In recent years the US government has begun to collect a vast number of descriptions and imagery of prison inmate tattoos. The FBI’s ‘Tatt-C’ project has openly dedicated research to the use of tattoo recognition technologies that can be implemented to identify suspects and even predictively police. This poses serious concerns for privacy, free expression, religious freedom, and the right of association. Using a data-set of one million tattoo captions from the Florida Department of Corrections, this program uses a recurrent neural network to hallucinate suggestions of personal tattoos based on the user input’s name, gender and physical attributes.
This project sees a trained machine learning algorithm on a database of over one million tattoo descriptions and now it can generate totally plausible ideas for new ones on its own. Note: I am not responsible for the vulgarity of these ideas, the machine did it.
Project documentation: https://www.instagram.com/p/Beqx7BpHbOq/