Road network maps facilitate a great number of applications in our everyday life. However, their automatic creation is a difficult task, and so far, published methodologies cannot provide reliable solutions. The common and most recent approach is to design a road detection algorithm from remote sensing imagery based on a Convolutional Neural Network, followed by a result refinement post-processing step. In this project I proposed a deep learning model that utilized the Multi-Task Learning technique to improve the performance of the road detection task by incorporating prior knowledge constraints. Multi-Task Learning is a mechanism whose objective is to improve a model's generalization performance by exploiting information retrieved from the training signals of related tasks as an inductive bias, and, as its name suggests, solve multiple tasks simultaneously. Carefully selecting which tasks will be jointly solved favors the preservation of specific properties of the target object, in this case, the road network. My proposed model is a Multi-Task Learning U-Net with a ResNet34 encoder, pre-trained on the ImageNet dataset, that solves for the tasks of Road Detection Learning, Road Orientation Learning, and Road Intersection Learning. Combining the capabilities of the U-Net model, the ResNet encoder and the constrained Multi-Task Learning mechanism, my model achieved better performance both in terms of image segmentation and topology preservation against the baseline single-task solving model. The project was based on the publicly available SpaceNet Roads Dataset.

With the rapid growth in point cloud acquisition technologies the recent years we have the ability to measure large quantities of 3D points of significantly detailed and geometrically composite scenes such as urban environments. This advantage can be exploited and used for direct analysis on point clouds. A direct point cloud analysis has several advantages over for example 3D surface reconstruction, such as the end result having more details and the computation being less expensive. In order to make a point cloud representation a suitable alternative for other types of 3D city models, they need to be semantically enriched, resulting in a rich point cloud. One element of this enrichment is the detection of objects, such as windows. Extracting these from facades is specifically what this research revolves around, which can be done by taking advantage of the fact that they show up as holes, since lasers of the point cloud scanner do not properly reflect on them. Two different general approaches are taken to detect windows in a by mobile laser scanner obtained point cloud of Noordereiland, Rotterdam, The Netherlands.