SOTAVerified

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 AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 551–600 of 2759 papers

TitleStatusHype
DSANet: Dynamic Segment Aggregation Network for Video-Level Representation LearningCode1
Decoupled Spatial-Temporal Attention Network for Skeleton-Based Action RecognitionCode1
Graph in Graph Neural NetworkCode1
Graph Convolution with Low-rank Learnable Local FiltersCode1
Co-occurrence Feature Learning from Skeleton Data for Action Recognition and Detection with Hierarchical AggregationCode1
Approximated Bilinear Modules for Temporal ModelingCode1
Actions as Moving PointsCode1
Dynamic Perceiver for Efficient Visual RecognitionCode1
EgoVLPv2: Egocentric Video-Language Pre-training with Fusion in the BackboneCode1
Hierarchical Contrast for Unsupervised Skeleton-based Action Representation LearningCode1
Hierarchically Self-Supervised Transformer for Human Skeleton Representation LearningCode1
FLAVR: Flow-Agnostic Video Representations for Fast Frame InterpolationCode1
Higher Order Recurrent Space-Time Transformer for Video Action PredictionCode1
Counterfactual Debiasing Inference for Compositional Action RecognitionCode1
InHARD - Industrial Human Action Recognition Dataset in the Context of Industrial Collaborative RoboticsCode1
Home Action Genome: Cooperative Compositional Action UnderstandingCode1
MoViNets: Mobile Video Networks for Efficient Video RecognitionCode1
Deep Analysis of CNN-based Spatio-temporal Representations for Action RecognitionCode1
Cross-Architecture Self-supervised Video Representation LearningCode1
Hybrid Relation Guided Set Matching for Few-shot Action RecognitionCode1
HyRSM++: Hybrid Relation Guided Temporal Set Matching for Few-shot Action RecognitionCode1
ImageNet-21K Pretraining for the MassesCode1
MViTv2: Improved Multiscale Vision Transformers for Classification and DetectionCode1
Improved Residual Networks for Image and Video RecognitionCode1
Self-distillation for surgical action recognitionCode1
IntegralAction: Pose-driven Feature Integration for Robust Human Action Recognition in VideosCode1
ARID: A New Dataset for Recognizing Action in the DarkCode1
Integrating Human Parsing and Pose Network for Human Action RecognitionCode1
Composable Augmentation Encoding for Video Representation Learning—0
Complex Video Action Reasoning via Learnable Markov Logic Network—0
Complex Human Action Recognition in Live Videos Using Hybrid FR-DL Method—0
Ani-GIFs: A benchmark dataset for domain generalization of action recognition from GIFs—0
Action Recognition with Coarse-to-Fine Deep Feature Integration and Asynchronous Fusion—0
Comparative Validation of Machine Learning Algorithms for Surgical Workflow and Skill Analysis with the HeiChole Benchmark—0
Comparative Evaluation of Action Recognition Methods via Riemannian Manifolds, Fisher Vectors and GMMs: Ideal and Challenging Conditions—0
Action Recognition via Pose-Based Graph Convolutional Networks with Intermediate Dense Supervision—0
ActAR: Actor-Driven Pose Embeddings for Video Action Recognition—0
A New Representation of Skeleton Sequences for 3D Action Recognition—0
Distributed non-parametric deep and wide networks—0
Distribution of Action Movements (DAM): A Descriptor for Human Action Recognition—0
CompactFlowNet: Efficient Real-time Optical Flow Estimation on Mobile Devices—0
Combining Spatio-Temporal Appearance Descriptors and Optical Flow for Human Action Recognition in Video Data—0
A New Adjacency Matrix Configuration in GCN-based Models for Skeleton-based Action Recognition—0
Combining Deep Learning Classifiers for 3D Action Recognition—0
Combining ConvNets with Hand-Crafted Features for Action Recognition Based on an HMM-SVM Classifier—0
A New Action Recognition Framework for Video Highlights Summarization in Sporting Events—0
Action Recognition Utilizing YGAR Dataset—0
Combined CNN Transformer Encoder for Enhanced Fine-grained Human Action Recognition—0
Combating Missing Modalities in Egocentric Videos at Test Time—0
Colo-SCRL: Self-Supervised Contrastive Representation Learning for Colonoscopic Video Retrieval—0
Show:102550
← PrevPage 12 of 56Next →
All datasetsSomething-Something V2UCF101HMDB-51Something-Something V1AVA v2.2EPIC-KITCHENS-100NTU RGB+DNTU RGB+D 120Diving-48ActivityNetAVA v2.1H2O (2 Hands and Objects)

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1MViTv2-B (IN-21K + Kinetics400 pretrain)Top-5 Accuracy93.4—Unverified
2RSANet-R50 (8+16 frames, ImageNet pretrained, 2 clips)Top-5 Accuracy91.1—Unverified
3MVD (Kinetics400 pretrain, ViT-H, 16 frame)Top-1 Accuracy77.3—Unverified
4InternVideoTop-1 Accuracy77.2—Unverified
5DejaVidTop-1 Accuracy77.2—Unverified
6InternVideo2-1BTop-1 Accuracy77.1—Unverified
7VideoMAE V2-gTop-1 Accuracy77—Unverified
8MVD (Kinetics400 pretrain, ViT-L, 16 frame)Top-1 Accuracy76.7—Unverified
9Hiera-L (no extra data)Top-1 Accuracy76.5—Unverified
10TubeViT-LTop-1 Accuracy76.1—Unverified
#ModelMetricClaimedVerifiedStatus
1FTP-UniFormerV2-L/143-fold Accuracy99.7—Unverified
2OmniVec23-fold Accuracy99.6—Unverified
3OmniVec3-fold Accuracy99.6—Unverified
4VideoMAE V2-g3-fold Accuracy99.6—Unverified
5BIKE3-fold Accuracy98.8—Unverified
6SMART3-fold Accuracy98.64—Unverified
7ZeroI2V ViT-L/143-fold Accuracy98.6—Unverified
8OmniSource (SlowOnly-8x8-R101-RGB + I3D-Flow)3-fold Accuracy98.6—Unverified
9PERF-Net (multi-distilled S3D)3-fold Accuracy98.6—Unverified
10Text4Vis3-fold Accuracy98.2—Unverified