Depth Estimation
Depth estimation is a task in computer vision that aims to measure the distance of each pixel relative to the camera. This task extracts depth information from monocular or stereo images, with traditional methods based on multi-view geometry, while newer methods directly estimate depth by minimizing regression loss or learning to generate new views from sequences. Depth estimation has significant application value in areas such as autonomous driving, robot navigation, and augmented reality, with common evaluation metrics including Root Mean Square Error (RMS), and major benchmark datasets including KITTI and NYUv2.
Stanford2D3D Panoramic
NYU-Depth V2
EVP
DCM
eBDtheque
Bhattacharjee et al.
ScanNetV2
Distill Any Depth
Cityscapes test
SwinMTL
DIODE
AIP-Brown
KITTI 2015
Mars DTM Estimation
GLPDepth
ScanNet
Atlas (plain)
4D Light Field Dataset
LFattNet
KITTI Eigen split
LightDepth
Matterport3D
UniFuse
Taskonomy
X-TC (Cross-Task Consistency)