Research for image caption based on global attention mechanism

Tong Wu; Tao Ku; Hao Zhang

doi:10.1117/12.2552711

31 January 2020 Research for image caption based on global attention mechanism

Tong Wu, Tao Ku, Hao Zhang

Proceedings Volume 11427, Second Target Recognition and Artificial Intelligence Summit Forum; 114272U (2020) https://doi.org/10.1117/12.2552711
Event: Second Target Recognition and Artificial Intelligence Summit Forum, 2019, Changchun, China

Abstract

Convolution Neural Networks (CNN) and Recurrent Neural Networks (RNN), which are the main research methods of image caption, have developed rapidly. Nevertheless lacking of global consciousness in image caption has not been completely solved. Separation from the bottom-up visual attention mechanism and the top-down visual attention mechanism has been widely used in image description and visual question and answers. In this article, we put forward image description based on a global attention mechanism research methods. The global attention prior channel is added to the infrastructure to extract the global information features while learning the local features. The attention to the object and other outstanding level of image region is calculated, and the global image features are enhanced.

Citation Download Citation

Tong Wu, Tao Ku, and Hao Zhang "Research for image caption based on global attention mechanism", Proc. SPIE 11427, Second Target Recognition and Artificial Intelligence Summit Forum, 114272U (31 January 2020); https://doi.org/10.1117/12.2552711

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available