Assessing Medical Optical Devices with INNs – joint work with Lena Maier Hein’s team

Paper accepted to IPCAI 2019. Find arxiv paper here https://arxiv.org/abs/1903.03441 Uncertainty-aware performance assessment of optical imaging modalities with invertible neural networks. Tim J. Adler, Lynton Ardizzone, Anant Vemuri, Leonardo Ayala, Janek Gröhl, Thomas Kirchner, Sebastian Wirkert, Jakob Kruse, Carsten Rother, Ullrich Köthe, Lena Maier-Hein. IPCAI 2019.

More Info

CVPR 19 Paper on Panoptic Segmentation accepted

Panoptic Segmentation, Alexander Kirillov, Kaiming He, Ross Girshick, Carsten Rother, Piotr Dollar, CVPR 2019 arxiv version https://arxiv.org/abs/1801.00868

More Info

Check out our ICLR Paper and Blog on Analyzing Inverse Problems with Invertible Neural Networks

Here is a gentle introduction to our Invertible Neural Network architecture to tackle ambiguous inverse problems. [Update: Accepted at ICLR 2019!]

More Info

Our excellent cluster “Structures” – where ML is used to find structures in data and the physical world – got funded by DFG

STRUCTURES: A unifying approach to emergent phenomena in the physical world, mathematics, and complex data.

More Info

“CEREALS – Cost-Effective REgion-based Active Learning for Semantic Segmentation”

Check out our paper in cooperation with the Robert Bosch GmbH. [pdf] [suppl.]

More Info

Geometric Image Synthesis

Check our new paper “Geometric Image Synthesis” [ArXiv] [video]

More Info

Robust Vision Challenge at CVPR 2018

We are happy to be involved in the organization of the Robust Vision Challenge at CVPR 2018 together with Andreas Geiger and a great team from MPI Tübingen, ETH Zurich and TU Munich. Visit http://www.robustvision.net for more information and stay tuned for details on how to participate with your stereo, flow, multi-view or segmentation algorithm!

More Info

DSAC – Differentiable RANSAC for Camera Localization

Ever wondered how to train a computer vision pipeline, which contains RANSAC, in an end-to-end fashion? See our project page or  arXiv.

More Info