Heedless Backbones

DAMamba 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)

Cityscapes (val)

Cityscapes (test)

ADE20K (val)

ADE20K (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

Mask2Former

HTC++

HTC

Panoptic FPN

axis

Select an option

Results

Parameters (M)

Images / Second

GFLOPs

Publication Date

Filters

pretrain dataset

Select an option

---------

MegData73M

JFT-3B

JFT-300M

ImageNet-1k

ImageNet-22k

pretrain method

Select an option

----------

Supervised

Sup. + TL

FCMAE

MAE

CL

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
DAMamba-T	26.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	284.0	43.4
DAMamba-T	26.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	284.0	44.8
DAMamba-S	45.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	395.0	44.5
DAMamba-S	45.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	395.0	45.1
DAMamba-B	86.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	520.0	44.9
DAMamba-B	86.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	520.0	45.3

model	params (m)	pretrain	finetune	gflops	IN-1k
DAMamba-T	26.0	IN-1k : Sup. : 300	— : — : —	4.8	83.8/—
DAMamba-S	45.0	IN-1k : Sup. : 300	— : — : —	10.3	84.8/—
DAMamba-B	86.0	IN-1k : Sup. : 300	— : — : —	16.3	85.2/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
DAMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	284.0	48.5	70.3	53.3	—	—	—
DAMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	284.0	50.4	71.4	55.5	—	—	—
DAMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	395.0	49.8	71.2	54.7	—	—	—
DAMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	395.0	51.2	72.1	56.1	—	—	—
DAMamba-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	520.0	50.6	71.9	55.5	—	—	—
DAMamba-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	520.0	51.4	72.3	56.4	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
DAMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	284.0	43.4	67.2	46.7	—	—	—
DAMamba-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	284.0	44.8	68.6	48.6	—	—	—
DAMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	395.0	44.5	68.4	48.2	—	—	—
DAMamba-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	395.0	45.1	69.2	49.1	—	—	—
DAMamba-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	520.0	44.9	68.9	48.7	—	—	—
DAMamba-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	520.0	45.3	69.5	48.9	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
DAMamba-T	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	937.0	51.2	—	—	50.3	—	—
DAMamba-S	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	1050.0	52.0	—	—	51.2	—	—
DAMamba-B	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	1178.0	52.3	—	—	51.9	—	—