From Perspective to Fisheye Depth Estimation and Open-Vocabulary Segmentation
Paper • 2608.27860 • Published
The 400 fisheye images and distance maps used to evaluate KITTI-360 in From Perspective to Fisheye Depth Estimation and Open-Vocabulary Segmentation (code: https://github.com/Suchisrit/DEX).
KITTI-360/RGB/*.png: 1400 x 1400 fisheye images (sequence 2013_05_28_drive_0010_sync, cameras image_02 and image_03, the _90 variants listed in splits/kitti360_test.txt of the code repository)KITTI-360/Depth/*.npy: 1400 x 1400 float32 Euclidean distance maps in meters, 0 where there is no measurementDownload into the data/ folder of the code repository so that the evaluation script finds it:
huggingface-cli download Rit888/DEX_KITTI360_Eval --repo-type dataset --local-dir data
Derived from KITTI-360:
@article{Liao2022PAMI,
title = {{KITTI}-360: A Novel Dataset and Benchmarks for Urban Scene Understanding in 2D and 3D},
author = {Yiyi Liao and Jun Xie and Andreas Geiger},
journal = {Pattern Analysis and Machine Intelligence (PAMI)},
year = {2022}
}