We are looking for motivated students who want to pursue their Bachelor / Master thesis at the transition between Deep Learning and Computer Graphics & Visualization. Two thesis works are available in der area of deep learning for terrain data synthesis and deep weather. In the first work we intend to use generative networks to learn from satellite images and ground cover data to synthesize new orthophotos that can be used in navigation systems or 3D viewers. In the second work we aim at exploring the use of deep learning networks for improving the prediction of weather. We intend to look at the training of deep networks using recorded data from the past and enhancing the learning process by weather forecast ensembles. Another work is on the reconstruction of photorealistic renderings of water surfaces from learned representations. Here it is in particular the use of transparency in feature learning.
Bachelor and Master thesis in the following areas