Multi-Granularity Feature Interaction and Multi-Region Selection Based Triplet Visual Question Answering

Accurately locating the question-related regions in one given image is crucial for visual question answering (VQA). The current approaches suffer two limitations: (1) Dividing one image into multiple regions may lose parts of semantic information and original relationships between regions; (2) Choosing only one or all image regions to predict the answer may correspondingly result in the insufficiency or redundancy of information. Therefore, how to effectively mine the relationship between image regions and choose the relevant image regions are vital. In this paper, we propose a novel <b>M</b>ulti-granularity feature interaction and <b>M</b>ulti-region selection-based triplet VQA model (M2TVQA). To tackle the first limitation, we propose the multi-granularity feature interaction strategy that adaptively supplements the global coarse-granularity features with the regional fine-granularity features. To overcome the second limitation, we design the Top-<inline-formula><tex-math notation="LaTeX">$K$</tex-math></inline-formula> learning strategy to adaptively select <inline-formula><tex-math notation="LaTeX">$K$</tex-math></inline-formula> most relevant image regions to the question, even if the selected regions are far away in space. Such a strategy can select as many relevant image regions as possible and reduce introducing noise. Finally, we construct the multi-modality triplet to predict the answer of VQA. Extended experiments on two public outside knowledge datasets OK-VQA and KRVQA verify the effectiveness of the proposed model.