Heedless Backbones

TransXNet 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

---------

COCO (val)

COCO (test)

PASCAL VOC 2007 (val)

PASCAL VOC 2007 (test)

metric

Select an option

----------

mAP

AP₅₀

AP₇₅

mAP_s

mAP_m

mAP_l

GFLOPs

head

Select an option

---------

Mask R-CNN

Cascade Mask R-CNN

RetinaNet

DINO

HTC++

HTC

FCOS

Faster R-CNN

axis

Select an option

Results

Parameters (M)

Images / Second

GFLOPs

Publication Date

Filters

Legend

Legend attribute

Select an option

Family

Pretrain Dataset

Instance Head

Instance Training Epochs

Legend attribute (second)

Select an option

----------

Family

Pretrain Method

Instance Head

Instance Training Epochs

model	params (m)	pretrain	head	train	GFLOPs	mAP
TransXNet-T	12.8	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	187.0	43.1
TransXNet-T	12.8	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	205.0	44.5
TransXNet-S	26.9	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	242.0	46.4
TransXNet-S	26.9	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	261.0	47.7
TransXNet-B	48.0	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	317.0	47.6
TransXNet-B	48.0	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	336.0	48.8

model	params (m)	pretrain	finetune	gflops	IN-1k	IN-V2
TransXNet-T	12.8	IN-1k : Sup. : 300	— : — : —	1.8	81.6/—	70.7/—
TransXNet-S	26.9	IN-1k : Sup. : 300	— : — : —	4.5	83.8/—	73.8/—
TransXNet-B	48.0	IN-1k : Sup. : 300	— : — : —	8.3	84.6/—	75.0/—
TransXNet-B	48.0	IN-1k : Sup. : 300	IN-1k : 30 : 384	24.2	85.5/—	76.1/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
TransXNet-T	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	187.0	43.1	64.1	46.0	26.2	46.7	57.4
TransXNet-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	205.0	44.5	66.5	48.6	—	—	—
TransXNet-S	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	242.0	46.4	67.7	50.0	28.9	50.3	61.1
TransXNet-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	261.0	47.7	69.9	52.3	—	—	—
TransXNet-B	IN-1k : Sup. : 300	RetinaNet	COCO (train) : 12	317.0	47.6	69.0	51.1	31.3	51.7	62.2
TransXNet-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	336.0	48.8	70.8	53.5	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
TransXNet-T	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	205.0	40.6	63.7	43.8	—	—	—
TransXNet-S	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	261.0	43.1	66.9	46.4	—	—	—
TransXNet-B	IN-1k : Sup. : 300	Mask R-CNN	COCO (train) : 12	336.0	43.8	68.0	47.2	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
TransXNet-T	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 128 : 512	121.0	—	—	—	45.5	—	—
TransXNet-S	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 128 : 512	179.0	—	—	—	48.5	—	—
TransXNet-B	IN-1k : Sup. : 300	Panoptic FPN	ADE20K (train) : 128 : 512	256.0	—	—	—	49.9	—	—