Counterfactual Dual-Bias VQA: A Multimodality Debias Learning for Robust Visual Question Answering.

已完结 10 由芳发布于 2025/5/18 19:14:20

DOI:10.1109/tnnls.2025.3562085
作者：Boyue Wang,Xiaoqian Ju,Junbin Gao,Xiaoyan Li,Yongli Hu,Baocai Yin
文献类型：期刊论文
补充材料：只需要正文
文献链接：https://doi.org/10.1109/tnnls.2025.3562085

Institute of Electrical and Electronics Engineers (IEEE)

Abstract：Visual question answering (VQA) models often face two language bias challenges. First, they tend to rely solely on the question to predict the answer, often overlooking relevant information

应助信息

求助已完结，文件已删除 如有需求请重新发布求助一键发布

感谢使用Book学术
本站所有文献仅供个人学习和参考，请勿将文件进行传播。共同遵守网站规定和相关知识产权规定。

上一个求助:其它数据库Naturally Fractured Carbonate Reservoir Characterization: A Case Study of a Mature High-Pour Point Oil Field in Hungary 下一个求助:ElsevierInvestigation into the synergistic roles of exposed crystal planes and S vacancies during the photo-Fenton degradation of sulfadiazine via pyrite with three different morphologies