My research project
Deep generative models for photorealistic dynamic scene modelling
My project focuses on relighting objects to the illumination conditions of a target scene so that they can be believably composited into the target scene. This involves illumination estimation of the target scene, shape estimation of the object and relighting the object. I am using deep neural networks such as auto-encoders and generative adversarial networks to create a model that is generalised for any object.