Application of ensemble Learning in visual question-answering

Abstract

Visual Question Answering (VQA) is a field that combines two different techniques: computer vision and natural language processing. Computer vision is used to process the image or video, and NLP uses for the processing of natural language. VQA is a technology that automatically answers the question based on the context of images or videos. The VQA is one of the Vision-language tasks that require a high level of language and image understanding, making this a difficult and complex problem. In this dissertation, we explore and apply an ensemble of diverse VQA models combined with Weighted Average techniques to increase the accuracy.