Note A presentation can be seen here. I will publish an accompanying video soon.
Following a rapid development in state-of-the-art image capitoning models the goal of this master thesis is to train a neural network inspired by SemStyle paper to achieve image captioning in Shakespearian style. For instance given an image:
The network should be able to generate a caption along the lines of:
A fearless hero facing down a horde of his most terrifying enemies with his beloved football
More results coming soon!
Classic: A bird sitting on a tree branch
Shakespearian: Thou didst usurp the night owl in a forest and bid the tree
- http://openaccess.thecvf.com/content_cvpr_2017/papers/Gan_StyleNet_Generating_Attractive_CVPR_2017_paper.pdf
- http://openaccess.thecvf.com/content_ECCV_2018/papers/Tianlang_Chen_Factual_or_Emotional_ECCV_2018_paper.pdf
- http://openaccess.thecvf.com/content_CVPR_2019/papers/Guo_MSCap_Multi-Style_Image_Captioning_With_Unpaired_Stylized_Text_CVPR_2019_paper.pdf