International Display Workshops General Incorporated Association

9:20 AM - 9:40 AM

[VHF4/AIS2-2] Estimation of Helmholtz-Kohlrausch Effect using Deep Learning

*Shuhei Yamamoto1, Kakeru Amano1, Hisakazu Aoyanagi2, Hiroaki Takamatsu2, Yoshifumi Shimodaira1, Gosuke Ohashi1 (1. Shizuoka University(Japan), 2. NEC Display Solutions, Ltd.(Japan))

Helmholtz-Kohlrausch effect, deep learning, natural image, data augmentation, fine-tuning

https://doi.org/10.36463/idw.2020.0580

The purpose of this study is to estimate Helmholtz-Kohlrausch effect by using deep neural networks. Data augmentation and fine-tuning based on empirical knowledge of subjective evaluation experiments enable deep neural networks to estimate Helmholtz-Kohlrausch effect.