Heedless Backbones

MobileViT 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)

PASCAL VOC 2012 (val)

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

Trans2Seg

DeepLabv3

PSPNet

Linear

resolution

Select an option

----------

512x2048

640x2560

axis

Select an option

Results

Parameters (M)

Images / Second

GFLOPs

Publication Date

Filters

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

No Results

model	params (m)	pretrain	finetune	gflops	IN-1k
MobileViT-XXS	1.3	IN-1k : Sup. : 300	— : — : —	None	69.0/88.9
MobileViT-XS	2.3	IN-1k : Sup. : 300	— : — : —	None	74.8/92.3
MobileViT-S	5.6	IN-1k : Sup. : 300	— : — : —	None	78.4/94.1

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
MobileViT-XS	IN-1k : Sup. : 300	SSDLite	COCO (train) : 200	1.6	24.8	—	—	—	—	—
MobileViT-S	IN-1k : Sup. : 300	SSDLite	COCO (train) : 200	None	27.7	—	—	—	—	—

PASCAL VOC 2012 (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
MobileViT-XXS	IN-1k : Sup. : 300	DeepLabv3	PASCAL VOC 2012 (train + SBD + COCO) : 50 : 512	None	—	—	—	73.6	—	—
MobileViT-XS	IN-1k : Sup. : 300	DeepLabv3	PASCAL VOC 2012 (train + SBD + COCO) : 50 : 512	5.7	—	—	—	77.1	—	—
MobileViT-S	IN-1k : Sup. : 300	DeepLabv3	PASCAL VOC 2012 (train + SBD + COCO) : 50 : 512	None	—	—	—	79.1	—	—