notesum.ai

Published at December 9

cs.RO

cs.CV

Released Date: December 9, 2024

Authors: Jesse Hagenaars, Yilun Wu¹, Federico Paredes-Vallés, Stein Stroobants¹, Guido de Croon

Aff.: ¹TU Delft

Depth Cutoff	10m	20m	30m	10m	20m	30m
	outdoor_day1			outdoor_night1
Zhu et al. [39] ³³3The network uses events as input but was trained with intensity frames in the loss function. Therefore, it is included as a reference but is not directly comparable to self-supervised methods trained solely on event data.	1.40	2.07	2.65	2.18	2.70	3.64
Zhu et al. [39]	3.90	3.79	4.89	5.55	4.57	5.72
Zhu et al. [38]	2.72	3.84	4.40	3.13	4.02	4.89
Ours	2.25	3.36	4.23	3.25	3.83	4.50
\hdashlineOurs (Dense)	1.96	2.67	3.29	2.92	3.56	4.28