Skip to content

DynaSLAM is a SLAM system robust in dynamic environments for monocular, stereo and RGB-D setups

License

Notifications You must be signed in to change notification settings

BertaBescos/DynaSLAM

Repository files navigation

DynaSLAM

[Project] [arXiv] [Journal]

DynaSLAM is a visual SLAM system that is robust in dynamic scenarios for monocular, stereo and RGB-D configurations. Having a static map of the scene allows inpainting the frame background that has been occluded by such dynamic objects.

DynaSLAM: Tracking, Mapping and Inpainting in Dynamic Scenes
Berta Bescos, José M. Fácil, Javier Civera and José Neira
RA-L and IROS, 2018

We provide examples to run the SLAM system in the TUM dataset as RGB-D or monocular, and in the KITTI dataset as stereo or monocular.

News

  • DynaSLAM supports now both OpenCV 2.X and OpenCV 3.X.

Getting Started

git clone https://github.com/BertaBescos/DynaSLAM.git
cd DynaSLAM
cd DynaSLAM
chmod +x build.sh
./build.sh
  • Place the mask_rcnn_coco.h5 model in the folder DynaSLAM/src/python/.

RGB-D Example on TUM Dataset

These associations files are given in the folder ./Examples/RGB-D/associations/ for the TUM dynamic sequences.

  • Execute the following command. Change TUMX.yaml to TUM1.yaml,TUM2.yaml or TUM3.yaml for freiburg1, freiburg2 and freiburg3 sequences respectively. Change PATH_TO_SEQUENCE_FOLDER to the uncompressed sequence folder. Change ASSOCIATIONS_FILE to the path to the corresponding associations file. PATH_TO_MASKS and PATH_TO_OUTPUT are optional parameters.

    ./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUMX.yaml PATH_TO_SEQUENCE_FOLDER ASSOCIATIONS_FILE (PATH_TO_MASKS) (PATH_TO_OUTPUT)
    

If PATH_TO_MASKS and PATH_TO_OUTPUT are not provided, only the geometrical approach is used to detect dynamic objects.

If PATH_TO_MASKS is provided, Mask R-CNN is used to segment the potential dynamic content of every frame. These masks are saved in the provided folder PATH_TO_MASKS. If this argument is no_save, the masks are used but not saved. If it finds the Mask R-CNN computed dynamic masks in PATH_TO_MASKS, it uses them but does not compute them again.

If PATH_TO_OUTPUT is provided, the inpainted frames are computed and saved in PATH_TO_OUTPUT.

Stereo Example on KITTI Dataset

  • Download the dataset (grayscale images) from http://www.cvlibs.net/datasets/kitti/eval_odometry.php

  • Execute the following command. Change KITTIX.yamlto KITTI00-02.yaml, KITTI03.yaml or KITTI04-12.yaml for sequence 0 to 2, 3, and 4 to 12 respectively. Change PATH_TO_DATASET_FOLDER to the uncompressed dataset folder. Change SEQUENCE_NUMBER to 00, 01, 02,.., 11. By providing the last argument PATH_TO_MASKS, dynamic objects are detected with Mask R-CNN.

./Examples/Stereo/stereo_kitti Vocabulary/ORBvoc.txt Examples/Stereo/KITTIX.yaml PATH_TO_DATASET_FOLDER/dataset/sequences/SEQUENCE_NUMBER (PATH_TO_MASKS)

Monocular Example on TUM Dataset

  • Download a sequence from http://vision.in.tum.de/data/datasets/rgbd-dataset/download and uncompress it.

  • Execute the following command. Change TUMX.yaml to TUM1.yaml,TUM2.yaml or TUM3.yaml for freiburg1, freiburg2 and freiburg3 sequences respectively. Change PATH_TO_SEQUENCE_FOLDERto the uncompressed sequence folder. By providing the last argument PATH_TO_MASKS, dynamic objects are detected with Mask R-CNN.

./Examples/Monocular/mono_tum Vocabulary/ORBvoc.txt Examples/Monocular/TUMX.yaml PATH_TO_SEQUENCE_FOLDER (PATH_TO_MASKS)

Monocular Example on KITTI Dataset

  • Download the dataset (grayscale images) from http://www.cvlibs.net/datasets/kitti/eval_odometry.php

  • Execute the following command. Change KITTIX.yamlby KITTI00-02.yaml, KITTI03.yaml or KITTI04-12.yaml for sequence 0 to 2, 3, and 4 to 12 respectively. Change PATH_TO_DATASET_FOLDER to the uncompressed dataset folder. Change SEQUENCE_NUMBER to 00, 01, 02,.., 11. By providing the last argument PATH_TO_MASKS, dynamic objects are detected with Mask R-CNN.

./Examples/Monocular/mono_kitti Vocabulary/ORBvoc.txt Examples/Monocular/KITTIX.yaml PATH_TO_DATASET_FOLDER/dataset/sequences/SEQUENCE_NUMBER (PATH_TO_MASKS)

Citation

If you use DynaSLAM in an academic work, please cite:

@article{bescos2018dynaslam,
  title={{DynaSLAM}: Tracking, Mapping and Inpainting in Dynamic Environments},
  author={Bescos, Berta, F\'acil, JM., Civera, Javier and Neira, Jos\'e},
  journal={IEEE RA-L},
  year={2018}
 }

Acknowledgements

Our code builds on ORB-SLAM2.

DynaSLAM

About

DynaSLAM is a SLAM system robust in dynamic environments for monocular, stereo and RGB-D setups

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages