Datasets:

lmb-freiburg
/

entitynet

license: mit
language:
  - en
pretty_name: EntityNet
size_categories:
  - 10M<n<100M
task_categories:
  - image-to-text
  - text-to-image
  - image-classification
  - visual-question-answering

EntityNet: Using Knowledge Graphs to harvest datasets for efficient CLIP model training

Dataloader and instructions can be found here: https://github.com/lmb-freiburg/entitynet

If you use the dataset, code, or results, please cite:

@inproceedings{ging2025entitynet,
  author    = {Simon Ging and Sebastian Walter and Jelena Bratuli{\'c} and Johannes Dienert and Hannah Bast and Thomas Brox},
  title     = {Using Knowledge Graphs to Harvest Datasets for Efficient {CLIP} Model Training},
  booktitle = {Pattern Recognition, 47th {DAGM} German Conference, {DAGM} {GCPR} 2025, Freiburg, Germany, September 23--26, 2025, Proceedings},
  series    = {Lecture Notes in Computer Science},
  publisher = {Springer Nature Switzerland},
  year      = {2025},
  pages     = {287--302},
  isbn      = {978-3-032-12840-9},
  doi       = {10.1007/978-3-032-12840-9_19}
}