MARVIS: Motion & Geometry Aware Real and Virtual Image Segmentation [IROS 2024]

This work proposes a novel approach for segmentation on real and virtual image regions, exploiting synthetic images combined with domain-invariant information, a Motion Entropy Kernel, and Epipolar Geometric Consistency. Our segmentation network does not need to be re-trained if the domain changes.

This repository is the official implementation of the paper:

"MARVIS: Motion & Geometry Aware Real and Virtual Image Segmentation" [Preprint] [Poster]

Citation

@misc{wu2024marvismotiongeometry,
      title={MARVIS: Motion & Geometry Aware Real and Virtual Image Segmentation}, 
      author={Jiayi Wu and Xiaomin Lin and Shahriar Negahdaripour and Cornelia Fermüller and Yiannis Aloimonos},
      year={2024},
      eprint={2403.09850},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2403.09850}, 
}