There is a key problem in the medical visual question answering task that how to effectively realize the feature fusion of language and medical images with limited datasets. In order to better utilize multi-scale information of medical images, previous methods directly embed the multi-stage visual feature maps as tokens of same size respectively and fuse them with text representation. However, this will cause the confusion of visual features at different stages. To this end, we propose a simple but powerful multi-stage feature fusion method, MF2-MVQA, which stage-wise fuses multi-level visual features with textual semantics. MF2-MVQA achieves the State-Of-The-Art performance on VQA-Med 2019 and VQA-RAD dataset. The results of visualization also verify that our model outperforms previous work.

Modern face recognition systems (FRS) still fall short when the subjects are wearing facial masks, a common theme in the age of respiratory pandemics. An intuitive partial remedy is to add a mask detector to flag any masked faces so that the FRS can act accordingly for those low-confidence masked faces. In this work, we set out to investigate the potential vulnerability of such FRS, equipped with a mask detector, on large-scale masked faces. As existing face recognizers and mask detectors have high performance in their respective tasks, it is a challenge to simultaneously fool them and preserve the transferability of the attack. To this end, we devise realistic facial masks that exhibit partial face patterns (i.e., faced masks) and stealthily add adversarial textures that can not only lead to significant performance deterioration of the SOTA deep learning-based FRS, but also remain undetected by the SOTA facial mask detector, thus successfully fooling both systems at the same time. The proposed method unveils the vulnerability of the FRS when dealing with masked faces wearing faced masks.