Most existing Visual Question Answering (VQA) models overly rely on language priors between questions and answers. In this paper, we present a novel method of language attention-based VQA that learns decomposed linguistic representations of questions and utilizes the representations to infer answers for overcoming language priors. We introduce a modular language attention mechanism to parse a question into three phrase representations: type representation, object representation, and concept representation. We use the type representation to identify the question type and the possible answer set (yes/no or specific concepts such as colors or numbers), and the object representation to focus on the relevant region of an image. The concept representation is verified with the attended region to infer the final answer. The proposed method decouples the language-based concept discovery and vision-based concept verification in the process of answer inference to prevent language priors from dominating the answering process. Experiments on the VQA-CP dataset demonstrate the effectiveness of our method.
Abstract. For 304L large structural parts used in nuclear power, it is hard and costly to fabricate and machine traditionally. Wire arc additive manufacturing (WAAM) has low cost and high material utilization, which provides an efficient way to fabricate the large structural parts. So in this study, WAAM is used to fabricate the parts of 304L stainless steel. Through the tensile test and metallographic analysis, the mechanical properties and microstructure of the 304L stainless steel fabricated by WAAM were explored. The results indicate that with the layers depositing, the cooling rate becomes slower, the dendrites become thicker and the morphology becomes more stable. Due to the existence of dendrites, the grain boundary strengthening effect is different between the transverse direction and longitudinal direction, and resulting in anisotropy of mechanical properties. However, the mechanical properties of the parts correspond to the forged piece, which lays the foundation for future applications.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.