Heedless Backbones

LBVim 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
LBVim-Ti	6.0	IN-1k : Sup. : 300	— : — : —	1.4	73.7
LBVim-300	15.0	IN-1k : Sup. : 300	— : — : —	3.1	77.7
LBVim-S	24.0	IN-1k : Sup. : 300	— : — : —	4.9	79.6
LBVim-528	44.0	IN-1k : Sup. : 300	— : — : —	9.0	81.1

model	params (m)	pretrain	finetune	gflops	IN-1k
LBVim-Ti	6.0	IN-1k : Sup. : 300	— : — : —	1.4	73.7/—
LBVim-300	15.0	IN-1k : Sup. : 300	— : — : —	3.1	77.7/—
LBVim-S	24.0	IN-1k : Sup. : 300	— : — : —	4.9	79.6/—
LBVim-528	44.0	IN-1k : Sup. : 300	— : — : —	9.0	81.1/—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
LBVim-Ti	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	141.0	—	—	—	40.2	—	—
LBVim-300	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	178.0	—	—	—	43.7	—	—
LBVim-S	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	219.0	—	—	—	44.2	—	—
LBVim-528	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	306.0	—	—	—	45.5	—	—