Heedless Backbones

P2T 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

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

Classification Resolution

Legend attribute (second)

Select an option

----------

Family

Pretrain Method

Classification Resolution

model	params (m)	pretrain	finetune	GFLOPs	Top-1
P2T-Tiny	11.6	IN-1k : Sup. : 300	— : — : —	1.8	79.8
P2T-Small	24.1	IN-1k : Sup. : 300	— : — : —	3.7	82.4
P2T-Base	36.1	IN-1k : Sup. : 300	— : — : —	6.5	83.5
P2T-Large	54.5	IN-1k : Sup. : 300	— : — : —	9.8	83.9

model	params (m)	pretrain	finetune	gflops	IN-1k
P2T-Tiny	11.6	IN-1k : Sup. : 300	— : — : —	1.8	79.8/94.9
P2T-Small	24.1	IN-1k : Sup. : 300	— : — : —	3.7	82.4/96.0
P2T-Base	36.1	IN-1k : Sup. : 300	— : — : —	6.5	83.5/96.6
P2T-Large	54.5	IN-1k : Sup. : 300	— : — : —	9.8	83.9/96.7

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
P2T-Tiny	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	206.0	41.3	62.0	44.1	24.6	44.8	56.0
P2T-Tiny	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	225.0	43.3	65.7	47.3	—	—	—
P2T-Small	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	260.0	44.4	65.3	47.6	27.0	48.3	59.4
P2T-Small	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	279.0	45.5	67.7	49.8	—	—	—
P2T-Base	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	344.0	46.1	67.5	49.6	30.2	50.6	60.9
P2T-Base	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	363.0	47.2	69.3	51.6	—	—	—
P2T-Large	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	449.0	47.2	68.4	50.9	32.4	51.6	62.2
P2T-Large	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	467.0	48.3	70.2	53.3	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
P2T-Tiny	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	225.0	39.6	62.5	42.3	—	—	—
P2T-Small	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	279.0	41.4	64.6	44.5	—	—	—
P2T-Base	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	363.0	42.7	66.1	45.9	—	—	—
P2T-Large	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	467.0	43.5	67.3	46.9	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
P2T-Tiny	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 64 : 512	31.6	—	—	—	43.4	80.8	54.5
P2T-Small	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 64 : 512	42.7	—	—	—	46.7	82.0	58.4
P2T-Base	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 64 : 512	58.5	—	—	—	48.7	82.9	60.7
P2T-Large	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 64 : 512	77.7	—	—	—	49.4	83.3	61.9