Skip to content

Latest commit

 

History

History
31 lines (23 loc) · 1.21 KB

README.md

File metadata and controls

31 lines (23 loc) · 1.21 KB
Raw

Salienteye: Maximizing Engagement While Maintaining Artistic Style on Instagram Using Deep Neural Networks

Style similarity experiment of Paper: Salienteye: Maximizing Engagement While Maintaining Artistic Style on Instagram Using Deep Neural Networks

Requirements:

Python3 Numpy Keras 2.1.2 Tensorflow 1.4.0

Run the test

python style_test_oneforall

The default experiment is to plot the style classification confusion matrix of the following four account: thephotosociety, cats_of_instagram,travelalberta and clarklittle

Dataset

We selected the 200 most recent photos from each of the seven accounts (All collected in 2019.4 and 2019.5). Of those photos, the 100 most recent photos are used tocreate a test set and the other 100 photos are used as reference photos to represent the style of the account.

Citation

If you find this useful, please use the following citation

@inproceedings{wang2020salienteye,
  title={Salienteye: Maximizing Engagement While Maintaining Artistic Style on Instagram Using Deep Neural Networks},
  author={Wang, Lili and Liu, Ruibo and Vosoughi, Soroush},
  booktitle={Proceedings of the 2020 International Conference on Multimedia Retrieval},
  pages={331--335},
  year={2020}
}