A web generative sculpture made with generated human skin.
For the Conference xCoAx 2020, I presented this installation which generates a threedimensional artificial skin. The Artificial Skin is generated using Machine Learning through a collection of Tinder profile pictures. The images are scraped from the matches of an account created for research purposes. The collection of images is then classified through a Semantic Segmentation approach to be able to recognize skin regions in any image where a human body is visible. After the ML model is trained, I provide any random combination of 2D shapes using the same Semantic Segmentation Schema, which allows the system to generate imagery of skin blobs.
Team
Andres Villa Torres.
Technologies
Python, PyTorch, Computer Vision, WEB, Colab, Java, three.js, Machine Learning.
Partners
xCoAx, Pro Helvetia, MUMUTH, Forum Stadtpark, IEM.