Benchmark dataset for depth estimation from lightfield images

Citation Author(s):
Mingtao
Feng
Yaonan
Wang
Jian
Liu
Liang
Zhang
Hasan
Zaki
Ajmal
Mian
Submitted by:
Ajmal Mian
Last updated:
Mon, 08/01/2022 - 03:25
DOI:
10.21227/yk0j-e526
Data Format:
Link to Paper:
Links:
License:
0
0 ratings - Please login to submit your rating.

Abstract 

We propose a real world data set comprising light field images of 19 objects captured with the Lytro Illum camera in outdoor scenes and their corresponding 3D point clouds, as ground truth, captured with the 3dMD scanner. This data set allows more precise 3D pointcloud level comparison of algorithms for the task of depth estimation or 3D point cloud reconstruction from light field images.

Instructions: 

There are 19 folders corresponding to the objects imaged with the light field camera multiple times. This data is in .mat format i.e. Matlab data files. Each folder also contains a .wrl file i.e. a point cloud (and mesh) obtained with the 3dMD scanner which uses two view points to capture 180 degrees.

A camera parameters file is also provided for the Lytro Illum camera.