VIXEN: Visual Text Comparison Network for Image Difference Captioning
Proceedings of the AAAI Conference on Artificial Intelligence · 24.03.2024 · 38 (2) · 846-854
DOI: 10.1609/aaai.v38i2.27843 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 24.03.2024. Verfasst von Alexander Black und 4 weiteren Autor:innen. Die Arbeit wurde bisher 10-mal zitiert (Crossref).
Abstract (Originalsprache)
We present VIXEN - a technique that succinctly summarizes in text the visual differences between a pair of images in order to highlight any content manipulation present. Our proposed network linearly maps image features in a pairwise manner, constructing a soft prompt for a pretrained large language model. We address the challenge of low volume of training data and lack of manipulation variety in existing image difference captioning (IDC) datasets by training on synthetically manipulated images from the recent InstructPix2Pix dataset generated via prompt-to-prompt editing framework. We augment this dataset with change summaries produced via GPT-3. We show that VIXEN produces state-of-the-art, comprehensible difference captions for diverse image contents and edit types, offering a potential mitigation against misinformation disseminated via manipulated image content. Code and data are available at http://github.com/alexblck/vixen
Zitieren
APA
Black, A., Shi, J., Fan, Y., Bui, T., & Collomosse, J. (2024). VIXEN: Visual Text Comparison Network for Image Difference Captioning. Proceedings of the AAAI Conference on Artificial Intelligence, 38(2), 846-854. https://doi.org/10.1609/aaai.v38i2.27843
Vancouver
Black A, Shi J, Fan Y, Bui T, Collomosse J. VIXEN: Visual Text Comparison Network for Image Difference Captioning. Proceedings of the AAAI Conference on Artificial Intelligence. 2024;38(2):846-854. doi:10.1609/aaai.v38i2.27843
BibTeX
@article{Black2024,
author = {Black, Alexander and Shi, Jing and Fan, Yifei and Bui, Tu and Collomosse, John},
title = {VIXEN: Visual Text Comparison Network for Image Difference Captioning},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2024},
volume = {38},
number = {2},
pages = {846--854},
doi = {10.1609/aaai.v38i2.27843}
}
EUVIMED