[TCSVT] Each Part Matters: Local Patterns Facilitate Cross-view Geo-localization

LPN

[Paper]

NEWs

We upload the codes of SAFA+ours and CVFT+ours

Prerequisites

• Python 3.6
• GPU Memory >= 8G
• Numpy > 1.12.1
• Pytorch 0.3+
• scipy == 1.2.1
• [Optional] apex (for float16) Requirements & Quick Start

Getting started

Dataset & Preparation

Download University-1652 upon request. You may use the request template.

Or download CVUSA / CVACT.

For CVUSA, I follow the training/test split in (https://github.com/Liumouliu/OriCNN).

Train & Evaluation

Train & Evaluation University-1652

sh run.sh

Train & Evaluation CVUSA

python prepare_cvusa.py  
sh run_cvusa.sh

Train & Evaluation CVACT

python prepare_cvact.py  
sh run_cvact.sh

Citation

@ARTICLE{wang2021LPN,
  title={Each Part Matters: Local Patterns Facilitate Cross-View Geo-Localization}, 
  author={Wang, Tingyu and Zheng, Zhedong and Yan, Chenggang and Zhang, Jiyong and Sun, Yaoqi and Zheng, Bolun and Yang, Yi},
  journal={IEEE Transactions on Circuits and Systems for Video Technology}, 
  year={2022},
  volume={32},
  number={2},
  pages={867-879},
  doi={10.1109/TCSVT.2021.3061265}}

@article{zheng2020university,
  title={University-1652: A Multi-view Multi-source Benchmark for Drone-based Geo-localization},
  author={Zheng, Zhedong and Wei, Yunchao and Yang, Yi},
  journal={ACM Multimedia},
  year={2020}
}

Related Work