TY - GEN
T1 - Super-Identity Convolutional Neural Network for Face Hallucination
AU - Zhang, Kaipeng
AU - Zhang, Zhanpeng
AU - Cheng, Chia Wen
AU - Hsu, Winston H.
AU - Qiao, Yu
AU - Liu, Wei
AU - Zhang, Tong
N1 - Acknowledgement. This work was supported in part by MediaTek Inc and the Ministry of Science and Technology, Taiwan, under Grant MOST 107-2634-F-002 -007. We also benefit from the grants from NVIDIA and the NVIDIA DGX-1 AI Supercomputer.
PY - 2018
Y1 - 2018
N2 - Face hallucination is a generative task to super-resolve the facial image with low resolution while human perception of face heavily relies on identity information. However, previous face hallucination approaches largely ignore facial identity recovery. This paper proposes Super-Identity Convolutional Neural Network (SICNN) to recover identity information for generating faces closed to the real identity. Specifically, we define a super-identity loss to measure the identity difference between a hallucinated face and its corresponding high-resolution face within the hypersphere identity metric space. However, directly using this loss will lead to a Dynamic Domain Divergence problem, which is caused by the large margin between the high-resolution domain and the hallucination domain. To overcome this challenge, we present a domain-integrated training approach by constructing a robust identity metric for faces from these two domains. Extensive experimental evaluations demonstrate that the proposed SICNN achieves superior visual quality over the state-of-the-art methods on a challenging task to super-resolve 12 × 14 faces with an 8 × upscaling factor. In addition, SICNN significantly improves the recognizability of ultra-low-resolution faces.
AB - Face hallucination is a generative task to super-resolve the facial image with low resolution while human perception of face heavily relies on identity information. However, previous face hallucination approaches largely ignore facial identity recovery. This paper proposes Super-Identity Convolutional Neural Network (SICNN) to recover identity information for generating faces closed to the real identity. Specifically, we define a super-identity loss to measure the identity difference between a hallucinated face and its corresponding high-resolution face within the hypersphere identity metric space. However, directly using this loss will lead to a Dynamic Domain Divergence problem, which is caused by the large margin between the high-resolution domain and the hallucination domain. To overcome this challenge, we present a domain-integrated training approach by constructing a robust identity metric for faces from these two domains. Extensive experimental evaluations demonstrate that the proposed SICNN achieves superior visual quality over the state-of-the-art methods on a challenging task to super-resolve 12 × 14 faces with an 8 × upscaling factor. In addition, SICNN significantly improves the recognizability of ultra-low-resolution faces.
KW - Convolutional neural networks
KW - Domain-integrated training
KW - Face hallucination
KW - Super identity
UR - https://www.scopus.com/pages/publications/85055085791
UR - https://www.scopus.com/inward/citedby.url?scp=85055085791&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-01252-6_12
DO - 10.1007/978-3-030-01252-6_12
M3 - Conference contribution
AN - SCOPUS:85055085791
SN - 9783030012519
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 196
EP - 211
BT - Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
A2 - Ferrari, Vittorio
A2 - Sminchisescu, Cristian
A2 - Weiss, Yair
A2 - Hebert, Martial
PB - Springer
T2 - 15th European Conference on Computer Vision, ECCV 2018
Y2 - 8 September 2018 through 14 September 2018
ER -