notesum.ai

Published at November 6

WiFlexFormer: Efficient WiFi-Based Person-Centric Sensing

cs.CV

cs.AI

cs.ET

cs.LG

Released Date: November 6, 2024

Authors: Julian Strohmayer¹, Matthias Wödlinger, Martin Kampel¹

Aff.: ¹Computer Vision Lab, TU Wien

Arxiv: http://arxiv.org/abs/2411.04224v1

Refer to caption

	Inference Time [ms]
Model	Amplitude	DFS xx.
RF-Net [2]	-	427.20 $\pm$ 13.6
SLNet [25]	-	322.31 $\pm$ 7.25
EfficientNetV2s [21]	67.72 $\pm$ 0.64	68.66 $\pm$ 0.78
THAT [11]	37.87 $\pm$ 0.49	-
ShuffleNetV2x0.5 [13]	22.16 $\pm$ 0.55	22.59 $\pm$ 0.50
ResNet18 [6]	9.67 $\pm$ 0.19	12.01 $\pm$ 0.42
WiFlexFormer [ours]	9.26 $\pm$ 0.39	11.06 $\pm$ 0.63