Exploring duality in visual question-driven top-down saliency
Publication Type
Journal Article
Publication Date
7-2020
Abstract
Top-down, goal-driven visual saliency exerts a huge influence on the human visual system for performing visual tasks. Text generations, like visual question answering (VQA) and visual question generation (VQG), have intrinsic connections with top-down saliency, which is usually involved in both VQA and VQG processes in an unsupervised manner. However, it is shown that the regions that humans choose to look at to answer questions are very different from the unsupervised attention models. In this brief, we aim to explore the intrinsic relationship between top-down saliency and text generations, and to figure out whether an accurate saliency response benefits text generation. To this end, we propose a dual supervised network with dynamic parameter prediction. Dual-supervision explicitly exploits the probabilistic correlation between the primal task top-down saliency detection and the dual task text generation, while dynamic parameter prediction encodes the given text (i.e., question or answer) into the fully convolutional network. Extensive experiments show the proposed top-down saliency method achieves the best correlation with human attention among various baselines. In addition, the proposed model can be guided by either questions or answers, and output the counterpart. Furthermore, we show that combining human-like visual question-saliency improves the performance of both answer and question generations.
Keywords
Task analysis, Visualization, Feature extraction, Training, Pipelines, Learning systems, Knowledge discovery, Dual learning, saliency, visual question answering (VQA), visual question generation (VQG)
Discipline
Information Security
Research Areas
Intelligent Systems and Optimization
Publication
IEEE Transactions on Neural Networks and Learning Systems
Volume
31
Issue
7
First Page
2672
Last Page
2679
ISSN
2162-237X
Identifier
10.1109/TNNLS.2019.2933439
Publisher
Institute of Electrical and Electronics Engineers
Citation
HE, Shengfeng; HAN, Chu; HAN, Guoqiang; and QIN, Jing.
Exploring duality in visual question-driven top-down saliency. (2020). IEEE Transactions on Neural Networks and Learning Systems. 31, (7), 2672-2679.
Available at: https://ink.library.smu.edu.sg/sis_research/7857
Additional URL
https://doi.org/10.1109/TNNLS.2019.2933439