List your company
ResearchFormerly WizWand

Video Classification on UCF101 (3-split average)

98.6AccuracyPERF-Net
84.24887.97491.795.426Dec 22, 2015Jan 28, 2017Mar 7, 2018Apr 14, 2019May 21, 2020Jun 28, 2021Aug 6, 2022
Updated 8mo ago

Evaluation results

MethodLinks
PERF-NetPretrain=ImageNet+K700, Modalities=RGB+Flow+Pose2022.0898.6
SlowOnly-R101-RGB + I3D-FlowPretrain=OmniSource, Modalities=RGB+Flow2022.0898.6
SMARTPretrain=ImageNet+K400, Modalities=RGB+Flow2022.0898.6
EVL ViT-L/14@336pxPretrain=CLIP+K400, Modalities=RGB, Resolution=336px2022.0898.6
EVL ViT-L/14Pretrain=CLIP+K400, Modalities=RGB, Resolution=224px2022.0898.5
Two-Stream LGD-3DPretrain=ImageNet+K600, Modalities=RGB+Flow2022.0898.2
I3DRGB+FlowBackbone=3D Inception2021.0398
Two-Stream I3DPretrain=ImageNet+K400, Modalities=RGB+Flow2022.0898
BQNBackbone=TSM R50, Inference Protocol=3 crops x 2 clips2021.0397.6
SlowOnly-R101Pretrain=OmniSource, Modalities=RGB2022.0897.3
LGD-3DPretrain=ImageNet+K600, Modalities=RGB2022.0897
TEABackbone=R502021.0396.9
FASTER32Pretrain=K400, Modalities=RGB2022.0896.9
R(2+1)D-34Pretrain=K400, Modalities=RGB2022.0896.8
S3DPretrain=ImageNet+K400, Modalities=RGB2022.0896.8
PAN FullBackbone=TSM R502021.0396.5
STMBackbone=R502021.0396.2
TSMBackbone=R502021.0395.9
STCPretrain=K400, Modalities=RGB2022.0895.8
I3DPretrain=ImageNet+K400, Modalities=RGB2022.0895.6
Dynamic Image Network2020.0295.5
DI Four-StreamBackbone=ResNeXt1012021.0395.5
Temporal Squeeze Network2020.0295.2
TVNetBackbone=BNInception2021.0394.5
TSNBackbone=BN-Inception2020.0294.2
TSNRGB+FlowBackbone=BNInception2021.0394
ECOPretrain=K400, Modalities=RGB2022.0893.6
VideoPromptPretrain=CLIP, Modalities=RGB2022.0893.6
Two-Stream Fusion+iDT2020.0293.5
StNetBackbone=R502021.0393.5
Two-Stream I3D2020.0293.4
TDD+iDT2020.0291.5
VGG16 + Images + IDT-FVbackbone=VGG16, training_data=video + web action images, motion_fusion=IDT-FV2015.1291.1
VGG19 + Images + IDT-FVbackbone=VGG19, training_data=video + web action images, motion_fusion=IDT-FV2015.1290.8
C3D+iDT+SVM2020.0290.4
RCNN using LSTMtemporal_modeling=LSTM2015.1288.6
Two-stream CNNinput=RGB + Optical Flow2015.1288
Two-Streamfusion=SVM2020.0288
iDT+HSV2020.0287.9
IDT-FVfeature=Improved Dense Trajectories2015.1285.9
iDT+Fisher vector2020.0284.8