Review of VQA : Datasets and Approaches

Visual Question Answering (VQA) is a fairly recent problem that has amassed much attention from two highly significant research fields by integrating natural language processing and computer vision techniques. It is a multimodal task in which an algorithm must answer questions about images. To begin, the paper will evaluate and analyze the various existing datasets for VQA tasks and explore their limitations and specialties. The paper then reviews various papers for novel approaches to the VQA problem. Finally, it explores the field's future scope and prospective possibilities.