Heedless Backbones

GlobalMamba 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

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
GlobalMamba-T	30.0	IN-1k : Sup. : 300	— : — : —	5.3	82.8
GlobalMamba-S	50.0	IN-1k : Sup. : 300	— : — : —	9.5	83.9
GlobalMamba-B	89.0	IN-1k : Sup. : 300	— : — : —	17.0	84.1

model	params (m)	pretrain	finetune	gflops	IN-1k
GlobalMamba-T	30.0	IN-1k : Sup. : 300	— : — : —	5.3	82.8/—
GlobalMamba-S	50.0	IN-1k : Sup. : 300	— : — : —	9.5	83.9/—
GlobalMamba-B	89.0	IN-1k : Sup. : 300	— : — : —	17.0	84.1/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
GlobalMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	None	47.6	69.4	52.2	—	—	—
GlobalMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	None	49.0	70.5	53.7	—	—	—
GlobalMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	None	49.0	70.5	53.5	—	—	—
GlobalMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	None	50.1	—	54.9	—	—	—
GlobalMamba-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	None	49.3	71.4	54.2	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
GlobalMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	None	42.9	66.5	46.0	—	—	—
GlobalMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	None	43.8	67.5	47.1	—	—	—
GlobalMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	None	43.9	67.5	47.0	—	—	—
GlobalMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	None	44.3	68.4	47.8	—	—	—
GlobalMamba-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	None	44.2	68.4	47.7	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
GlobalMamba-T	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	None	49.0	—	—	48.1	—	—
GlobalMamba-S	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	None	51.4	—	—	50.9	—	—
GlobalMamba-B	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	None	51.7	—	—	51.2	—	—