The Future Value Creation Research Center, GSI-NU will hold the visiting scientist’s lecture on August 2.

Following Lecture  will be held on Aug. 2, Friday.
         

[Date/Time]     Aug.  2, 2019 (Fri), 14:45 - 16:15

[Location]      Lecture Room 143, Engineering Building 1,
               　　  Department of Engineering, Nagoya University
 

[Speaker]       Prof. Chia-Wen LIN

                          Designated Professor, 　Graduate School of Informatics, Nagoya University
                          Professor,
                          Department of Electrical Engineering,
                          National Tsing Hua University, Taiwan)
 

[Title]     Identity-preserving face processing for better face recognition
 

[Abstract]    
  Face detection/recognition is essential in video  surveillance applications. However, the performance of  current face detection and recognition schemes can be easily significantly degraded for difficult cases in  real-world video surveillance applications such as  detecting and recognizing low-resolution faces in the  wild, and recognizing faces with illumination variations, viewpoint changes, and facial expression  variations. In this lecture, I will show some methods and their results in addressing the difficult problems  in face recognition for real-world video surveillance  applications, including identity-preserving face hallucination, face augmentation and normalization. Our approaches are based on Siamese Generative Adversarial Networks (SiGANs) to achieve identity-preserving face hallucination and normalization. We incorporate the reconstruction error and identity label information in the loss function of SiGAN. By  iteratively optimizing the loss functions of the generator and discriminator of SiGAN, we cannot only  maximize the visual fidelity between the reconstructed faces and their ground-truths, but also ensure the                reconstructed information is useful for identity recognition. Experimental results on large-scale  public face datasets demonstrate the efficacy of the
proposed approaches.
 

[Bio]          
Prof. Chia-Wen Lin received his PhD degree in Electrical Engineering from National Tsing Hua University (NTHU), Hsinchu, Taiwan in 2000.
He is  currently a Professor with the Department of  Electrical Engineering, NTHU, Taiwan. He also serves  as Deputy Director of the AI Research Center of NTHU.   His research interests include image/video processing, computer vision, and video networking.
Prof. Lin is an IEEE Fellow. He is a Distinguished  Lecturer of IEEE Circuits and Systems Society (CASS)  during 2018-2019. He has served as Associate Editor of IEEE T-IP, IEEE T-MM, IEEE T-CSVT, and IEEE Multimedia.  He is currently President of Chinese Image Processing and Pattern Recognition Association, Taiwan. He has  served as TPC Chair of IEEE ICME in 2010 and ICIP 2019,  the Conference Chair of IEEE VCIP 2018.
 

[Contact]      
Ichiro IDE
(Graduate School of, Informatics, Nagoya University)
TEL: 052-789-3313 / E-mail: ide@i.nagoya-u.ac.jp