GEOMAR I Helmholtz Centre for Ocean Research Kiel
2015 Ph.D. in Computer Science:
"Automated detection in benthic images for megafauna classification and marine resource exploration: supervised and unsupervised methods for classification and regression tasks in benthic images with efficient integration of expert knowledge" (Bielefeld University)
2010 M.Sc. in Computer Science in the Natural Sciences:
"Towards Improved Epilepsia Diagnosis by Unsupervised Segmentation of Neuropathology Tissue Sections using Ripley's-L Features" (Bielefeld University)
2008 B.Sc. in Computer Science in the Natural Sciences:
"Web 2.0 techniques for exploratory image database analysis" (Bielefeld University)
My work aims to monitor the deep-sea with methods of pattern recognition and image processing. The focus lies on the exploration of large archives of 2D and 3D images of benthic and pelagic images.
Monitoring poly-metallic nodule abundance
In cooperation with the JPIO Partners, BGR and Saltation Marine IT, the amount, size and finally mass of polymetallic nodules lying on the seafloor is to be estimated. To monitor the minerals as well as the diverse biology in this sensitive habitat, precise measurements are required. This includes detections of the nodules as well as the biology.
The first part of this project was targeted to derive the percentage of the seafloor covered in nodules. Subsequently more detailed segmentation yielded single nodule sizes, and, combined with automatically detected laser markers, the relative size of the nodules in cm.
Automated detection of benthic mega fauna
One of the major projects, is the automated detection of mega fauna (as well as flora and Lebensspuren created by those animals). Images are provided by collaboration partners at the Alfred Wegener Institute and NOC Southampton. One result is an automated detection system (called iSIS, intelligent screening of image series) that incorporates the expertise of the biologists while keeping their input as small as possible. Parameters for different steps of the system are derived from a fully-annotated workshop transect. As the central part, supervised machine-learning is applied by Support Vector Machines.
Manual image annotation
A basis for the application of supervised and semi-supervised machine learning algorithms is a training set of reference samples. To obtain such a reference set, image annotation is required which was facilitated in different ways. Initially, whole images can be annotated using BIIGLE. More features and efficiency is currently implemented for the successor DIAS (Discol Image Annotation System) being developed for the JPIO Mining Impact project.
Web based visualization tools
A wide range of research focused visualizations tools has been implemented. These mostly facilitate link and brush interfaces to fuse representations of complex feature based descriptions with images or plots.
DELPHI—fast and adaptive computational laser point detection and visual footprint quantification for arbitrary underwater image collections
Timm schoening, Thomas Kuhn, Melanie Bergmann, Tim W. Nattkemper
Frontiers in Marine Science, 2015
Seabed classification using a bag-of-prototypes feature representation
Timm Schoening, Thomas Kuhn, Tim W. Nattkemper
Ultra-fast segmentation and quantification of poly-metallic nodule coverage in high resolution digital images
Timm Schoening, Björn Steinbrink, Daniel Brün, Thomas Kuhn, Tim W. Nattkemper
The impact of human expert knowledge on automated object detection in benthic images
Timm Schoening, Melanie Bergmann, Tim W. Nattkemper
Investigation of hidden parameters influencing the automated object detection in images from the deep seafloor of the HAUSGARTEN observatory
Timm Schoening, Melanie Bergmann, Anthe Boetius, Tim W. Nattkemper OCEANS 2012
Semi-Automated Image Analysis for the Assessment of Megafaunal Densities at the Arctic Deep-Sea Observatory HAUSGARTEN
Timm Schoening, Melanie Bergmann, Jörg Ontrup, James Taylor, Jennifer Dannheim, et al. (2012)
Biigle Tools - A Web 2.0 approach for Visual Bioimage Database Mining
Timm Schoening, Nils Ehnert, Jörg Ontrup, Tim W. Nattkemper