International Display Workshops General Incorporated Association

13:20 〜 13:40

[VHF1/DES1-2] GAN Based Image-to-Image Translation Model for Nighttime Road Scene Dataset

*Rebeka Sultana1、Yuki Hikosaka1、Gosuke Ohashi1 (1.Shizuoka University (Japan))

Image-to-image translation, GAN model, daytime road scene, nighttime road scene, ADAS

https://doi.org/10.36463/idw.2022.0637

A large-scale dataset boosts the performance of deep learning models. However, in-vehicle camera image datasets contain many daytime and few nighttime scenes. Therefore, this study proposes an image-to-image translation model to convert daytime road scenes to nighttime road scenes by preserving visual appearance and contents to increase nighttime dataset.