Публікація:
CLIPTraVeLGAN for Semantically Robust Unpaired Image Translation

Завантаження...
Зображення мініатюри

Дата

2023

Назва журналу

ISSN журналу

Назва тома

Видавництво

Дослідницькі проекти

Організаційні підрозділи

Видання журналу

Анотація

In this paper a novel approach for semantically robust unpaired image translation is presented. CLIPTraVeLGAN replaces the Siamese network in TraVeLGAN with a contrastively pretrained language-image model (CLIP) with frozen weights. This approach significantly simplifies the model selection and training process of TraVeLGAN, making it more robust and easier to use.

Опис

The work was performed as part of the state budget research project “Development of methods and algorithms for combined learning of deep neuro-neo-fuzzy systems under short training set conditions” (state registration number 0122U001701) of Artificial Intelligence Department of Kharkiv National University of Radio Electronics.

Ключові слова

Image-to-image translation, GAN, CLIP, Transfer knowledge

Бібліографічний опис

Bodyanskiy Y. CLIPTraVeLGAN for Semantically Robust Unpaired Image Translation / Y. Bodyanskiy, N. Ryabova, R. Lavrynenko // Computational Linguistics and Intelligent Systems (COLINS-2023) : Proc. 7th Int. Conf., April 20–21, 2023. – Kharkiv, 2023. – Volume I (Machine Learning Workshop). – pp.1-12.

DOI