W

colorization-bias

public

2 stars

0 forks

1 issues

Commits

List of commits on branch main.

Unverified

fc272b961f343e6433c9b8865cb096b94e72a938

Add examples to table in README.md

ffrankstapel committed 3 years ago

Unverified

384ffa0aa1f2fb7256e616e80bf1cb150f007d5b

Add human examples

ffrankstapel committed 3 years ago

Unverified

2de5675d2bc10bc7359ce9c9aa9207dab181f545

Add examples images

WWeersProductions committed 3 years ago

Verified

2b8312a5bdf80d756cacdb85b5232d178e46e9d2

Initial commit

WWeersProductions committed 3 years ago

README

The README file for this repository.

Bias in Automated Image Colorization: Metrics and Error Types

This repository contains examples of colorized images. All images are colorized using the artistic model from DeOldify.

Images

For each reason, we selected the image with the highest value. Images of humans are examples that were not chosen through specific values.

Ground Truth	Colorized	Image ID	Reason	Value
		00000136	human	-
		00002262	human	-
		00005581	third_bottom_diff	77.414
		00006607	human	-
		00006700	third_top_diff	69.757
		00012765	center_average_diff	81.819
		00023303	human	-
		00024091	gold_dots_diff	62.353
		00024474	human	-
		00024567	third_left_diff	75.708

Structure

`images/`

Contains two files for each image and a values.md that explains why the image was chosen.