Home

Keynote Speakers

 

Prof. Guoping Qiu
The University of Nottingham, UK

 

 

 

Guoping Qiu is a Professor of Visual Information Processing in the School of Computer Science at the University of Nottingham, UK. His research areas are neural networks and image processing. His PhD thesis (1993) was entitled “An Investigation of Neural Networks for Image Processing Applications”and he developed some of the earliest image processing applications of neural networks - image compression (ca. 1991), learning image resolution enhancement (ca. 1996, aka super-resolution), compression artifact removal (ca. 1997) and high dynamic range (HDR) tone mapping (ca. 2004). He also developed one of the earliest forms of representation learning where he designed self-organized competitive learning models to learn image representation features (ca. 2000, aka bag-of-visual-words). He and his students have pioneered a computational approach to high dynamic range (HDR) image tone mapping in which they optimized a two-term cost function where one favors linear mapping and the other encourages uniform distribution mapping. This computational approach marked a breakthrough from traditional heuristic methods and won a best paper award at the 18th International Conference on Pattern Recognition (ICPR2006). Under this framework, his group developed some of the best practical HDR tone mapping solutions - a closed-form algorithm, a neural network solution, and a tree data structured computationally efficient technique. These methods have been widely adopted and featured in award-winning photo editing software routinely used by 100s of millions worldwide. More recently his research has been focusing on deep learning and its applications. Amongst the highlights are one of the earliest applications of deep convolutional neural networks in image processing - downscaling, decolorizing, HDR tone mapping and inverse tone-mapping/halftoning (ca. 2016), a deep feature consistent variational autoencoder (ca. 2016), and a spectral regularization algorithm for combating mode collapse in generative adversarial networks (GANs) (ca. 2018). Other applications include urban region function recognition (ca. 2019), image dehaze (ca. 2020), medical (digital pathology) image analysis (ca. 2018), depth estimation (ca. 2019), visual quality assessment (ca. 2020), inverse tone mapping HDR videos (ca. 2021), visual language modeling (ca. 2022), computational image editing and AIGC (ca. 2022), and 3D reconstruction (ca. 2023). His latest work investigates how AI can help promoting mental health and wellbeing.

 

Speech Title "High Dynamic Range – the Last Frontier of Digital Imaging"

 

Abstract: Many years of research and development plus billions of dollars investment in technology have made digital photography device ubiquitous and very advanced. Despite huge progress, there are still the occasions, for example taking a photo of an evening party at a restaurant, where the image quality will still come out poorly. Either the dark shadows are too dark such that no details are visible, or the light areas are so bright such that they are completely saturated, and no details are visible. Even after turning on the high dynamic range (HDR) function in your camera (which is now a feature in every smartphone), or adjusting the various control buttons, the situations will not improve much. And yet the photographer on the scene can see every detail both in the dark and in the bright regions. The question is, why? In this talk I will show that this difficulty is caused by the high dynamic range of the light intensities of the scene, and we call this the HDR problem. I will show from first principle that HDR is the cause of many difficulties in digital imaging (photography) and correct some of the misconceptions in many recent literatures on image processing problems such as (low-light or dark) image enhancement, especially those so-called end-to-end blackbox solutions based on deep learning. I will demonstrate both theoretically and in practice that HDR is the last technical obstacle, the last frontier, of digital imaging.

 

Prof. James Tin-Yau Kwok (IEEE Fellow)
Hong Kong University of Science and Technology, Hong Kong, China

 

 

 

James Kwok is a Professor in the Department of Computer Science and Engineering, Hong Kong University of Science and Technology. He is an IEEE Fellow.

Prof Kwok received his B.Sc. degree in Electrical and Electronic Engineering from the University of Hong Kong and his Ph.D. degree in computer science from the Hong Kong University of Science and Technology. He then joined the Department of Computer Science, Hong Kong Baptist University as an Assistant Professor. He returned to the Hong Kong University of Science and Technology and is now a Professor in the Department of Computer Science and Engineering. He is serving as an Associate Editor for the IEEE Transactions on Neural Networks and Learning Systems, Neural Networks, Neurocomputing, Artificial Intelligence Journal, International Journal of Data Science and Analytics, and on the Editorial Board of Machine Learning. He is also serving / served as Senior Area Chairs of major machine learning / AI conferences including NeurIPS, ICML, ICLR, IJCAI, and as Area Chairs of conferences including AAAI and ECML. He is on the IJCAI Board of Trustees. He is recognized as the Most Influential Scholar Award Honorable Mention for "outstanding and vibrant contributions to the field of AAAI/IJCAI between 2009 and 2019". Prof Kwok will be the IJCAI-2025 Program Chair.