Visual Question Answering (VQA) usually combines visual inputs like image and video with a natural language question concerning the input and generates a natural language answer as the output. This is by nature a multi-disciplinary research problem, involving computer vision (CV), natural language processing (NLP), knowledge representation and reasoning (KR), etc.

Further, VQA is an ambitious undertaking, as it must overcome the challenges of general image understanding and the question-answering task, as well as the difficulties entailed by using large-scale databases with mixed-quality inputs. However, with the advent of deep learning (DL) and driven by the existence of advanced techniques in both CV and NLP and the availability of relevant large-scale datasets, we have recently seen enormous strides in VQA, with more systems and promising results emerging.

This book provides a comprehensive overview of VQA, covering fundamental theories, models, datasets, andpromising future directions. Given its scope, it can be used as a textbook on computer vision and natural language processing, especially for researchers and students in the area of visual question answering. It also highlights the key models used in VQA.

Autorentext

Dr. Qi Wu is Senior Lecturer at the University of Adelaide and Chief Investigator at the ARC Centre of Excellence for Robotic Vision. He is also Director of Vision-and-Language Methods at the Australian Institute for Machine Learning. Dr Wu has been in the Computer Vision field for 10 years and he has a strong track record, having pioneered the field of Vision-and-Language, one of the most interesting and technically challenging areas of Computer Vision. This area, which has emerged over the last 5 years, represents the application of computer vision technology to problems that are closer to Artificial Intelligence. Dr Wu has made breakthroughs in methods and conceptual understanding to advance the field and is recognised as an international leader in the discipline. Beyond publishing some of the seminal papers in the area, he has organised a series of workshops in CVPR, ICCV and ACL. and authored key benchmarks that define the field. Recently, he led a team that won second place in VATEX Video Captioning Challenge, the first place in both TextVQA Challenge and MedicalVQA Challenge. His achievements have been recognised with the Australian Academy of Science J G Russel Award in 2019, one of four awards to ECRs across Australia; and an NVIDIA Pioneer Research Award.

Dr. Peng Wang is Professor at the School of Computer Science, Northwestern Polytechnical University, China. He previously served at the School of Computer Science, University of Adelaide, for four years. His research interests include computer vision, machine learning, and artificial intelligence.

Dr. Xin Wang is currently Assistant Professor at the Department of Computer Science and Technology, Tsinghua University. His research interests include cross-modal multimedia intelligence and inferable recommendations in social media. He has published several high-quality research papers for top conferences including ICML, KDD, WWW, SIGIR ACM Multimedia, etc. In addition to being selected for the 2017 China Postdoctoral innovative talents supporting program, he received the ACM China Rising Star Award in 2020.

Dr. Xiaodong He is Deputy Managing Director of JD AI Research; Head of the Deep Learning, NLP and Speech Lab; and Technical Vice President of JD.com. He is also Affiliate Professor at the University of Washington (Seattle), where he serves on doctoral supervisory committees. His research interests are mainly in artificial intelligence areas including deep learning, natural language, computer vision, speech, information retrieval, and knowledge representation. He has published more than 100 papers in ACL, EMNLP, NAACL, CVPR, SIGIR, WWW, CIKM, NIPS, ICLR, ICASSP, Proc. IEEE, IEEE TASLP, IEEE SPM, and other venues. He has received several awards including the Outstanding Paper Award at ACL 2015. He is Co-inventor of the DSSM, which is now broadly applied to language, vision, IR, and knowledge representation tasks. He also led the development of the CaptionBot, the world-first image captioning cloud service, deployed in 2016. He and colleagues have won major AI challenges including the 2008 NIST MT Eval, IWSLT 2011, COCO Captioning Challenge 2015, and VQA 2017. His work has been widely integrated into influential software and services including Microsoft Image Caption Services, Bing & Ads, Seeing AI, Word, and PowerPoint. He has held editorial positions with several IEEE journals, served as Area Chair for NAACL-HLT 2015 and served on the organizing committees/program committees of major speech and language processing conferences. He is IEEE Fellow and Member of the ACL.

Wenwu Zhu is currently Professor in the Department of Computer Science and Technology at Tsinghua University and Vice Dean of National Research Center for Information Science and Technology. Prior to his current post, he was Senior Researcher and Research Manager at Microsoft Research Asia. He was Chief Scientist and Director at Intel Research China from 2004 to 2008. He worked at Bell Labs New Jersey as Member of Technical Staff during 1996-1999. He received his Ph.D. degree from New York University in 1996.

His current research interests are in the area of data-driven multimedia networking and multimedia intelligence. He has published over 350 referred papers and is Inventor or Co-inventor of over 50 patents. He received eight Best Paper Awards, including ACM Multimedia 2012 and IEEE Transactions on Circuits and Systems for Video Technology in 2001 and 2019.

He served as EiC for IEEE Transactions on Multimedia (2017-2019). He serves as Chair of the steering committee for IEEE Transactions on Multimedia, and he serves as Associate EiC for IEEE Transactions for Circuits and Systems for Video technology. He serves as General Co-Chair for ACM Multimedia 2018 and ACM CIKM 2019, respectively. He is AAAS Fellow, IEEE Fellow, SPIE Fellow, and Member of The Academy of Europe (Academia Europaea).

Inhalt

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Visual Question Answering in AI tasks . . . . . . . . . . . . . . . . . . . . . . . . 41.3 Categorisation of VQA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3.1 Classi ed by Data Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3.2 Classi ed by Task Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.3.3 Others . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.4 Book Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Part I Preliminaries2 Deep Learning Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.1 Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.2 Convolutional Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.3 Recurrent Neural Networks and variants . . . . . . . . . . . . . . . . . . . . . . . 182.4 Encoder-Decoder Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.5 Attention Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.6 Memory Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.7 Transformer Networks and BERT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.8 Graph Neural Networks Basics…