Dataset Description

A large dataset containing millions of pairs of ground-level and aerial/satellite images from across the United States.

Dataset Download

The CVUSA dataset is available as five files form Google Drive.

• metadata.tar: A README and an example script.
• flickr.tar: ground-level images from Flickr
• flickr_aerial.tar: aerial images associated with the Flickr images
• streetview.tar: ground-level images from Google Street View
• streetview_aerial.tar: aerial images associated with the Google Street View images

This subset of CVUSA is often used for localization evaluation. If you use this subset, please also cite the Zhai et al. CVPR 2017 paper.

• git repo
• bingmap.tar.gz
• splits.tar.gz
• streetview.tar.gz

Disclaimer: The street-view panorama imagery is owned and copyrighted by Google Inc. The flickr images are owned by Flickr, Inc. The aerial images are from Bing Maps and are subject to their Terms of Service. For commercial usage of the dataset, one must seek explicit permissions from each of these rightful owners.

Dataset Citation

If you use the dataset, please cite the following publications:

1. Workman S, Souvenir R, Jacobs N. 2015. Wide-Area Image Geolocalization with Aerial Reference Imagery. In: IEEE International Conference on Computer Vision (ICCV). 1–9. DOI: 10.1109/ICCV.2015.451.
   bibtex | website | code | doi

   ×Close

   Bibtex Entry for "Wide-Area Image Geolocalization with Aerial Reference Imagery"

   @inproceedings{workman2015localize,
     annotation = {spotlight,dataset_cvusa,waldo,nvidia,remote_sensing,social_media,localization},
     author = {Workman, Scott and Souvenir, Richard and Jacobs, Nathan},
     booktitle = {IEEE International Conference on Computer Vision (ICCV)},
     code = {https://github.com/scottworkman/deeplyfound},
     pages = {1--9},
     pdf = {http://www.cv-foundation.org/openaccess/content_iccv_2015/papers/Workman_Wide-Area_Image_Geolocalization_ICCV_2015_paper.pdf},
     project = {http://cs.uky.edu/%7escott/research/deeplyfound/},
     thumbnail = {http://cs.uky.edu/%7ejacobs/thumbnails/crossview.jpg},
     title = {Wide-Area Image Geolocalization with Aerial Reference Imagery},
     doi = {10.1109/ICCV.2015.451},
     note = {Acceptance rate: 30.3\%},
     year = {2015}
   }
   

Related Publication(s)

1. Salem T, Workman S, Jacobs N. 2020. Learning a Dynamic Map of Visual Appearance. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR). DOI: 10.1109/CVPR42600.2020.01245.
   bibtex | website | doi | tweet

   ×Close

   Bibtex Entry for "Learning a Dynamic Map of Visual Appearance"

   @inproceedings{salem2020dynamic,
     author = {Salem, Tawfiq and Workman, Scott and Jacobs, Nathan},
     booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
     title = {Learning a Dynamic Map of Visual Appearance},
     doi = {10.1109/CVPR42600.2020.01245},
     year = {2020},
     annotation = {google_crossview,localization,career,remote_sensing,spotlight,timelapse,cvusa},
     date-modified = {2020-03-31 16:54:47 -0400},
     project = {https://tsalem.github.io/DynamicMaps/},
     note = {Acceptance rate: 25\%},
     pdf = {http://openaccess.thecvf.com/content_CVPR_2020/papers/Salem_Learning_a_Dynamic_Map_of_Visual_Appearance_CVPR_2020_paper.pdf},
     video = {https://www.youtube.com/watch?v=OVTvgWLlJmM},
     tweet = {https://twitter.com/jacobsn/status/1272637141797961729},
     thumbnail = {http://mvrl.cs.uky.edu/thumbnails/dynamic_visual.jpg}
   }
   

2. Greenwell C, Workman S, Jacobs N. 2019. Implicit Land Use Mapping Using Social Media Imagery. In: IEEE Applied Imagery and Pattern Recognition (AIPR). DOI: 10.1109/AIPR47015.2019.9174570.
   bibtex | doi

   ×Close

   Bibtex Entry for "Implicit Land Use Mapping Using Social Media Imagery"

   @inproceedings{greenwell2019implicit,
     annotation = {abstract,google_crossview,amazon,nvidia,remote_sensing,social_media,cvusa},
     author = {Greenwell, Connor and Workman, Scott and Jacobs, Nathan},
     booktitle = {IEEE Applied Imagery and Pattern Recognition (AIPR)},
     title = {Implicit Land Use Mapping Using Social Media Imagery},
     note = {(oral)},
     doi = {10.1109/AIPR47015.2019.9174570},
     year = {2019}
   }
   

3. Salem T, Greenwell C, Blanton H, Jacobs N. 2019. Learning to Map Nearly Anything. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS). DOI: 10.1109/IGARSS.2019.8900646.
   bibtex | doi

   ×Close

   Bibtex Entry for "Learning to Map Nearly Anything"

   @inproceedings{salem2019anything,
     annotation = {career,remote_sensing,social_media,cvusa},
     author = {Salem, Tawfiq and Greenwell, Connor and Blanton, Hunter and Jacobs, Nathan},
     booktitle = {IEEE International Geoscience and Remote Sensing Symposium (IGARSS)},
     title = {Learning to Map Nearly Anything},
     note = {(oral)},
     doi = {10.1109/IGARSS.2019.8900646},
     year = {2019}
   }
   

4. Greenwell C, Workman S, Jacobs N. 2018. What Goes Where: Predicting Object Distributions from Above. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS). DOI: 10.1109/IGARSS.2018.8519251.
   bibtex | website | doi

   ×Close

   Bibtex Entry for "What Goes Where: Predicting Object Distributions from Above"

   @inproceedings{greenwell2018objects,
     annotation = {career,remote_sensing,social_media,cvusa},
     author = {Greenwell, Connor and Workman, Scott and Jacobs, Nathan},
     booktitle = {IEEE International Geoscience and Remote Sensing Symposium (IGARSS)},
     project = {https://connorgreenwell.github.io/publication/what_goes_where/},
     title = {What Goes Where: Predicting Object Distributions from Above},
     pdf = {https://arxiv.org/abs/1808.00995},
     doi = {10.1109/IGARSS.2018.8519251},
     year = {2018}
   }
   

5. Zhai M, Bessinger Z, Workman S, Jacobs N. 2017. Predicting Ground-Level Scene Layout from Aerial Imagery. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR). DOI: 10.1109/CVPR.2017.440.
   bibtex | code | doi | tweet

   ×Close

   Bibtex Entry for "Predicting Ground-Level Scene Layout from Aerial Imagery"

   @inproceedings{zhai2017layout,
     annotation = {google_crossview,localization,spotlight,nvidia,amazon,career,remote_sensing,cvusa},
     author = {Zhai, Menghua and Bessinger, Zach and Workman, Scott and Jacobs, Nathan},
     booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
     code = {https://github.com/viibridges/crossnet},
     doi = {10.1109/CVPR.2017.440},
     note = {Acceptance rate: 29.2\%},
     pdf = {http://arxiv.org/abs/1612.02709},
     tweet = {https://twitter.com/jacobsn/status/837012306814980097},
     thumbnail = {http://cs.uky.edu/%7ejacobs/thumbnails/predict_layout.jpg},
     title = {Predicting Ground-Level Scene Layout from Aerial Imagery},
     year = {2017}
   }
   

6. Workman S, Souvenir R, Jacobs N. 2015. Wide-Area Image Geolocalization with Aerial Reference Imagery. In: IEEE International Conference on Computer Vision (ICCV). 1–9. DOI: 10.1109/ICCV.2015.451.
   bibtex | website | code | doi

   ×Close

   Bibtex Entry for "Wide-Area Image Geolocalization with Aerial Reference Imagery"

   @inproceedings{workman2015localize,
     annotation = {spotlight,dataset_cvusa,waldo,nvidia,remote_sensing,social_media,localization},
     author = {Workman, Scott and Souvenir, Richard and Jacobs, Nathan},
     booktitle = {IEEE International Conference on Computer Vision (ICCV)},
     code = {https://github.com/scottworkman/deeplyfound},
     pages = {1--9},
     pdf = {http://www.cv-foundation.org/openaccess/content_iccv_2015/papers/Workman_Wide-Area_Image_Geolocalization_ICCV_2015_paper.pdf},
     project = {http://cs.uky.edu/%7escott/research/deeplyfound/},
     thumbnail = {http://cs.uky.edu/%7ejacobs/thumbnails/crossview.jpg},
     title = {Wide-Area Image Geolocalization with Aerial Reference Imagery},
     doi = {10.1109/ICCV.2015.451},
     note = {Acceptance rate: 30.3\%},
     year = {2015}
   }
   

7. Workman S, Jacobs N. 2015. On the Location Dependence of Convolutional Neural Network Features. In: IEEE/ISPRS Workshop: Looking from above: When Earth observation meets vision (EARTHVISION). 1–9. DOI: 10.1109/CVPRW.2015.7301385.
   bibtex | doi

   ×Close

   Bibtex Entry for "On the Location Dependence of Convolutional Neural Network Features"

   @inproceedings{workman2015geocnn,
     annotation = {waldo,localization,calibration,cvusa,social_media,remote_sensing},
     author = {Workman, Scott and Jacobs, Nathan},
     booktitle = {IEEE/ISPRS Workshop: Looking from above: When Earth observation meets vision (EARTHVISION)},
     note = {Acceptance rate: 30\%},
     pages = {1--9},
     pdf = {http://www.cv-foundation.org/openaccess/content_cvpr_workshops_2015/W13/papers/Workman_On_the_Location_2015_CVPR_paper.pdf},
     title = {On the Location Dependence of Convolutional Neural Network Features},
     doi = {10.1109/CVPRW.2015.7301385},
     year = {2015}
   }