Introduction
This is a Tensorflow implementation of Deep Ordinal Regression Network for Monocular Depth Estimation.
See utils/confutil.py for details.
KITTI
*Dataset path
input:
~/_DepthPredic/DATASET/KITTI_eigen_dense/[%A]/[%B]/[%C]/[%D]/data/[filename].jpg
gt:~/_DepthPredic/DATASET/KITTI_eigen_dense/[%A]/[%B]/[%C]/[%D]/groundtruth/[filename].png
%A = { train-set, test-set, valid-set }
%B = { 2011_09_26, ... }
%C = { 2011_09_26_drive_0002_sync, ... }
%D = { image_02, image_03 }
*CSV path
~/_DepthPredic/DATASET/KITTI_eigen_dense/[%A].csv
%A = { train, test, valid }
*CSV Format
[1] ../../DATASET/KITTI_eigen_dense/train-set/2011_09_26/2011_09_26_drive_0001_sync/image_02/data/0000000049.jpg, ../../DATASET/KITTI_eigen_dense/train-set/2011_09_26/2011_09_26_drive_0001_sync/image_02/groundtruth/0000000049.png
[2] ../../DATASET/KITTI_eigen_dense/train-set/2011_09_26/2011_09_26_drive_0001_sync/image_02/data/0000000057.jpg, ../../DATASET/KITTI_eigen_dense/train-set/2011_09_26/2011_09_26_drive_0001_sync/image_02/groundtruth/0000000057.png
[3]...
NYU
*Dataset path
input:
~/_DepthPredic/DATASET/NYU/[%A]/[%B]/[filename].jpg
gt:~/_DepthPredic/DATASET/NYU/[%A]/[%B]/[filename].png
%A = { train-set, test-set, valid-set }
%B = { basement_0001a_out, ... }
*CSV path
~/_DepthPredic/DATASET/NYU/[%A].csv
%A = { train, test, valid }