GPS2Vec+: Learning Multi-context Aware Location Representations from Large-scale Geotagged Images

Pretrained models for our papers:

Learning Multi-context Aware Location Representations from Large-scale Geotagged Images

GPS2Vec: Towards Generating Worldwide GPS Embeddings

Requirements

• numpy
• math
• utm
• keras

Usage

Download Pre-trained Models

models_tag Semantic context trained on the 1M Flickr dataset.

models_visual Visual context trained on the YLI-GEO dataset.

Citation

If you use the pre-trained models please cite our papers.

@inproceedings{yin2021-gps2vec+,
  title={Learning Multi-context Aware Location Representations from Large-scale Geotagged Images},
  author={Yin, Yifang and Zhang, Ying and Liu, Zhenguang and Liang, Yuxuan and Wang, Sheng and Shah, Rajiv Ratn and Zimmermann, Roger},
  year={2021},
  booktitle={Proceedings of the 29th ACM International Conference on Multimedia},
  pages = {899–-907},
}

@inproceedings{yin2019-gps2vec,
author = {Yin, Yifang and Liu, Zhenguang and Zhang, Ying and Wang, Sheng and Shah, Rajiv Ratn and Zimmermann, Roger},
title = {GPS2Vec: Towards Generating Worldwide GPS Embeddings},
year = {2019},
booktitle = {Proceedings of the 27th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems},
pages = {416–-419},
}