Dr. Marc A. Kastner

About me

Action Semantic Alignment for Image Captioning

Back to publications

Authors: Da Huo, Marc A. Kastner, Takahiro Komamizu, Ichiro Ide

Abstract:

Image captioning is one of the main goals in vision and language processing, which aims to generate proper descriptions of images. Recently, the attention mechanisms became crucial in captioning tasks, as they can capture global dependencies between modalities. Moreover, research have used objects detected from the input image as anchor points, so called object tags, to ease such alignments resulting in good performance for this task. In this paper, we newly introduce action information as a prior to further improve this. We propose adding action tags for training in image captioning. The action tags can learn alignment at action semantic level and catch the previously ignored dimension of action, that could be very important in image captioning. We found that adjusting action tags can be used to describe images in a dynamic style. Furthermore, we found it can actually lead to a significant enhancement compared with other methods in captioning performance for common metrics.

Type: 5th IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR2022)

Publication date: August 2022

DOI: 10.1109/MIPR54900.2022.00041


If you have questions or ideas about this research, feel free to leave a comment below or send me an email. I will reply quickly.
© 2013-2023 Marc A. Kastner. Powered by KirbyCMS. Some rights reserved. Privacy policy.