About Us

SIAT Video Group (SVG) is comprised of members from the High Performance Computing Center (HPCC) in the Shenzhen Institutes of Advanced Technology (SIAT) of the Chinese Academy of Sciences (CAS). We have long been engaged in the field of Multimedia Communications and Visual Signal Processing for 2D/3D, VR/AR videos, including video coding, visual signal pre/post-processing, and computational visual perception. We are also pursuing exciting and challenging problems in the innovation areas, such as VR/AR, and AI etc.

Join Us

Recent News

[Nov 8, 2017]
Prof. C.C. Jay Kuo (USC/MCL, IEEE Fellow, AAAS Fellow, SPIE Fellow) was invited to SIAT and give us a talk, titled Prediction of Satisfied User Ratio (SUR) Curves and Just-Noticeable-Difference (JND) Points for Coded Video, at 10:30-11:30am,room 1119,building F. And at afternoon he was invited to Shenzhen University and give us a talk, titled Why Deep Learning Works So Well? What is Next?. welcome to attend.

[Sep 20, 2017]
Co-authored paper, titled Fuzzy SVM Based Fast CU Decision for HEVC Encoding with Emphasis on Misclassification Cost Intra Coding, has been accepted by IEEE Transactions on Broadcasting.

[Aug. 24, 2017]
Our paper titled Effective Data Driven CU Size Decision Approaches for HEVC Intra Coding has been accepted by IEEE Transactions on Circuits System and Video Technology.

[July 28, 2017]
Our Guangdong Provincial Science and Technology Major Project has been granted, 3 million CNY. http://www.gdczt.gov.cn/zwgk/ggtz/201707/t20170727_856523.htm.

[July 25, 2017]
Co-authored paper, titled Multi-Class Ranking Based Most Probable Prediction Unit Selection for HEVC Encoding, has been accepted by VCIP 2017, to be held in Florida, USA.

[July 24, 2017]
Co-authored paper, titled Compression Artifacts Reduction for Depth Map by Deep Intensity Guidance, has been accepted by PCM 2017, to be held in Harbin, China, September 28-29, 2017

[July 14, 2017]
Prof. C.C. Jay Kuo (USC/MCL, IEEE Fellow, AAAS Fellow, SPIE Fellow) are invited to SIAT and give us a talk, titled A Data Driven Approach to Perceptual Video Coding. (pdf)

[July 7, 2017]
Workshop on Immersive Visual Signal Processing is held in SIAT. Prof. Yo-Sung Ho (GIST, IEEE Fellow), Prof. Long Xu (NAO,CAS), Dr. You Yang (HUST) and Dr. Xu Wang were invited to give us talks. (pdf)

[May 28, 2017]
Our International Joint Research Project between SIAT and USC/MCL has been granted by Shenzhen Science and Technology Innovation Commission. http://www.szsti.gov.cn/notices/2017/5/25/3

[May 19, 2017]
Our paper titled "Instant Coherent Group Motion Filtering by Group Motion Representations " has been accepted for publication by Neurocomputing.

[May 12, 2017]
Our paper titled "Binary and Multi-Class Learning Based Low Complexity Optimization for HEVC Encoding" has been accepted for publication by IEEE Transactions on Broadcasting.

Research Areas

Our team are mainly engaged in the field of Multimedia Communications and Visual Signal Processing and develop practical systems for image, 2D/3D and VR video applications. The research topics are:

Immersive UHD Video Coding and Communication
High Efficiency Video Coding and Transmission
Low Complexity Video Coding
Image and Video Super-Resolution

3D/Multiview Video Coding and Processing
Multiview Depth Video Coding
Depth Image Based Rendering
Texture-Depth Joint Video Processing

Computational Visual Perception
3D Video Quality/QoE
Image/Video Quality Assessment
Visual Attention
Vision Based Pattern Recognition
Crowd Behavior Analysis
Crowd Detection

Learn More

Research Achievements

We have published more than 80 high quality SCI/EI journal and conference papers, more than 20 of them are published on IEEE Transactions, including IEEE Trans. Broadcast. IEEE Trans. Image Process. IEEE Trans. Circuits Syst. Video Technol., IEEE Trans. Industrial Electronics, IEEE Trans. Industrial Informatics. Also, we have filed 15 CN and US patents, and 10 of them are granted. More details can be referred in our Publication page.


In addition, we have been rewarded by 2009 Second Prize Scientific and Technological Advancement Award from Ministry of Education, China, First Prize Zhejiang Provincial Science and Technology Award, First Prize Ningbo Municipal Science and Technology Award, 2015 First Prize Ningbo Municipal Science and Technology Award, 2016 Third Prize Zhejiang Provincial Science and Technology Award, and so on.

Read more

Download Center

In our download center, you can download some data we have shared.

Some academic papers we have published.
A Synthesized Video Quality Database for academic research.
Some interesting video we captured or made.
Some software we developed.

Note: All downloads can't be used for commercial purposes, if you have any questions, please contact us: (yun.zhang@siat.ac.cn)

Download Center

VOD System

2D/3D VOD System

SIAT3DPlayer is a versatile player with the following three major functions: (1) Video recording. It supports the single-camera recording and dual camera recording, dual camera can record 3D video (left/right, up/down, etc.) or record seamless panoramic video. 4 kinds of video encryption methods are supported while recoding or transmission; (2) Video playback. The SIAT3DPlayer can play 2D/3D videos with various coding formats and 3D data formats. , and can automatically identify part of the 3D video format (left/right, up/down) and automatically set 3D playback effect. (3) Video-on-demand and live 3D broadcasting, SIAT3DPlayer is a video-on-demand client that supports the http protocol. It can play the videos on our website: http://codec.siat.ac.cn/vod/index.html and third party website. In addition, it supports live 3D broadcasting on RTP protocol.

Live VR Video System

SIATVRPlayer is a new developed video solution for VR video system, which is mainly comprised of 360° panorama picture producer, high efficiency video coder and VR video player. The 360° panorama picture is collaged from pictures of different view angles captured by a number of GoPro cameras. The VR video player is capable of displaying both the live 360°panorama video through RTMP and the available VR video online through URL. It provides three kinds of play mode: 1) VOD on HDTV. Pictures from monocular viewpoint is projected and displayed on 2D screen, one view or double view. The viewpoint can be changed arbitrary by moving the mouse and other input devices. 2) VOD on Head Mounted Device (HMD). Pictures from binocular viewpoint are generated and transferred to the HMD, so as to provide immersive experiences. The viewpoint can be seamless changed by turning head around without delay. 3) Live VR broadcasting.