Perceptual quality assessment has been a long-standing problem that attracts attentions from
both academia and industry. The just-noticeable-difference (JND) methodology has been proposed to
measure subjective experience of the human visual system in recent years. In this talk, I will try
to give a brief review on JND-based subjective quality measurement, objective quality assessment
metrics, and applications to perceptual coding. Then, I will present our recent work that applies
Convolutional Neural Networks (CNN) to develop objective quality metrics. To be specific, we have
developed a Full-Reference metric that uses 2D convolutional layers to extract spatial features and
Convolutional Neural Network with 3D kernels (C3D) to learn spatiotemporal features. C3DVQA
combines feature learning and score pooling into one spatiotemporal feature learning process.
What's more, I would share an on-going work that aims to develop a No-Reference IQA metric. It
adopts a coarse-to-fine multi-task learning strategy to reconstruct objective error maps in two
subtasks optimized with different loss functions. The network is designed to be nested such that
discriminative features learned from subtasks are efficiently shared by the primary task.
Experimental results are given to show the performance of the proposed methods.
About the Speaker
Haiqiang Wang received his Ph.D. degree in 2018 from the University of Southern California. From
2018 to 2019, he was a Sr. researcher at Tencent Inc., Shenzhen, China. He is currently a research
assistant professor at Peng Cheng Laboratory, Shenzhen, China. His research interests include deep
learning, image/video processing, as well as quality assessment. Dr. Wang is a recipient of the
2017 CAPOCELLI PRIZE from the Data Compression Conference (DCC).
About the Series
The Lecture Series is organized every winter term and consist of talks with international
speakers. The speakers are experts from various fields and professions. The talks are aimed at
covering the whole spectrum of visual computing and at discussing the relevance of quantification.
This years lecture series will be held entirely online.
This is an online event only.
Registration: If you would like to attend, please email
Leonel Merino to