The Helmholtz-Zentrum Geesthacht (HZG), in Geesthacht, near Hamburg, and in Teltow, near Berlin, conducts materials and coastal research. For further information please refer to: www.hzg.de
HZG is one of the 19 national institutions of the Hermann von Helmholtz Association of German Research Centres e.V. (HGF). Around 1,000 employees carry out basic research and development work in close cooperation with national and international research institutions, research-oriented clinics and economic and public institutions.
At the Institute of Materials Research some 200 scientists conduct research in the five sub-institutes “Materials Mechanics”, “Materials Physics”, “Materials Technology”, “Magnesium Innovation Centre MagIC” and “Metallic Biomaterials”.
Research Software Developer/Scientist (m/f/d) deep learning for segmentation in X-ray tomography
Reference code: 50049323_2 – 2020/WP 16Commencement date: as soon as possibleThe place of employment is Hamburg.
Helmholtz-Zentrum Geesthacht (HZG) operates an outstation at DESY in Hamburg in order to provide access to highly brilliant synchrotron radiation within its German Engineering Materials Science Center (GEMS). At the PETRA III synchrotron radiation source, HZG jointly operates the High Energy Materials Science Beamline (HEMS, P07), the Imaging Beamline (IBL, P05) and the Nanofocus end station of the Micro- and Nanofocus X-ray Scattering Beamline (MiNaXS, P03) along with supporting laboratory instrumentation. HZG is regularly producing large amounts of 3D and 4D X-ray tomography data. Segmentation is typically required in order to analyze and interpret these data sets which are often challenging in terms of contrast, texture, noise, size or heterogeneity. Here, deep-learning-based segmentation approaches have shown to be successful. Therefore, we aim to develop a framework for the semi-automatic and interactive segmentation of X-ray tomography data using machine learning methods. This includes the automatic selection of the most suitable deep-learning architecture, a guided interactive and iterative strategy for the annotation of training data, and the deployment of a browser-based web service. Furthermore, we examine deep-learning methods for the segmentation of identical objects (instance segmentation). The position is initially limited to 2 years.
Your tasks
- development and implementation of a pipeline for semi-automatic segmentation
- development and implementation of quality metrics for the evaluation of deep-learning-based segmentation
- development and implementation of convergence criteria for an interactive annotation strategy
- development and implementation of a pipeline for automated image denoising
- development of deep-learning-based approaches for instance segmentation
- integration of developed methods and pipelines as a web service within the HPC environment at DESY
- annotation of tomographic data sets for the training of neural networks
- application of the developed algorithms and services to user data
- close collaboration with project partners and the Helmholtz.AI framework
- publish and present scientific results at international conferences and workshops
Your profile
- PhD in computer science, physics, mathematics or equivalent
- profound knowledge in the areas of artificial intelligence, deep/machine learning and image processing
- experience in deep learning and software/algorithm development
- experience with image segmentation
- proficient in relevant programming languages and machine learning libraries (PyTorch, Keras, TensorFlow, python, C/C++, CUDA etc.)
- excellent spoken and written command of the English language
- team player with good communication skills
- motivation to work in an interdisciplinary and international team
These qualifications are considered assets:
- experience with instance segmentation
- experience with web services and software frameworks
- experience in X-ray tomography
- experience with image data analysis and handling of large data sets
- experience in high performance, parallel, or distributed computing
- experience with image enhancement or image registration
- good communication skills in German
We offer you
- multinational work environment with over 1,000 colleagues from more than 50 nations
- extensive options of vocational training (i. a. expert seminars, language courses or leadership seminars)
- flexible working hours and various models to ensure the compatibility of family and career
- excellent infrastructure, including a scientific in-house library as well as modern work spaces
- an appropriate salary related to the German public tariff (TV-AVH) plus the usual social benefits for the public employment sector
The promotion of equal rights is a matter of course for us. Severely disabled persons and those equaling severely disabled persons who are equally suitable for the position will be considered preferentially within the framework of legal requirements.
Interested? Then we are looking forward to receiving your comprehensive application documents (cover letter, CV, transcripts, certificates etc.) indicating the reference number 2020/WP 16.Please start the online application process for this offered position via www.hzg.de category Career & Campus.
Application deadline: January 14th, 2021Contact: Mr. Tristan Tietz
Link: https://www.hzg.de/campus_career/vacencies/index.php.de