VIS30K Link

This repository serves as a link to information for Graphics Replicability Stamp Initiative to the VIS30K dataset as well as the VISImageNavigator.

VIS30K Dataset

The dataset is hosted permanently on the IEEE DataPort at DOI 10.21227/4hy6-vh52 under a CC BY 4.0 license. We do not replicate the data here because it is larger than 20GB and the IEEE Data port promises an indefinite dataset storage.

In addition to being available from the IEEE DataPort, the data is also available through a direct download link from https://www.cse.osu.edu/~chen.8028/VIS30K/VIS30K.tar.gz.

Further meta data is hosted at http://go.osu.edu/vis30k. In addition, the CNN algorithm training data and validation data as well as image corpus and text corpus are also available.

VISImageNavigator

The dataset can also be explored online using the VISImageNavigator described in the paper: https://visimagenavigator.github.io/. The source code for this website is available at https://github.com/VisImageNavigator/VisImageNavigator.Release.

Paper details

@ARTICLE{Chen:2021:VCF,
  author      = {Jian Chen and Meng Ling and Rui Li and Petra Isenberg and Tobias Isenberg and Michael Sedlmair and Torsten M{\"o}ller and Robert S. Laramee and Han-Wei Shen and Katharina W{\"u}nsche and Qiru Wang},
  title       = {{VIS30K}: A Collection of Figures and Tables from {IEEE} Visualization Conference Publications},
  journal     = {IEEE Transactions on Visualization and Computer Graphics},
  year        = {2021},
  volume      = {27},
  number      = {9},
  month       = sep,
  pages       = {3826--3833},
  doi         = {10.1109/TVCG.2021.3054916},
  shortdoi    = {10/gmsvxd},
  doi_url     = {https://doi.org/10.1109/TVCG.2021.3054916},
  oa_hal_url  = {https://hal.inria.fr/hal-03123279},
  arxiv_url   = {https://arxiv.org/abs/2101.01036},
  url         = {https://visimagenavigator.github.io/},
}

Videos

A 30 second overview as well as the presentation at IEEE VIS 2021 of the approach and dataset can be found on YouTube.