Home

Awesome

Geometry-Aware Learning of Maps for Camera Localization

MapNet This is the project page for the CVPR 2018 Spotlight paper "Geometry-Aware Learning of Maps for Camera Localization". Our algorithm MapNet allows you to estimate the pose of a camera in a known scene, from a sequence of images, in a completely data-driven manner. More importantly, it enables you to use unsupervised videos from that scene to continue improving the deep network.

Paper | Supplementary Material

Code and Models

Project Page at Nvidia | Github | Trained PyTorch Models

Citation

If you find this code useful for your research, please cite our paper

@inproceedings{mapnet2018,
  title={Geometry-Aware Learning of Maps for Camera Localization},
  author={Samarth Brahmbhatt and Jinwei Gu and Kihwan Kim and James Hays and Jan Kautz},
  booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2018}
}

Supplementary Material

Supplementary Video

Supplementary Video

Attention Maps Video

Attention Maps