Heedless Backbones

MaxViT 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)

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

axis

Select an option

Results

Parameters (M)

Images / Second

GFLOPs

Publication Date

Filters

pretrain dataset

Select an option

---------

JFT-3B

ImageNet-1k

ImageNet-22k

JFT-300M

pretrain method

Select an option

----------

Supervised

FCMAE

MAE

CL

Legend

Legend attribute

Select an option

Family

Pretrain Dataset

Pretrain Method

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
MaxViT-T	31.0	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	475.0	52.1
MaxViT-S	69.0	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	595.0	53.1
MaxViT-B	120.0	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	856.0	53.4

model	params (m)	pretrain	finetune	gflops	IN-1k
MaxViT-T	31.0	IN-1k : Sup. : 300	— : — : —	5.6	83.62/—
MaxViT-T	31.0	IN-1k : Sup. : 300	IN-1k : 30 : 384	17.7	85.24/—
MaxViT-T	31.0	IN-1k : Sup. : 300	IN-1k : 30 : 512	33.7	85.72/—
MaxViT-S	69.0	IN-1k : Sup. : 300	— : — : —	11.7	84.45/—
MaxViT-S	69.0	IN-1k : Sup. : 300	IN-1k : 30 : 384	36.1	85.74/—
MaxViT-S	69.0	IN-1k : Sup. : 300	IN-1k : 30 : 512	67.6	86.19/—
MaxViT-B	120.0	IN-1k : Sup. : 300	— : — : —	23.4	84.95/—
MaxViT-B	120.0	IN-1k : Sup. : 300	IN-1k : 30 : 384	74.2	86.34/—
MaxViT-B	120.0	IN-1k : Sup. : 300	IN-1k : 30 : 512	138.5	86.66/—
MaxViT-B	120.0	IN-22k : Sup. : 90	IN-1k : 30 : 384	74.2	88.24/—
MaxViT-B	120.0	IN-22k : Sup. : 90	IN-1k : 30 : 512	138.3	88.38/—
MaxViT-L	212.0	IN-1k : Sup. : 300	— : — : —	43.9	85.17/—
MaxViT-L	212.0	IN-1k : Sup. : 300	IN-1k : 30 : 384	133.1	86.4/—
MaxViT-L	212.0	IN-1k : Sup. : 300	IN-1k : 30 : 512	245.4	86.7/—
MaxViT-L	212.0	IN-22k : Sup. : 90	IN-1k : 30 : 384	128.7	88.32/—
MaxViT-L	212.0	IN-22k : Sup. : 90	IN-1k : 30 : 512	245.2	88.46/—
MaxViT-XL	475.0	IN-22k : Sup. : 90	IN-1k : 30 : 384	293.7	88.51/—
MaxViT-XL	475.0	IN-22k : Sup. : 90	IN-1k : 30 : 512	535.2	88.7/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
MaxViT-T	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	475.0	52.1	71.9	56.8	—	—	—
MaxViT-S	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	595.0	53.1	72.5	58.1	—	—	—
MaxViT-B	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	856.0	53.4	72.9	58.1	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
MaxViT-T	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	475.0	44.6	69.1	48.4	—	—	—
MaxViT-S	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	595.0	45.4	69.8	49.5	—	—	—
MaxViT-B	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	856.0	45.7	70.3	50.0	—	—	—