TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Action Recognition	AVA v2.1	SlowFast (Kinetics-600 pretraining)	mAP (Val)	26.8	# 8
Action Recognition	AVA v2.1	SlowFast (Kinetics-400 pretraining)	mAP (Val)	26.3	# 9
Action Recognition	AVA v2.1	SlowFast++ (Kinetics-600 pretraining, NL)	mAP (Val)	28.3	# 4
Action Recognition	AVA v2.1	SlowFast (Kinetics-600 pretraining, NL)	mAP (Val)	27.3	# 7
Action Recognition	AVA v2.2	SlowFast, 16x8 R101+NL (Kinetics-600 pretraining)	mAP	27.5	# 28
Action Recognition	AVA v2.2	SlowFast, 8x8, R101 (Kinetics-400 pretraining)	mAP	23.8	# 37
Action Recognition	AVA v2.2	SlowFast, 4x16, R50 (Kinetics-400 pretraining)	mAP	21.9	# 38
Action Recognition	AVA v2.2	SlowFast, 8x8 R101+NL (Kinetics-600 pretraining)	mAP	27.1	# 31
Action Classification	Charades	SlowFast (Kinetics-600 pretraining, NL)	MAP	45.2	# 18
Action Classification	Charades	SlowFast (Kinetics-600 pretraining)	MAP	42.1	# 27
Action Classification	Charades	SlowFast (Kinetics-400 pretraining, NL)	MAP	42.5	# 25
Action Recognition	Diving-48	SlowFast	Accuracy	77.6	# 16
Action Recognition	H2O (2 Hands and Objects)	SlowFast	Actions Top-1	77.69	# 9
Action Recognition	H2O (2 Hands and Objects)	SlowFast	RGB	Yes	# 1
Action Recognition	H2O (2 Hands and Objects)	SlowFast	Hand Pose	No	# 1
Action Recognition	H2O (2 Hands and Objects)	SlowFast	Object Pose	No	# 1
Action Recognition	H2O (2 Hands and Objects)	SlowFast	Object Label	No	# 1
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101 + NL)	Acc@5	93.9	# 81
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101 + NL)	Acc@1	79.8	# 99
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101)	Acc@1	78.9	# 115
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101)	Acc@5	93.5	# 91
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-101)	Acc@1	77.9	# 125
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-101)	Acc@5	93.2	# 97
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-50)	Acc@1	77	# 139
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-50)	Acc@5	92.6	# 103
Action Classification	Kinetics-400	SlowFast 4x16 (ResNet-50)	Acc@1	75.6	# 149
Action Classification	Kinetics-400	SlowFast 4x16 (ResNet-50)	Acc@5	92.1	# 107
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101 + NL)	Top-1 Accuracy	81.8	# 46
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101 + NL)	Top-5 Accuracy	95.1	# 39
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101)	Top-1 Accuracy	81.1	# 49
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101)	Top-5 Accuracy	95.1	# 39
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-101)	Top-1 Accuracy	80.4	# 51
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-101)	Top-5 Accuracy	94.8	# 42
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-50)	Top-1 Accuracy	79.9	# 52
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-50)	Top-5 Accuracy	94.5	# 43
Action Classification	Kinetics-600	SlowFast 4x16 (ResNet-50)	Top-1 Accuracy	78.8	# 54
Action Classification	Kinetics-600	SlowFast 4x16 (ResNet-50)	Top-5 Accuracy	94	# 44
Action Recognition	Something-Something V2	SlowFast	Top-1 Accuracy	61.7	# 107

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-recognition-in-videos-on-ava-v21)](https://paperswithcode.com/sota/action-recognition-in-videos-on-ava-v21?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-recognition-on-h2o-2-hands-and-objects)](https://paperswithcode.com/sota/action-recognition-on-h2o-2-hands-and-objects?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-recognition-on-diving-48)](https://paperswithcode.com/sota/action-recognition-on-diving-48?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-classification-on-charades)](https://paperswithcode.com/sota/action-classification-on-charades?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-recognition-on-ava-v2-2)](https://paperswithcode.com/sota/action-recognition-on-ava-v2-2?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-classification-on-kinetics-600)](https://paperswithcode.com/sota/action-classification-on-kinetics-600?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-classification-on-kinetics-400)](https://paperswithcode.com/sota/action-classification-on-kinetics-400?p=slowfast-networks-for-video-recognition)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/slowfast-networks-for-video-recognition/action-recognition-in-videos-on-something)](https://paperswithcode.com/sota/action-recognition-in-videos-on-something?p=slowfast-networks-for-video-recognition)`

SlowFast Networks for Video Recognition

ICCV 2019 · Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, Kaiming He ·

We present SlowFast networks for video recognition. Our model involves (i) a Slow pathway, operating at low frame rate, to capture spatial semantics, and (ii) a Fast pathway, operating at high frame rate, to capture motion at fine temporal resolution. The Fast pathway can be made very lightweight by reducing its channel capacity, yet can learn useful temporal information for video recognition. Our models achieve strong performance for both action classification and detection in video, and large improvements are pin-pointed as contributions by our SlowFast concept. We report state-of-the-art accuracy on major video recognition benchmarks, Kinetics, Charades and AVA. Code has been made available at: https://github.com/facebookresearch/SlowFast

PDF Abstract ICCV 2019 PDF ICCV 2019 Abstract

Code

Add Remove Mark official

facebookresearch/SlowFast official

6,541

open-mmlab/mmaction2

4,209

facebookresearch/pytorchvideo

↳ Quickstart in

Colab

Spaces

PyTorch Hub

3,293

PaddlePaddle/PaddleVideo

1,505

wangxiang1230/SSTAP

See all 15 implementations

Tasks

Add Remove

Action Classification

Action Detection

Action Recognition

Action Recognition In Videos

General Classification

Video Recognition

Datasets

ImageNet

Kinetics

Kinetics 400

Charades

Something-Something V2

Kinetics-600

AVA

H2O (2 Hands and Objects)

Results from the Paper

Edit

Ranked #4 on Action Recognition on AVA v2.1

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Action Recognition	AVA v2.1	SlowFast (Kinetics-600 pretraining)	mAP (Val)	26.8	# 8	Compare
Action Recognition	AVA v2.1	SlowFast (Kinetics-400 pretraining)	mAP (Val)	26.3	# 9	Compare
Action Recognition	AVA v2.1	SlowFast++ (Kinetics-600 pretraining, NL)	mAP (Val)	28.3	# 4	Compare
Action Recognition	AVA v2.1	SlowFast (Kinetics-600 pretraining, NL)	mAP (Val)	27.3	# 7	Compare
Action Recognition	AVA v2.2	SlowFast, 4x16, R50 (Kinetics-400 pretraining)	mAP	21.9	# 38	Compare
Action Recognition	AVA v2.2	SlowFast, 8x8 R101+NL (Kinetics-600 pretraining)	mAP	27.1	# 31	Compare
Action Classification	Charades	SlowFast (Kinetics-600 pretraining, NL)	MAP	45.2	# 18	Compare
Action Classification	Charades	SlowFast (Kinetics-600 pretraining)	MAP	42.1	# 27	Compare
Action Classification	Charades	SlowFast (Kinetics-400 pretraining, NL)	MAP	42.5	# 25	Compare
Action Recognition	H2O (2 Hands and Objects)	SlowFast	Actions Top-1	77.69	# 9	Compare
			RGB	Yes	# 1	Compare
			Hand Pose	No	# 1	Compare
			Object Pose	No	# 1	Compare
			Object Label	No	# 1	Compare
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101 + NL)	Acc@5	93.9	# 81	Compare
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101 + NL)	Acc@1	79.8	# 99	Compare
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101)	Acc@1	78.9	# 115	Compare
Action Classification	Kinetics-400	SlowFast 16x8 (ResNet-101)	Acc@5	93.5	# 91	Compare
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-101)	Acc@1	77.9	# 125	Compare
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-101)	Acc@5	93.2	# 97	Compare
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-50)	Acc@1	77	# 139	Compare
Action Classification	Kinetics-400	SlowFast 8x8 (ResNet-50)	Acc@5	92.6	# 103	Compare
Action Classification	Kinetics-400	SlowFast 4x16 (ResNet-50)	Acc@1	75.6	# 149	Compare
Action Classification	Kinetics-400	SlowFast 4x16 (ResNet-50)	Acc@5	92.1	# 107	Compare
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101 + NL)	Top-1 Accuracy	81.8	# 46	Compare
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101 + NL)	Top-5 Accuracy	95.1	# 39	Compare
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101)	Top-1 Accuracy	81.1	# 49	Compare
Action Classification	Kinetics-600	SlowFast 16x8 (ResNet-101)	Top-5 Accuracy	95.1	# 39	Compare
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-101)	Top-1 Accuracy	80.4	# 51	Compare
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-101)	Top-5 Accuracy	94.8	# 42	Compare
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-50)	Top-1 Accuracy	79.9	# 52	Compare
Action Classification	Kinetics-600	SlowFast 8x8 (ResNet-50)	Top-5 Accuracy	94.5	# 43	Compare
Action Classification	Kinetics-600	SlowFast 4x16 (ResNet-50)	Top-1 Accuracy	78.8	# 54	Compare
Action Classification	Kinetics-600	SlowFast 4x16 (ResNet-50)	Top-5 Accuracy	94	# 44	Compare

Results from Other Papers

Task	Dataset	Model	Metric Name	Metric Value	Rank	Compare
Action Recognition	AVA v2.2	SlowFast, 16x8 R101+NL (Kinetics-600 pretraining)	mAP	27.5	# 28	See all
Action Recognition	AVA v2.2	SlowFast, 8x8, R101 (Kinetics-400 pretraining)	mAP	23.8	# 37	See all
Action Recognition	Diving-48	SlowFast	Accuracy	77.6	# 16	See all
Action Recognition	Something-Something V2	SlowFast	Top-1 Accuracy	61.7	# 107	See all

Methods

Add Remove

No methods listed for this paper. Add relevant methods here

Edit Social Preview

SlowFast Networks for Video Recognition

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit