Keynote and Invited Speakers

"Depth prediction from monocular: towards unsupervised and self-adaptive reconstruction" PD Dr. Habil. Federico Tombari Technical University of Munich, Germany

Depth prediction from a single image is a computer vision topic that has recently attracted a lot of attention as an interesting alternative to traditional geometry-based approaches for a variety of applications that involve 3D reconstruction and mapping, e.g. in the field of robotics, augmented reality, photogrammetry, computer graphics to name a few. This is motivated by recent solutions in the field of deep learning, in particular Convolutional Neural Networks, that proved to be effective in densely and accurately predicting depth maps given an appropriate training set. Nevertheless, challenges still remain in particular with regards to the domain shift and efficiency aspects. In this talk, I will give an overview of the state of the art in depth prediction, with a focus on current trends and solutions towards unsupervised and self-adaptive learning aimed at overcoming the aforementioned limitations. I will also discuss the use of depth prediction for SLAM and semantic reconstruction, and well as applications in the field of robotics, autonomous driving, augmented reality and photogrammetry.