Heedless Backbones

MobileMamba 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

---------

COCO (val)

COCO (test)

PASCAL VOC 2007 (val)

PASCAL VOC 2007 (test)

metric

Select an option

----------

mAP

AP₅₀

AP₇₅

mAP_s

mAP_m

mAP_l

GFLOPs

head

Select an option

---------

Mask R-CNN

Cascade Mask R-CNN

RetinaNet

DINO

HTC++

HTC

FCOS

Faster R-CNN

DETR

SSDLite

Sparse R-CNN

ATSS

axis

Select an option

Results

Parameters (M)

Images / Second

GFLOPs

Publication Date

Filters

Legend

Legend attribute

Select an option

Family

Pretrain Dataset

Instance Head

Instance Training Epochs

Legend attribute (second)

Select an option

----------

Family

Pretrain Method

Instance Head

Instance Training Epochs

model	params (m)	pretrain	head	train	GFLOPs	mAP
MobileMamba-B1	17.1	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	178.0	40.6
MobileMamba-B1	17.1	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	151.0	39.6
MobileMamba-B1	17.1	IN-1k : Sup. : 300	SSDLite	COCO (train) : 600	1.7	24.0

model	params (m)	pretrain	finetune	gflops	IN-1k
MobileMamba-T2	8.8	IN-1k : Sup. : 300	— : — : —	0.255	73.6/91.422
MobileMamba-T2	8.8	IN-1k : Sup. : 1000	— : — : —	0.255	76.9/93.1
MobileMamba-T4	14.2	IN-1k : Sup. : 300	— : — : —	0.413	76.1/92.772
MobileMamba-T4	14.2	IN-1k : Sup. : 1000	— : — : —	0.413	78.9/94.16
MobileMamba-S6	15.0	IN-1k : Sup. : 300	— : — : —	0.652	78.0/93.992
MobileMamba-S6	15.0	IN-1k : Sup. : 1000	— : — : —	0.652	80.7/95.182
MobileMamba-B1	17.1	IN-1k : Sup. : 300	— : — : —	1.08	79.9/94.924
MobileMamba-B1	17.1	IN-1k : Sup. : 1000	— : — : —	1.08	82.2/95.872
MobileMamba-B2	17.1	IN-1k : Sup. : 300	— : — : —	2.427	81.6/95.89
MobileMamba-B2	17.1	IN-1k : Sup. : 1000	— : — : —	2.427	83.3/96.438
MobileMamba-B4	17.1	IN-1k : Sup. : 300	— : — : —	4.313	82.5/96.252
MobileMamba-B4	17.1	IN-1k : Sup. : 1000	— : — : —	4.313	83.6/96.606

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
MobileMamba-B1	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	178.0	40.6	61.8	43.8	22.4	43.5	55.9
MobileMamba-B1	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	151.0	39.6	59.8	42.4	21.5	43.1	53.9
MobileMamba-B1	IN-1k : Sup. : 300	SSDLite	COCO (train) : 600	1.7	24.0	39.5	24.0	3.1	23.4	46.9

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
MobileMamba-B1	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	178.0	37.4	58.9	39.9	17.1	39.9	56.4

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
MobileMamba-B4	IN-1k : Sup. : 300	DeepLabv3	ADE20K (train) : 128 : 512	4.7	—	—	—	36.6	76.3	47.1
MobileMamba-B4	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 256 : 512	5.6	—	—	—	42.5	79.9	53.7
MobileMamba-B4	IN-1k : Sup. : 300	PSPNet	ADE20K (train) : 128 : 512	4.5	—	—	—	36.9	76.2	47.9