Skip to content
Code for "Prior Guided Dropout for Robust Visual Localization in Dynamic Environments" in ICCV 2019
Python Shell
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
attentionmapnet
common copyright for files Jan 2, 2020
datainfo datainfo for training Nov 21, 2019
dataset_loader copyright for files Jan 2, 2020
experiments Add copyright Dec 19, 2019
mapnet copyright for files Jan 2, 2020
posenet copyright for files Jan 2, 2020
LICENSE.md Update LICENSE.md Nov 19, 2019
README.md Update acknowledgements Dec 19, 2019
environment.yml env.yml Nov 21, 2019

README.md

Prior Guided Dropout for Robust Visual Localization in Dynamic Environments

Prior Guided Dropout for Robust Visual Localization in Dynamic Environments
Zhaoyang Huang, Yan Xu, Jianping Shi, Xiaowei Zhou, Hujun Bao, Guofeng Zhang

License

Licensed under the CC BY-NC-SA 4.0 license, see LICENSE.

Citation

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

@inproceedings{huang2019prior,
  title={Prior Guided Dropout for Robust Visual Localization in Dynamic Environments},
  author={Huang, Zhaoyang and Xu, Yan and Shi, Jianping and Zhou, Xiaowei and Bao, Hujun and Zhang, Guofeng},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision},
  pages={2791--2800},
  year={2019}
}

Environment

PGD-MapNet uses Conda to setup the environment

conda env create -f environment.yml
conda activate pgd-mapnet

The data is processed as suggested in geomapnet. The dynamic information computed from Mask_RCNN is stored in datainfo. The files should be put into the corresponding root dir of each scene.

Running

Training

cd experiments
bash runattmapnet.sh

Evaluation

cp logs/exp_beta[-3.0]gamma[-3.0]batch_size[64]model[attentionmapnet]mask_sampling[True]sampling_threshold[0.2]color_jitter[0.0]uniform_sampling[False]mask_image[False]dataset[RobotCar]scene[full]/config.json admapfull.json
bash run_eval.sh

Acknowledgements

Our code partially builds on geomapnet.

The work is affliated with ZJU-SenseTime Joint Lab of 3D Vision, and its intellectual property belongs to SenseTime Group Ltd.

You can’t perform that action at this time.