Jiang, Tingting is an associate professor in the School of Computer Science. She has served at Peking University since 2014. She received the B.S. degree in computer science from University of Science and Technology of China in Hefei, China, in 2001 and the Ph.D. degree in computer science from Duke University, Durham, North Carolina, USA, in 2007. Her research interests include computer vision, image and video quality assessment (IQA/VQA).
Dr. Jiang has published more than 30 research papers, and many of them are published in top-tier conferences and journals, such as CVPR, ICCV and IJCV. She has served in the Technical Program Committee of various international conferences including ICME and ICSI. She has served as the leading guest editor for the special issue of Sensing and Imaging. She is serving in the Editorial Board of Sensing and Imaging. She has also served in the CCF NOI scientific committee since 2014.
Dr. Jiang has more than ten research projects including NSFC, 973 programs, etc. Her research achievements are summarized as follows:
1) Shape analysis:
Shape is an important cue in computer vision. First, she learned class-specific shape models from images and then applied this shape prior knowledge for shape reconstruction and object detection as well as localization. Second, she studied how to segment shapes into parts for building part-based shape models for objects.
2) Sampling for subjective image quality assessment based on pairwise comparisons:
Pairwise comparisons have been adopted for IQA because it is more stable than absolute 5-scale testing method. However, the paired comparison approach leaves a heavier burden on participants with a larger number of comparisons. So it is natural to collect subjective test data via crowdsourcing. She focused on designing an efficient sampling strategy for IQA online. First, she proposed a random sampling strategy and later proposed an active sampling approach based on the similarity between images in order to optimize the effort devoted for subjective test.
3) Objective image/video quality assessment:
How to design efficient features for IQA is an important problem for feature-based IQA methods. She first manually designed some features for both IQA and VQA. Recently with the development of machine learning, she has proposed some IQA methods based on deep learning. For 3D image, she proposed a JND model based on depth perception. For 3D videos, she studied the JND model and QoE model based on the structure of low-level features in the two views, and proposed a new model for 3D VQA. For image enhancement, she also proposed an objective model to evaluate the enhanced image quality.
