The main goal of the project is to perform texture-based segmentation using deep learning (semantic segmentation) providing images synthesized using the photometric stereo method.
We want to see if the images obtained using the photometric stereo method have a higher value over just a single RGB image when both images are fed into a deep learning model that performs texture-based segmentation. 3-D printed phantoms that mimic the texture of the skin will be used for collecting data.