Bias in Automated Image Colorization: Metrics and Error Types
Not on the map. Why?
Abstract (arXiv)
We measure the color shifts present in colorized images from the ADE20K dataset, when colorized by the automatic GAN-based DeOldify model. We introduce fine-grained local and regional bias measurements between the original and the colorized images, and observe many colorization effects. We confirm a general desaturation effect, and also provide novel observations: a shift towards the training average, a pervasive blue shift, different color shifts among image categories, and a manual categorization of colorization errors in three classes.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 10 of its 20 references that are in the map. First posted on arXiv 16 Feb 2022.
Builds on
10 of 20 references.
References · 10
- 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,510
- 2016Colorful Image Colorizationcited by 1,113
- 2016Learning Representations for Automatic Colorizationcited by 251
- 2020Instance-aware Image Colorizationcited by 76
- 2016Deep Colorizationcited by 70
- 2019ChromaGAN: Adversarial Picture Colorization with Semantic Class Distributioncited by 60
- 2021Towards Vivid and Diverse Image Colorization with Generative Color Priorcited by 55
- 2016Learning Diverse Image Colorizationcited by 43
- 2018Image Colorization with Generative Adversarial Networkscited by 24
- 2019Image Colorization By Capsule Networkscited by 10
Built on by
Citing papers per year: 2024: 1.
1 paper.