TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Models/SkateFormer

SkateFormer

Reported on 60 benchmarks across 9 tasks · 1 paper · 12 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision25 results

  • VideoonN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Human Interaction RecognitiononNTU RGB+D
    Accuracy (Cross-Subject)· 2024-03-14
    97.1
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Human Interaction RecognitiononNTU RGB+D
    Accuracy (Cross-View)· 2024-03-14
    99.3
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Human Interaction RecognitiononNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    93.2
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Human Interaction RecognitiononNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    92.3
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • VideoonNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 92.2 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • VideoonNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 90.9 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • VideoonNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • VideoonNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 94.3 (Hulk(Finetune, ViT-L))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • VideoonNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 98.3 (ST-GCN [PYSKL, 2D Skeleton])
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • VideoonNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 92.2 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 90.9 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 94.3 (Hulk(Finetune, ViT-L))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 98.3 (ST-GCN [PYSKL, 2D Skeleton])
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Temporal Action LocalizationonNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 92.2 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 90.9 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 94.3 (Hulk(Finetune, ViT-L))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 98.3 (ST-GCN [PYSKL, 2D Skeleton])
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action LocalizationonNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508

Time Series14 results

  • Action DetectiononN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action DetectiononNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 92.2 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action DetectiononNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 90.9 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action DetectiononNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action DetectiononNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 94.3 (Hulk(Finetune, ViT-L))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action DetectiononNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 98.3 (ST-GCN [PYSKL, 2D Skeleton])
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action DetectiononNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 96.7 (DSCNet (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 95.6 (DSCNet (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 97.4 (DSCNet (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 99.6 (PoseC3D (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Action RecognitiononNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508

Methodology7 results

  • Zero-Shot LearningonN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Zero-Shot LearningonNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 92.2 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Zero-Shot LearningonNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 90.9 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Zero-Shot LearningonNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Zero-Shot LearningonNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 94.3 (Hulk(Finetune, ViT-L))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Zero-Shot LearningonNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 98.3 (ST-GCN [PYSKL, 2D Skeleton])
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Zero-Shot LearningonNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508

Robots7 results

  • Activity RecognitiononN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Activity RecognitiononNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 96.7 (DSCNet (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Activity RecognitiononNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 95.6 (DSCNet (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Activity RecognitiononNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Activity RecognitiononNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 97.4 (DSCNet (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Activity RecognitiononNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 99.6 (PoseC3D (RGB + Pose))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • Activity RecognitiononNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508

Natural Language Processing7 results

  • 3D Action RecognitiononN-UCLA
    Accuracy· 2024-03-14
    98.3
    best: 99.1 (DSCNet (RGB + Pose))
    SOTA
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • 3D Action RecognitiononNTU RGB+D 120
    Accuracy (Cross-Setup)· 2024-03-14
    91.4
    best: 92.2 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • 3D Action RecognitiononNTU RGB+D 120
    Accuracy (Cross-Subject)· 2024-03-14
    89.8
    best: 90.9 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • 3D Action RecognitiononNTU RGB+D 120
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • 3D Action RecognitiononNTU RGB+D
    Accuracy (CS)· 2024-03-14
    93.5
    best: 94.3 (Hulk(Finetune, ViT-L))
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • 3D Action RecognitiononNTU RGB+D
    Accuracy (CV)· 2024-03-14
    97.8
    best: 98.3 (ST-GCN [PYSKL, 2D Skeleton])
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508
  • 3D Action RecognitiononNTU RGB+D
    Ensembled Modalities· 2024-03-14
    4
    best: 6 (ProtoGCN)
    SkateFormer: Skeletal-Temporal Transformer for Human Action RecognitionarXiv:2403.09508