Heedless Backbones

ConvNeXt-dcls 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
ConvNeXt-dcls-T	28.6	IN-1k : Sup. : 300	— : — : —	5.0	82.5
ConvNeXt-dcls-S	50.2	IN-1k : Sup. : 300	— : — : —	9.5	83.7
ConvNeXt-dcls-B	88.6	IN-1k : Sup. : 300	— : — : —	16.5	84.1

model	params (m)	pretrain	finetune	gflops	IN-1k	IN-C-bar↓	IN-A	IN-R	IN-Sketch
ConvNeXt-dcls-T	28.6	IN-1k : Sup. : 300	— : — : —	5.0	82.5/—	39.7/—	23.9/—	47.8/—	34.7/—
ConvNeXt-dcls-S	50.2	IN-1k : Sup. : 300	— : — : —	9.5	83.7/—	35.2/—	33.7/—	50.4/—	36.7/—
ConvNeXt-dcls-B	88.6	IN-1k : Sup. : 300	— : — : —	16.5	84.1/—	34.3/—	36.8/—	52.6/—	38.4/—

COCO (val)

model	pretrain	head	train	gflops	mAP^b	AP^b₅₀	AP^b₇₅	mAP^b_s	mAP^b_m	mAP^b_l
ConvNeXt-dcls-T	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	751.0	51.2	69.9	55.7	—	—	—
ConvNeXt-dcls-S	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	844.0	52.8	71.6	57.6	—	—	—
ConvNeXt-dcls-B	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	987.0	53.0	71.5	57.7	—	—	—

COCO (val)

model	pretrain	head	train	gflops	mAP^m	AP^m₅₀	AP^m₇₅	mAP^m_s	mAP^m_m	mAP^m_l
ConvNeXt-dcls-T	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	751.0	44.5	67.5	48.3	—	—	—
ConvNeXt-dcls-S	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	844.0	45.6	69.0	49.3	—	—	—
ConvNeXt-dcls-B	IN-1k : Sup. : 300	Cascade Mask R-CNN	COCO (train) : 36	987.0	46.0	69.3	50.0	—	—	—

ADE20K (val)

model	pretrain	head	train	gflops	mIoU^ms	pAcc^ms	mAcc^ms	mIoU^ss	pAcc^ss	mAcc^ss
ConvNeXt-dcls-T	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	950.0	—	—	—	47.1	—	—
ConvNeXt-dcls-S	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	1045.0	—	—	—	48.4	—	—
ConvNeXt-dcls-B	IN-1k : Sup. : 300	UPerNet	ADE20K (train) : 128 : 512	1193.0	—	—	—	49.3	—	—