Action Recognition

Action Recognition is a computer vision task that involves recognizing human actions in videos or images. The goal is to classify and categorize the actions being performed in the video or image into a predefined set of action classes.

In the video domain, it is an open question whether training an action classification network on a sufficiently large dataset, will give a similar boost in performance when applied to a different temporal task or dataset. The challenges of building video datasets has meant that most popular benchmarks for action recognition are small, having on the order of 10k videos.

Please note some benchmarks may be located in the Action Classification or Video Classification tasks, e.g. Kinetics-400.

Papers

Recently Added Most Hyped Most Active Needs Verification Most Verified

Showing 2611–2620 of 2759 papers

Title	Date	Tasks	Status
Learning Causal Domain-Invariant Temporal Dynamics for Few-Shot Action Recognition	Feb 20, 2024	Action RecognitionDecoder	—Unverified
Learning End-to-End Action Interaction by Paired-Embedding Data Augmentation	Jul 16, 2020	Action RecognitionData Augmentation	—Unverified
Learning Ensembles of Potential Functions for Structured Prediction With Latent Variables	Dec 1, 2015	Action RecognitionActivity Recognition	—Unverified
Learning from Small Sample Sets by Combining Unsupervised Meta-Training with CNNs	Dec 1, 2016	Action RecognitionGeneral Classification	—Unverified
Learning from Weakly-labeled Web Videos via Exploring Sub-Concepts	Jan 11, 2021	Action RecognitionPseudo Label	—Unverified
Learning Graphs for Knowledge Transfer With Limited Labels	Jun 19, 2021	Action RecognitionBenchmarking	—Unverified
Learning Higher-order Object Interactions for Keypoint-based Video Understanding	May 16, 2023	Action LocalizationAction Recognition	—Unverified
Learning Human Pose Models from Synthesized Data for Robust RGB-D Action Recognition	Jul 4, 2017	Action RecognitionSkeleton Based Action Recognition	—Unverified
Learning Joint Representation of Human Motion and Language	Oct 27, 2022	Action RecognitionContrastive Learning	—Unverified
Learning Latent Spatio-Temporal Compositional Model for Human Action Recognition	Feb 1, 2015	Action RecognitionTemporal Action Localization	—Unverified

Show:10 25 50

← PrevPage 262 of 276Next →

All datasets Something-Something V2 UCF101 HMDB-51 Something-Something V1 AVA v2.2 EPIC-KITCHENS-100 NTU RGB+D NTU RGB+D 120 Diving-48 ActivityNet AVA v2.1 H2O (2 Hands and Objects)

Benchmark Results

#	Model	Metric	Claimed	Verified	Status
1	MViTv2-B (IN-21K + Kinetics400 pretrain)	Top-5 Accuracy	93.4	—	Unverified
2	RSANet-R50 (8+16 frames, ImageNet pretrained, 2 clips)	Top-5 Accuracy	91.1	—	Unverified
3	MVD (Kinetics400 pretrain, ViT-H, 16 frame)	Top-1 Accuracy	77.3	—	Unverified
4	InternVideo	Top-1 Accuracy	77.2	—	Unverified
5	DejaVid	Top-1 Accuracy	77.2	—	Unverified
6	InternVideo2-1B	Top-1 Accuracy	77.1	—	Unverified
7	VideoMAE V2-g	Top-1 Accuracy	77	—	Unverified
8	MVD (Kinetics400 pretrain, ViT-L, 16 frame)	Top-1 Accuracy	76.7	—	Unverified
9	Hiera-L (no extra data)	Top-1 Accuracy	76.5	—	Unverified
10	TubeViT-L	Top-1 Accuracy	76.1	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	FTP-UniFormerV2-L/14	3-fold Accuracy	99.7	—	Unverified
2	OmniVec	3-fold Accuracy	99.6	—	Unverified
3	VideoMAE V2-g	3-fold Accuracy	99.6	—	Unverified
4	OmniVec2	3-fold Accuracy	99.6	—	Unverified
5	BIKE	3-fold Accuracy	98.8	—	Unverified
6	SMART	3-fold Accuracy	98.64	—	Unverified
7	ZeroI2V ViT-L/14	3-fold Accuracy	98.6	—	Unverified
8	PERF-Net (multi-distilled S3D)	3-fold Accuracy	98.6	—	Unverified
9	OmniSource (SlowOnly-8x8-R101-RGB + I3D-Flow)	3-fold Accuracy	98.6	—	Unverified
10	Text4Vis	3-fold Accuracy	98.2	—	Unverified