Visual Question Answering Using Deep Learning Techniques

Abstract

This thesis reviews state of the art visual question answering (VQA) algorithms and datasets, describes out work on constructing the first Arabic VQA dataset, and the details of our approach on that dataset. This thesis is divided into 7 chapters as show below Chapter 1

Gives an introduction about recent advances in integrating language and vision research to show the motivation behind working on the VQA problem. It explains the big picture of the problem, then gives an overview on the work done in this thesis. Chapter 2

Gives an overview of the different categories of datasets used to benchmark VQA algorithms and the methodology of constructing those datasets, and state of the art approaches to deal with the problem Chapter 3

Gives the basic theoretical foundations that our work is based on. It explains