Corals in Colour has been created in response to the crisis surrounding our Coral Reef ecosystems. Sea temperatures and acidity levels are drastically increasing and as a result cause many coral species to go into a state of shock. This state of shock presents its self in the form of a dazzling fluorescent display of colours. Scientists are still unsure as to why corals respond in this way but they believe it could be kind of ‘sunscreening’ effect, distress singling to other corals or a last attempt to retain or attract algae. If their environment does not return back to normal conditions the corals will likely expel their symbiotic hosts, zooxanthellae, and become lifeless. This is called the bleaching process. Surprising, there is still hope for life to return back to this white skeletal structure however, this needs to happen before the calcium starts to deteriorate and die.
My curiosity lies within the scientific work that is taking place to help conserve and restore coral reef.
An important element of coral reef restoration is their classification (identification, labelling and catagorising) This is a painstakingly difficult job for researchers and ultimately slows down the progress of restoration and as we know, time is of the essence when it comes to climate issues.
This is where Machine Learning technologies come into the equation. Much like humans, machine learning techniques use the fundamental laws of recognising and learning from colour, texture, shape and form. For this particular project I collaborated with Fernando Pérez-García, a researcher in medical imaging and machine learning at UCL. Using open-access coral reef datasets, Fernando trained a generative machine learning model that hallucinates images of non-existing corals. From these images I have interpreted my own set of images using the same principals.