notesum.ai

Published at November 20

XMask3D: Cross-modal Mask Reasoning for Open Vocabulary 3D Semantic Segmentation

cs.CV

cs.AI

Released Date: November 20, 2024

Authors: Ziyi Wang¹, Yanbo Wang¹, Xumin Yu¹, Jie Zhou¹, Jiwen Lu¹

Aff.: ¹Department of Automation, Tsinghua University, China

Arxiv: http://arxiv.org/abs/2411.13243v1

Refer to caption

Method	Scannet									ScanNet200
	B15/N4			B12/N7			B10/N9			B170/N30			B150/N50
	hIoU	Base	Novel	hIoU	Base	Novel	hIoU	Base	Novel	hIoU	Base	Novel	hIoU	Base	Novel
LSeg-3D [23]	0.0	64.4	0.0	0.9	55.7	0.1	1.8	68.4	0.9	1.5	21.1	0.8	3.0	20.6	1.6
3DGenZ [31]	20.6	56.0	12.6	19.8	35.5	13.3	12.0	63.6	6.6	2.6	15.8	1.4	3.3	14.1	1.9
3DTZSL [5]	10.5	36.7	6.1	3.8	36.6	2.0	7.8	55.5	4.2	0.9	4.0	0.5	0.7	3.8	0.4
PLA [11]	65.3	68.3	62.4	55.3	69.5	45.9	53.1	76.2	40.8	11.4	20.9	7.8	10.1	20.9	6.6
OpenScene [34]	65.7	68.8	62.8	56.8	61.5	51.7	54.3	71.8	43.6	14.2	22.5	10.4	15.2	23.5	11.2
OV3D [20]	72.4	70.2	74.7	68.5	74.1	63.7	64.8	77.6	55.6	–	–	–	–	–	–
XMask3D	70.0	69.8	70.2	61.7	70.2	55.1	55.7	76.5	43.8	18.0	27.8	13.3	15.5	24.4	11.4