Heedless Backbones

FractalMamba++ 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

---------

Cityscapes (val)

Cityscapes (test)

ADE20K (val)

ADE20K (test)

PASCAL VOC 2007 (val)

PASCAL VOC 2007 (test)

metric

Select an option

mIoU^ms

pAcc^ms

mAcc^ms

mIoU^ss

pAcc^ss

mAcc^ss

GFLOPs

head

Select an option

---------

UPerNet

Mask2Former

Panoptic FPN

SETR

resolution

Select an option

----------

512x2048

640x2560

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

Semantic Segmentation Head

Semantic Segmentation Resolution

Semantic Segmentation Training Epochs

Legend attribute (second)

Select an option

----------

Family

Pretrain Method

Semantic Segmentation Head

Semantic Segmentation Resolution

Semantic Segmentation Training Epochs

model	params (m)	pretrain	head	train	GFLOPs	mIoU^ms
FractalMamba++ (T)	30.0	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 640 : 512	957.0	49.7
FractalMamba++ (M)	11.0	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 640 : 512	889.0	46.2
FractalMamba++ (N)	7.0	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 640 : 512	860.0	45.0

model	params (m)	pretrain	finetune	gflops	IN-1k
FractalMamba++ (T)	30.0	IN-1k : Sup. : 300	— : — : —	None	67.5/—
FractalMamba++ (M)	11.0	IN-1k : Sup. : 300	— : — : —	None	56.1/—
FractalMamba++ (N)	7.0	IN-1k : Sup. : 300	— : — : —	None	42.5/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
FractalMamba++ (T)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	260.0	47.8	69.8	52.5	—	—	—
FractalMamba++ (T)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	260.0	49.5	71.0	54.6	—	—	—
FractalMamba++ (M)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	199.0	44.1	66.2	48.0	—	—	—
FractalMamba++ (M)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	199.0	46.8	68.5	51.3	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
FractalMamba++ (T)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	260.0	43.2	66.8	46.6	—	—	—
FractalMamba++ (T)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	260.0	44.1	67.9	47.5	—	—	—
FractalMamba++ (M)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	199.0	40.3	63.2	43.4	—	—	—
FractalMamba++ (M)	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	199.0	42.2	65.4	45.3	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
FractalMamba++ (T)	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 640 : 512	957.0	49.7	—	—	48.8	—	—
FractalMamba++ (M)	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 640 : 512	889.0	46.2	—	—	45.8	—	—
FractalMamba++ (N)	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 640 : 512	860.0	45.0	—	—	44.5	—	—