Depth Estimation
Depth estimation, the process of determining the distance of objects from a camera, aims to reconstruct 3D scenes from visual data, crucial for applications like autonomous driving and robotics. Current research emphasizes improving accuracy and robustness, particularly in challenging scenarios like endoscopy and low-light conditions, often employing self-supervised learning techniques and novel neural network architectures such as transformers and diffusion models alongside traditional stereo vision methods. These advancements are driving progress in various fields, including medical imaging, autonomous navigation, and 3D scene reconstruction, by enabling more accurate and reliable perception of the environment.
Papers
DistillNeRF: Perceiving 3D Scenes from Single-Glance Images by Distilling Neural Fields and Foundation Model Features
Letian Wang, Seung Wook Kim, Jiawei Yang, Cunjun Yu, Boris Ivanovic, Steven L. Waslander, Yue Wang, Sanja Fidler, Marco Pavone, Peter Karkus
MEDeA: Multi-view Efficient Depth Adjustment
Mikhail Artemyev, Anna Vorontsova, Anna Sokolova, Alexander Limonov