Disparity Estimation Using a Dual/Quad-Pixel Sensor

Okutomi-Monno Lab.
Institute of Science Tokyo


Introduction of Dual/Quad-Pixel Disparity Estimation

Dual-pixel (DP) and Quad-pixel (QP) sensors are increasingly integrated into modern cameras. Although originally designed for autofocus through phase-shift information, these sensors provide unique data that can be exploited for 3D scene understanding. Our research aims to enhance 3D geometry reconstruction by effectively leveraging data from both dual-pixel and quad-pixel sensors.

FMDP: Leveraging a Foundation Model for Dual-Pixel Disparity Estimation
[Project Page]

FMDP
Proposed architecture for dual-pixel disparity estimation leveraging a foundation model.

Publication:
FMDP: Leveraging a Foundation Model for Dual-Pixel Disparity Estimation
Doehyung Lee, Zhuofeng Wu, Yusuke Monno, and Masatoshi Okutomi
International Conference on Machine Vision Applications (MVA), 2025.

Disparity Estimation Using a Quad-pixel Sensor [Project Page]

QPDNet
Proposed architecture for quad-pixel disparity estimation.

Publication:
Disparity Estimation Using a Quad-pixel Sensor
Zhuofeng Wu, Doehyung Lee, Zihua Liu, Kazunori Yoshizaki, Yusuke Monno, and Masatoshi Okutomi
British Machine Vision Conference (BMVC), 2024.