Yu-Gang Jiang

School of Computer Science
Fudan University

ygj AT fudan.edu.cn

My research is in the areas of multimedia content analysis and computer vision. I lead the BigVid Lab, conducting research on all aspects of extracting high-level information from big video data, such as video event recognition, object/scene recognition and large-scale visual search.

I participate regularly in international benchmark competitions. At the annual U.S. NIST TREC video retrieval evaluation (TRECVID), I have designed a few best-performing systems (among many submissions worldwide) in 2008 video concept detection task and 2010 multimedia event detection task. Recent techniques developed in our lab have produced top video content recognition accuracies in various competitions like the YouTube-8M Video Understanding Challenge.

My publications and citations on Google Scholar can be found here.

[07/2019] Our work on class-incremental object detection received Best Paper Award at IEEE ICME 2019.
[11/2018] Five papers accepted to AAAI 2019.
[07/2018] Four papers accepted to ECCV 2018.
[02/2017] Our work on video content recognition with regularized DNN has been accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence.
[06/2016] Our work on image search with hash codes received the 2016 IEEE Trans. on Multimedia Prize Paper Award Honorable Mention.
[02/2016] Due to many requests, researchers interested in getting the videos of the CCV dataset can fill up this form and send to me for a link.
[02/2015] We have released FCVID, one of the largest public Web video datasets with manual annotations (91,223 videos, 239 categories).

dataset / code

FCVID: Fudan-Columbia Video Dataset
91,223 Web videos annotated manually according to 239 categories
VCDB: a large-scale database for partial copy detection in videos
ECCV 2014 paper
Dataset for predicting video emotions
AAAI 2014 paper
Dataset for predicting video interestingness
AAAI 2013 paper
Part-level attributes for visual recognition (source code)
ECCV 2012 paper
CCV: a benchmark dataset for consumer video analysis
Note: To obtain a copy of the videos in CCV, please fill up this form and send to me for a link.
Domain adaptive semantic diffusion (DASD) for context-based visual annotation refinement
ICCV 2009 paper | source code
VIREO-374: keypoint-based LSCOM semantic concept detectors see who has used it
CU-VIREO374: fusing Columbia374 and VIREO374 for large scale semantic concept detection see who has used it

Selected other recent projects

Query-adaptive image search with hash codes
IEEE TMM 2013 paper
Violent scene detection in movies

Mailing Address: Room 413-5, Computer Science Building, 825 Zhangheng Road, Pudong, Shanghai 201203, China.   Tel: +86-21-51355532

All the materials shared on this personal website are to ensure timely dissemination of scholarly and technical works. Copyright and all rights therein are retained by Yu-Gang Jiang or by other copyright holders. People who copy these materials are expected to adhere to the corresponding terms and constraints. These works may not be reposted without the explicit permission of the copyright holders.