorized versions of a grayscale image. Our method results from the observation that the task of automated colorization is relatively easy given a low-resolution version of the color image. We first train a conditional PixelCNN to generate a low resolution color for a given grayscale image. Then, given the generated low-resolution color image and the original grayscale image as inputs, we train a second CNN to generate a high-resolution colorization of an image. We demonstrate that our approach produces more diverse and plausible colorizations than existing methods, as judged by human raters in a "Visual Turing Test".

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Colorization on ImageNet 500-image human evaluation (test)](<https://www.sota2.com/research/sota/image-colorization-on-imagenet-500-image-human-evaluation-test.md>): Image Colorization; ImageNet 500-image human evaluation (test); AMT Fooling Rate: 38.3.

- [Image Colorization on ImageNet (val)](<https://www.sota2.com/research/sota/image-colorization-on-imagenet-val.md>): Image Colorization; ImageNet (val); FID: 24.32.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/1705.07208>)

[PDF](<https://arxiv.org/pdf/1705.07208.pdf>)