Feature-Realistic Neural Fusion

General scene understanding for robotics requires flexible semantic representation, so that novel objects and structures which may not have been known at training time can be identified, segmented and grouped. We present an algorithm which fuses general learned features from a standard pre-trained network into a highly efficient 3D geometric neural field representation during real-time SLAM. The fused 3D feature maps inherit the coherence of the neural field's geometry representation. This means that tiny amounts of human labelling interacting at runtime enable objects or even parts of objects to be robustly and accurately segmented in an open set manner.

Overview Video

Bibtex

  @inproceedings{Mazur:etal:ICRA2023,
    title={Feature-Realistic Neural Fusion for Real-Time, Open Set Scene Understanding},
    author={Kirill Mazur and Edgar Sucar and Andrew Davison},
    booktitle={2023 International Conference on Robotics and Automation (ICRA)},
    year={2023},
  }