Chong-Rong Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Chong-Rong Li has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Hua-Xia Rui
Tsinghua University
1 shared publication - Sheng-Ke Qiu
Tsinghua University
1 shared publication - Xuan Zhang
Tsinghua University
1 shared publication - Yang Sun
Institute for Network Sciences and Cyberspace, Tsinghua University, Beijing, China
1 shared publication
Latest Publications
- Open Access
- 1 August 2017
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 887 (1) , 012079
Multi-task convolution network for face alignment
In order to solve face alignment more effectively, we proposed a multi-task deep convolution network for face alignment, which achieves good performance even in the case of large pose variations and severe occlusion. Instead of dealing with face alignment as a single task, we jointly trained the auxiliary task of pose estimation together with face alignment to guide the distribution of facial points. By doing so we are able to 1) avoid trapping in the local optimum due to the inaccurate face boxes, 2) improve the robustness in dealing with faces with pose variation and severe occlusion. Compared with the traditional methods, our method also improves the accuracy by providing better initialization instead of mean shape. Extensive experiments show that our method has great performance on various benchmarks.
- 1 August 2017
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 887 (1) , 012079
Open AccessMulti-task convolution network for face alignment
In order to solve face alignment more effectively, we proposed a multi-task deep convolution network for face alignment, which achieves good performance even in the case of large pose variations and severe occlusion. Instead of dealing with face alignment as a single task, we jointly trained the auxiliary task of pose estimation together with face alignment to guide the distribution of facial points. By doing so we are able to 1) avoid trapping in the local optimum due to the inaccurate face boxes, 2) improve the robustness in dealing with faces with pose variation and severe occlusion. Compared with the traditional methods, our method also improves the accuracy by providing better initialization instead of mean shape. Extensive experiments show that our method has great performance on various benchmarks.
- 1 January 2006
- Published by Zhejiang University Press in Journal of Zhejiang University-SCIENCE A
- Vol. 7 (S1) , 131-136
Evaluation of packet loss impairment on streaming video
Video compression technologies are essential in video streaming application because they could save a great amount of network resources. However compressed videos are also extremely sensitive to packet loss which is inevitable in today’s best effort IP network. Therefore we think accurate evaluation of packet loss impairment on compressed video is very important. In this work, we develop an analytic model to describe these impairments without the reference of the original video (NR) and propose an impairment metric based on the model, which takes into account both impairment length and impairment strength. To evaluate an impaired frame or video, we design a detection and evaluation algorithm (DE algorithm) to compute the above metric value. The DE algorithm has low computational complexity and is currently being implemented in the real-time monitoring module of our HDTV over IP system. The impairment metric and DE algorithm could also be used in adaptive system or be used to compare different error concealment strategies.
- 1 January 2006
- Published by Zhejiang University Press in Journal of Zhejiang University-SCIENCE A
- Vol. 7 (S1) , 131-136
Evaluation of packet loss impairment on streaming video
Video compression technologies are essential in video streaming application because they could save a great amount of network resources. However compressed videos are also extremely sensitive to packet loss which is inevitable in today’s best effort IP network. Therefore we think accurate evaluation of packet loss impairment on compressed video is very important. In this work, we develop an analytic model to describe these impairments without the reference of the original video (NR) and propose an impairment metric based on the model, which takes into account both impairment length and impairment strength. To evaluate an impaired frame or video, we design a detection and evaluation algorithm (DE algorithm) to compute the above metric value. The DE algorithm has low computational complexity and is currently being implemented in the real-time monitoring module of our HDTV over IP system. The impairment metric and DE algorithm could also be used in adaptive system or be used to compare different error concealment strategies.