<> "The repository administrator has not yet configured an RDF license."^^ . <> . . "Machine Learning for Instance Segmentation"^^ . "Volumetric Electron Microscopy images can be used for connectomics, the study of brain connectivity at the cellular level. \r\nA prerequisite for this inquiry is the automatic identification of neural cells, which requires machine learning algorithms and in particular efficient image segmentation algorithms.\r\nIn this thesis, we develop new algorithms for this task. \r\nIn the first part we provide, for the first time in this\r\nfield, a method for training a neural network to predict optimal input data for a watershed algorithm. \r\nWe demonstrate its superior performance compared to other segmentation methods of its category.\r\nIn the second part, we develop an efficient watershed-based algorithm for weighted graph\r\npartitioning, the \\emph{Mutex Watershed}, which uses negative edge-weights for the first time.\r\nWe show that it is intimately related to the multicut and has a cutting edge performance on a connectomics challenge. \r\nOur algorithm is currently used by the leaders of two connectomics challenges.\r\nFinally, motivated by inpainting neural networks, we create a method to learn the graph weights without any supervision."^^ . "2020" . . . . . . . "Steffen"^^ . "Wolf"^^ . "Steffen Wolf"^^ . . . . . . "Machine Learning for Instance Segmentation (PDF)"^^ . . . "steffen_wolf_thesis_compressed.pdf"^^ . . . "Machine Learning for Instance Segmentation (Other)"^^ . . . . . . "indexcodes.txt"^^ . . . "Machine Learning for Instance Segmentation (Other)"^^ . . . . . . "lightbox.jpg"^^ . . . "Machine Learning for Instance Segmentation (Other)"^^ . . . . . . "preview.jpg"^^ . . . "Machine Learning for Instance Segmentation (Other)"^^ . . . . . . "medium.jpg"^^ . . . "Machine Learning for Instance Segmentation (Other)"^^ . . . . . . "small.jpg"^^ . . "HTML Summary of #28353 \n\nMachine Learning for Instance Segmentation\n\n" . "text/html" . . . "530 Physik"@de . "530 Physics"@en . .