Research Interests

I aim to develop state-of-the-art machine learning methods for timely and relevant applications, such as the classification of vegetation from satellite time series or the detection of marine debris in the oceans. My Ph.D. research focused on deep learning models for satellite time series classification and crop type mapping. Today, I deploy general deep learning models on a global scale which requires tackling distribution shifts with transfer learning on a variety of different applications. In particular, I train deep learning models with the model-agnostic meta-learning algorithm (MAML) to tackle a variety of different problems with few labeled samples.

Short Bio

I did my Ph.D. research at the Technical University of Munich at the Chair of Remote Sensing Technology. During my Ph.D., I had the opportunity to visited the European Space Agency and the University of Oxford as a participant in the Frontier Development Lab in 2018. Also, I could visit the IRISA Obelix Group in Vannes, France, and the Lobell Lab at Stanford. Today, I am a postdoctoral researcher at the EPFL Laboratory for Environmental Computer Science and Earth Observation (ECEO) where I develop state-of-the-art machine learning methods for meaningful and important applications.

Selected Publications

A full list of publications available via google scholar

  • End-to-end learned early classification of time series for in-season crop type mapping
  • Marc Rußwurm, Nicolas Courty, Rémi Emonet, Sébastien Lefèvre, DevisTuia, RomainTavenard
    ISPRS Journal of Photogrammetry and Remote Sensing. Volume 196, February 2023, Pages 445-456.
  • Towards detecting floating objects on a global scale with learned spatial features using Sentinel 2
  • Mifdal, J., Carmo R., Rußwurm M.
    ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., V-3-2021, 285–293, 2021, 169:421 – 435.
  • Self-attention for raw optical satellite time series classification
  • Rußwurm, M. and Körner, M.
    Self-attention for raw optical satellite time series classification. ISPRS Journal of Photogrammetry and Remote Sensing, 169:421 – 435. 2020
  • Meta-learning for few-shot land cover classification
  • Rußwurm, M., Wang, S., Korner, M., and Lobell, D.
    In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 788–796. EarthVision 2020 Best Paper Award.
  • Segmenting flooded buildings via fusion of multiresolution, multisensor, and multitemporal satellite imagery
  • Tim G. J. Rudner, Marc Rußwurm, Jakub Fil, Ramona Pelich, Benjamin Bischke, Veronika Kopačková, Piotr Bilinski
    Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 33. No. 01. 2019. 2019..
  • Multi-Temporal Land Cover Classification with Sequential Recurrent Encoders
  • Marc Rußwurm and Marco Körner
    ISPRS International Journal of Geo-Information, 2018.
  • Temporal Vegetation Modelling using Long Short-Term Memory Networks for Crop Identification from Medium-Resolution Multi-Spectral Satellite Images
  • Marc Rußwurm and Marco Körner
    In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops 2017. (best paper award)

    Apps and Links

    Some projects, I work on.

    BreizhCrops - A benchmark dataset for crop type mapping.
    Beat The MAML - An interactive interfact for few-shot land cover classification.
    Marine Debris Explorer - An interactive Earthengine app to compare the performance of different marine debris detectors.