TY - GEN
T1 - Coloring with limited data
T2 - 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019
AU - Yoo, Seungjoo
AU - Bahng, Hyojin
AU - Chung, Sunghyo
AU - Lee, Junsoo
AU - Chang, Jaehyuk
AU - Choo, Jaegul
N1 - Funding Information:
This work was partially supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIP) (No. NRF2016R1C1B2015924).
Funding Information:
Acknowledgements. This work was partially supported bytheNationalResearchFoundationofKorea (NRF) grantfundedbytheKorean government(MSIP)(No. NRF2016R1C1B2015924). We thankall researchers at NAVER WEBTOONCorp.,especiallySungminKang. Jaegul Choo is the corresponding author.
Publisher Copyright:
© 2019 IEEE.
PY - 2019/6
Y1 - 2019/6
N2 - Despite recent advancements in deep learning-based automatic colorization, they are still limited when it comes to few-shot learning. Existing models require a significant amount of training data. To tackle this issue, we present a novel memory-augmented colorization model MemoPainter that can produce high-quality colorization with limited data. In particular, our model is able to capture rare instances and successfully colorize them. Also, we propose a novel threshold triplet loss that enables unsupervised training of memory networks without the need for class labels. Experiments show that our model has superior quality in both few-shot and one-shot colorization tasks.
AB - Despite recent advancements in deep learning-based automatic colorization, they are still limited when it comes to few-shot learning. Existing models require a significant amount of training data. To tackle this issue, we present a novel memory-augmented colorization model MemoPainter that can produce high-quality colorization with limited data. In particular, our model is able to capture rare instances and successfully colorize them. Also, we propose a novel threshold triplet loss that enables unsupervised training of memory networks without the need for class labels. Experiments show that our model has superior quality in both few-shot and one-shot colorization tasks.
KW - Computational Photography
KW - Computer Vision Theory
KW - Deep Learning
KW - Image and Video Synthesis
KW - Vi
KW - Vision + Graphics
UR - http://www.scopus.com/inward/record.url?scp=85078777565&partnerID=8YFLogxK
U2 - 10.1109/CVPR.2019.01154
DO - 10.1109/CVPR.2019.01154
M3 - Conference contribution
AN - SCOPUS:85078777565
T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
SP - 11275
EP - 11284
BT - Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019
PB - IEEE Computer Society
Y2 - 16 June 2019 through 20 June 2019
ER -