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/UniFormerV2-L

UniFormerV2-L

Reported on 22 benchmarks across 3 tasks

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

Computer Vision8 results

  • VideoonKinetics-700
    Top-1 Accuracy· uses extra data
    82.7
    best: 85.9 (InternVideo2-6B)
  • VideoonKinetics-700
    Top-5 Accuracy· uses extra data
    96.2
    best: 96.7 (UMT-L (ViT-L/16))
  • VideoonMiT
    Top 1 Accuracy· uses extra data
    47.8
    best: 53.1 (OmniVec2)
  • VideoonMiT
    Top 5 Accuracy· uses extra data
    76.9
    best: 78.2 (UMT-L (ViT-L/16))
  • VideoonKinetics-600
    Top-1 Accuracy· uses extra data
    90.1
    best: 91.9 (InternVideo2-6B)
  • VideoonKinetics-600
    Top-5 Accuracy· uses extra data
    98.5
    best: 98.9 (TubeVit-H)
  • VideoonActivityNet
    Top 1 Accuracy· uses extra data
    94.7
  • VideoonActivityNet
    Top 5 Accuracy· uses extra data
    99.5

Robots7 results

  • Activity RecognitiononHACS
    Top 1 Accuracy· uses extra data
    95.5
    best: 97 (InternVideo2-6B)
  • Activity RecognitiononHACS
    Top 5 Accuracy· uses extra data
    99.8
  • Activity RecognitiononSomething-Something V1
    Top 1 Accuracy· uses extra data
    62.7
    best: 70 (InternVideo)
  • Activity RecognitiononSomething-Something V1
    Top 5 Accuracy· uses extra data
    88
    best: 91.9 (VideoMAE V2-g)
  • Activity RecognitiononSomething-Something V2
    GFLOPs· uses extra data
    5154
    best: 13321 (InternVideo2-6B)
  • Activity RecognitiononSomething-Something V2
    Top-1 Accuracy· uses extra data
    73
    best: 77.3 (MVD (Kinetics400 pretrain, ViT-H, 16 frame))
  • Activity RecognitiononSomething-Something V2
    Top-5 Accuracy· uses extra data
    94.5
    best: 96.3 (DejaVid)

Time Series7 results

  • Action RecognitiononHACS
    Top 1 Accuracy· uses extra data
    95.5
    best: 97 (InternVideo2-6B)
  • Action RecognitiononHACS
    Top 5 Accuracy· uses extra data
    99.8
  • Action RecognitiononSomething-Something V1
    Top 1 Accuracy· uses extra data
    62.7
    best: 70 (InternVideo)
  • Action RecognitiononSomething-Something V1
    Top 5 Accuracy· uses extra data
    88
    best: 91.9 (VideoMAE V2-g)
  • Action RecognitiononSomething-Something V2
    GFLOPs· uses extra data
    5154
    best: 13321 (InternVideo2-6B)
  • Action RecognitiononSomething-Something V2
    Top-1 Accuracy· uses extra data
    73
    best: 77.3 (MVD (Kinetics400 pretrain, ViT-H, 16 frame))
  • Action RecognitiononSomething-Something V2
    Top-5 Accuracy· uses extra data
    94.5
    best: 96.3 (DejaVid)