List your company
ResearchFormerly WizWand

Action Recognition on SSV2

77.7Top-1 AccV-JEPA 2.1 ViT-G
5.31624.10842.961.692Mar 30, 2025Jun 7, 2025Aug 16, 2025Oct 25, 2025Jan 2, 2026Mar 13, 2026May 22, 2026
Updated 5mo ago

Evaluation results

MethodLinks
V-JEPA 2.1 ViT-GParam.=2B, Resolution=384x384, Frame clips=16, Temporal crops=2, Spatial crops=32026.0377.7
V-JEPA 2 ViT-gParam.=1B, Resolution=384x384, Frame clips=16, Temporal crops=2, Spatial crops=32026.0377.3
V-JEPA 2.1 ViT-gParam.=1B, Resolution=384x384, Frame clips=16, Temporal crops=2, Spatial crops=32026.0376.9
VJEPA2Type=Discriminative Pretraining, Pretrained on=Videos, Arch.=ViT-G, Param.=1B, Attentive probing=true2025.1275.3
V-JEPA ViT-HParam.=600M, Resolution=256x256, Frame clips=16, Temporal crops=2, Spatial crops=32026.0374.3
CASTGFLOPs/View=3912025.0371.6
OMNIVORE2025.0371.4
VJEPAType=Discriminative Pretraining, Pretrained on=Videos, Arch.=ViT-H, Param.=630M, Attentive probing=true2025.1271.4
VideoMAE-BGFLOPs/View=180, Parameter-efficient tuning with adapters=true2025.0370.8
BEVTGFLOPs/View=2822025.0370.6
DINOv3Param.=7B2026.0370.1
DINOv3 ViT-H+Param.=0.8B2026.0369.8
InternVideo2s2-1BParam.=1B, Resolution=256x256, Frame clips=16, Temporal crops=2, Spatial crops=32026.0369.7
Video Swin-LGFLOPs/View=2822025.0369.6
ST-AdapterGFLOPs/View=6072025.0369.5
NExT-Vid-GType=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-G, Param.=1.1B, Attentive probing=true2025.1269.5
MTV-HRGFLOPs/View=9302025.0368.5
VideoPrismType=Discriminative Pretraining, Pretrained on=Videos, Arch.=ViT-G, Param.=1B, Attentive probing=true2025.1268.5
VideoPrismParam.=1B2026.0368.5
AIMGFLOPs/View=404, Parameter-efficient tuning with adapters=true2025.0368.1
MFormer-HRGFLOPs/View=11852025.0368.1
ORVIT MF2025.0367.9
MVIT-BGFLOPs/View=1702025.0367.7
InternVideo2Type=Discriminative Pretraining, Pretrained on=Videos, Arch.=ViT-G, Param.=1B, Attentive probing=true2025.1267.3
InternVideo2-1BParam.=1B2026.0367.3
NExT-Vid-HType=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-H, Param.=600M, Attentive probing=true2025.1267
MVDType=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-L, Param.=200M, Attentive probing=true2025.1266.5
VideoMAEType=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-H, Param.=630M, Attentive probing=true2025.1266.2
VIVIT FEGFLOPs/View=9902025.0365.9
OmniMAEType=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-H, Param.=630M, Attentive probing=true2025.1265.4
NExT-Vid-LType=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-L, Param.=300M, Attentive probing=true2025.1263.9
EVLGFLOPs/View=5922025.0362.4
TimeSformer-BGFLOPs/View=23802025.0362.4
RVM-BSize(M)=1172025.1261.4
VideoMAEv2Type=Generative Pretraining, Pretrained on=Videos, Arch.=ViT-G, Param.=1.1B, Attentive probing=true2025.1261.2
4DS-BSize(M)=91, config=dist from e2025.1260.3
OV-Encoder (Codec)Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0260.1
RVM-SSize(M)=342025.1259.7
OV-Encoder (Frame)Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0258.7
OV-Encoder (Codec)Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0258.5
DINOv3Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0258.3
SigLIP2Backbone=ViT-L/16, Resolution=256, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0258.2
OV-Encoder (Frame)Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0257.7
DINOv3Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0257.4
AIMv2Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0257.2
VideoMAEv2Param.=1B2026.0356.1
SiamMAE-SSize(M)=272025.1256
PEcore GParam.=1.9B, Resolution=256x256, Frame clips=16, Temporal crops=2, Spatial crops=32026.0355.4
AIMv2Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0255.1
SigLIPBackbone=ViT-L/16, Resolution=256, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0252.7
SigLIP2Backbone=ViT-L/16, Resolution=256, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0252.6
VideoMAE-BSize(M)=872025.1252.3
SigLIPBackbone=ViT-L/16, Resolution=256, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0250.7
DINOv2Param.=1.1B, Resolution=256x256, Frame clips=16, Temporal crops=2, Spatial crops=32026.0350.7
MetaCLIPBackbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0250.6
DINOv2Type=Discriminative Pretraining, Pretrained on=Images, Arch.=ViT-G, Param.=1.1B, Attentive probing=true2025.1250.6
IJEPAType=Discriminative Pretraining, Pretrained on=Images, Arch.=ViT-H, Param.=630M, Attentive probing=true2025.1250
Siglip2Type=Discriminative Pretraining, Pretrained on=Images, Arch.=ViT-G, Param.=1.2B, Attentive probing=true2025.1249.9
SigLIP2Param.=1.2B, Resolution=256x256, Frame clips=16, Temporal crops=2, Spatial crops=32026.0349.9
4DS-BSize(M)=912025.1249.6
MetaCLIP2Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=16 Frames / 4096 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0249.3
CLIPBackbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0248.2
CLIPGFLOPs/View=140, Parameter-efficient tuning with adapters=true2025.0347.8
MetaCLIP2Backbone=ViT-L/14, Resolution=224, Input Configuration (Frames/Patches)=8 Frames / 2048 Patches, Evaluation Protocol (Attentive probe with frozen backbones)=Attentive probe with frozen backbones2026.0247.2
4DS-SSize(M)=242025.1239.9
OpenCLIPType=Discriminative Pretraining, Pretrained on=Images, Arch.=ViT-G, Param.=1.8B, Attentive probing=true2025.1234.8
SimVAK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0515.2
BDC-CLIPK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0514.8
GA2-CLIPK=16, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.1114.7
TC-CLIPK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0514
ViLT-CLIPK=16, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.1113.2
ViFi-CLIPK=16, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.1113
ViFi-CLIPK=16, Adaptation Protocol=Tuning pre-trained image VL models2025.1112.4
ViFi-CLIPK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0512.4
OSTK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0512.2
ActionCLIPK=16, Adaptation Protocol=Adapting pre-trained image VL models2025.1111.1
ActionCLIPK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0511.1
GA2-CLIPK=8, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.1110.6
CLIP image-FTK=16, Adaptation Protocol=Tuning pre-trained image VL models2025.1110.4
ViLT-CLIPK=8, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.1110.3
SimVAK (Shots)=8, Few-shot setting=true, Fine-tuned from CLIP=true2026.0510.3
X-CLIPK=16, Adaptation Protocol=Adapting pre-trained image VL models2025.1110
X-CLIPK (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.0510
GA2-CLIPK=4, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.119.9
ViFi-CLIPK=8, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.119.8
A5K=16, Adaptation Protocol=Adapting pre-trained image VL models2025.119.7
A5K (Shots)=16, Few-shot setting=true, Fine-tuned from CLIP=true2026.059.7
BDC-CLIPK (Shots)=8, Few-shot setting=true, Fine-tuned from CLIP=true2026.059.6
ViLT-CLIPK=4, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.119.4
TC-CLIPK (Shots)=8, Few-shot setting=true, Fine-tuned from CLIP=true2026.059.3
CLIP text-FTK=16, Adaptation Protocol=Tuning pre-trained image VL models2025.119.1
SimVAK (Shots)=4, Few-shot setting=true, Fine-tuned from CLIP=true2026.059
OSTK (Shots)=8, Few-shot setting=true, Fine-tuned from CLIP=true2026.058.9
TC-CLIPK (Shots)=4, Few-shot setting=true, Fine-tuned from CLIP=true2026.058.6
ViFi-CLIPK=8, Adaptation Protocol=Tuning pre-trained image VL models2025.118.5
ViFi-CLIPK (Shots)=8, Few-shot setting=true, Fine-tuned from CLIP=true2026.058.5
ActionCLIPK=8, Adaptation Protocol=Adapting pre-trained image VL models2025.118.4
ActionCLIPK (Shots)=8, Few-shot setting=true, Fine-tuned from CLIP=true2026.058.4
BDC-CLIPK (Shots)=4, Few-shot setting=true, Fine-tuned from CLIP=true2026.058.3
ViFi-CLIPK=4, Adaptation Protocol=Prompt tuning pre-trained image VL models2025.118.1