Abstract: Text-based Visual Question Answering (TextVQA) focuses on answering questions about the scene text in images. Most works in this field uses transformer based models to modeling the ...
Abstract: Text-to-image person retrieval (TIPR) aims to identify the target person using textual descriptions, facing challenge in modality heterogeneity. Prior works have attempted to address it by ...
一些您可能无法访问的结果已被隐去。
显示无法访问的结果