Heedless Backbones

TailProp 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
TailProp-T	29.0	IN-1k : Sup. : 300	— : — : —	4.5	82.9
TailProp-S	50.0	IN-1k : Sup. : 300	— : — : —	8.0	84.1
TailProp-B	68.0	IN-1k : Sup. : 300	— : — : —	11.0	84.4

model	params (m)	pretrain	finetune	gflops	IN-1k	IN-A	IN-Sketch
TailProp-T	29.0	IN-1k : Sup. : 300	— : — : —	4.5	82.9/—	—/—	—/—
TailProp-S	50.0	IN-1k : Sup. : 300	— : — : —	8.0	84.1/—	—/—	—/—
TailProp-B	68.0	IN-1k : Sup. : 300	— : — : —	11.0	84.4/—	37.2/—	23.1/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
TailProp-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	271.0	46.2	—	—	—	—	—
TailProp-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	271.0	47.8	—	—	—	—	—
TailProp-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	346.0	47.4	—	—	—	—	—
TailProp-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	346.0	49.4	—	—	—	—	—
TailProp-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	432.0	48.2	—	—	—	—	—
TailProp-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	432.0	50.3	—	—	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
TailProp-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	271.0	41.8	—	—	—	—	—
TailProp-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	271.0	43.1	—	—	—	—	—
TailProp-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	346.0	42.8	—	—	—	—	—
TailProp-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	346.0	44.2	—	—	—	—	—
TailProp-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	432.0	43.4	—	—	—	—	—
TailProp-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 36	432.0	44.8	—	—	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
TailProp-T	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	234.0	—	—	—	47.8	—	—
TailProp-S	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	253.0	—	—	—	50.0	—	—
TailProp-B	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	291.0	—	—	—	50.8	—	—