Heedless Backbones

MSVMamba Family

Paper | GitHub

axis

Select an option

Results

Parameters (M)

Images / Second

Publication Date

task

Select an option

---------

Object Detection

Instance Segmentation

Classification

Semantic Segmentation

Panoptic Segmentation

dataset

Select an option

---------

ImageNet-1k

ImageNet-A

ImageNet-R

ImageNet-Sketch

ImageNet-C

ImageNet-C-bar

ImageNet-V2

ImageNet-ReaL

PASCAL VOC 2007 (val)

PASCAL VOC 2007 (test)

metric

Select an option

----------

Top-1

Top-5

GFLOPs

resolution

Select an option

----------

224x224

384x384

512x512

axis

Select an option

Results

Parameters (M)

Images / Second

GFLOPs

Publication Date

Filters

pretrain dataset

Select an option

---------

ImageNet-1k

ImageNet-22k

JFT-300M

JFT-3B

MegData73M

pretrain method

Select an option

----------

Supervised

Sup. + TL

FCMAE

MAE

CL

MAP

Legend

Legend attribute

Select an option

Family

Pretrain Dataset

Classification Resolution

Legend attribute (second)

Select an option

----------

Family

Pretrain Method

Classification Resolution

model	params (m)	pretrain	finetune	GFLOPs	Top-1
MSVMamba-N	6.9	IN-1k : Sup. : 300	— : — : —	0.9	77.3
MSVMamba-M	11.9	IN-1k : Sup. : 300	— : — : —	1.5	79.8
MSVMamba-T	33.0	IN-1k : Sup. : 300	— : — : —	4.6	82.8

model	params (m)	pretrain	finetune	gflops	IN-1k
MSVMamba-N	6.9	IN-1k : Sup. : 300	— : — : —	0.9	77.3/—
MSVMamba-M	11.9	IN-1k : Sup. : 300	— : — : —	1.5	79.8/—
MSVMamba-T	33.0	IN-1k : Sup. : 300	— : — : —	4.6	82.8/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
MSVMamba-M	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	201.0	43.8	65.8	47.7	—	—	—
MSVMamba-M	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	201.0	46.3	68.1	50.8	—	—	—
MSVMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	252.0	46.9	68.8	51.4	—	—	—
MSVMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	252.0	48.3	69.5	53.0	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
MSVMamba-M	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	201.0	39.9	62.9	42.9	—	—	—
MSVMamba-M	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	201.0	41.8	65.1	44.9	—	—	—
MSVMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	252.0	42.2	65.6	45.4	—	—	—
MSVMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	252.0	43.2	66.8	46.9	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
MSVMamba-M	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	875.0	45.4	—	—	45.1	—	—
MSVMamba-T	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	942.0	48.5	—	—	47.6	—	—