notesum.ai

Published at November 25

cs.CV

Released Date: November 25, 2024

Authors: Katharina Bendig¹, René Schuster, Nicole Thiemer², Karen Joisten², Didier Stricker¹

Aff.: ¹DFKI - German Research Center for Artificial Intelligence; ²RPTU - University of Kaiserslautern-Landau

Dataset	Input	Stage	$acc_{T}[\%]\uparrow$	$acc_{id}[\%]\downarrow$	$mAP_{id}[\%]\downarrow$
DVS-Gesture [4]	Raw	Pre-training	92.42	98.33	69.64
	Noise std=32	Post-training	84.47	60.00	27.24
	Noise std=64	Post-training	82.20	61.67	27.04
	Noise std=128	Post-training	78.79	50.00	23.40
	Noise std=256	Post-training	66.29	36.67	20.07
	Anonymized	All stages	80.68	40.00	20.69
SEE [29]	Raw	Pre-training	77.14	72.5	43.27
	Noise std = 1	Post-training	66.43	47.5	25.63
	Noise std = 2	Post-training	60.00	35.00	17.98
	Noise std = 4	Post-training	55.00	32.5	17.08
	Anonymized	All stages	55.71	25.0	13.14