Team 'Artificially Intelligent' win at Govhack 2019
Team 'Artificially Intelligent' have taken out the top award at Govhack 2019 in the Digital Culture category.
AIML researchers Yasir Latif, Sam Bahrami, Boris Repasky, Mahdi Kazemi and Thomas Rowntree won for their work training two machine learning models using two specific data sets.
The first data set was from the History Trust of South Australia's photographic collection, and the second was a series of old colonist photographs from the State Library of South Australia.
The first machine learning model applied a concept called neural style transfer to current images and historical photos (dating back to the mid-19th century). This allowed the style of one image to be transferred to another image, enabling the user to reproduce any photographic style regardless of the age of the photo.
The second model allowed the user to identify the colonist photograph that most closely resembled them. This was achieved using facial similarity detection, a neural network that they trained to find the most similar face.
Both models could potentially be used digitally and physically; more details about how the team would achieve this are here.