Speaker: Pedro Hermosilla
Recent advances in machine learning for 3D data have revolutionized the fields of computer vision and computer graphics. These techniques have enabled researchers to train neural network architectures directly from 3D data. Among these technologies, neural networks for unstructured data or point clouds have gained a lot of attention in the past years since they are able to work with sparse 3D representations, saving large amounts of memory. However, these technologies do not come without a cost. In this presentation, I will talk about the challenges that these networks pose, how to overcome them, and how they can be used to solve different problems in the fields of computer vision, computer graphics, and bio-medicine.