AI restoration of endangered species

4K moving images.

2021. 20 sec/ each. 

Topaz gigapixel AI, After Effect, Premiere.

Minorities is a project stemming from the inherent bias of data, machine learning and artificial intelligence. 

This project has been started from biased data for machine learning. As Artificial Intelligence has issues about biased results, I would like to talk about the minor species that are not famous as much as cute endangered animals.

These creatures (Mushroom - Lampteromyces japonicus, Algae - Dictyosphaeria cavernosa, Coccophora langsdorfii) are on the endangered list in South Korea, also they belong to minor species: marine algae and fungi.

I finally decided to make them (Mushroom-Lampteromyces japonicus, Algae- Lampteromyces japonicus, Coccophora langsdorfii) the main character in this project..


As I was trying to get the scanned specimen from the Korean National Institute of Ecology, I found out it is quite hard to get access to those minor endangered species. They are living deep inside of the mountain, somewhere at the shore of Jeju Island and somewhere beside the rocks of the East sea.


Here, I’d like to ask the question

“What does it look like if the AI tries to restore the creatures?”


I begin from the one specimen image. 

Like photoshop, if you order to sharpen the images, then you will get better resolution images as the result.

But what if you keep continuing that order, again and again?

After 50 time attempts

The result was interesting to me, the images started to distort themselves as it made the images look sharp and high resolution.

The Original images

I downloaded this from https://www.nie.re.kr/, National Institute of Ecology.