List your company
ResearchFormerly WizWand

Video Action Recognition on HMDB-51 (3 splits)

84.6AccuracySCT-L
55.27262.88670.578.114Nov 29, 2017Oct 3, 2018Aug 7, 2019Jun 10, 2020Apr 14, 2021Feb 16, 2022Dec 21, 2022
Updated 8mo ago

Evaluation results

MethodLinks
SCT-LPretrain=K4002021.0884.6
SMARTPretrain=ImageNet+K400, Modalities=RGB+Flow2022.0884.3
SlowOnly-R101-RGB + I3D-FlowPretrain=OmniSource, Modalities=RGB+Flow2022.0883.8
EVL ViT-L/14Pretrain=CLIP+K400, Modalities=RGB, Resolution=224px2022.0883.6
SCT-MPretrain=K4002021.0883.2
PERF-NetPretrain=ImageNet+K700, Modalities=RGB+Flow+Pose2022.0883.2
EVL ViT-L/14@336pxPretrain=CLIP+K400, Modalities=RGB, Resolution=336px2022.0883.2
SCT-SPretrain=K4002021.0881.5
SCT-LPretrain=ImageNet2021.0881.4
Two-Stream I3DPre-training=Kinetics 300k2017.1180.7
I3DVideo Input Type=Decoded, Input Modality=RGB + Flow, Backbone Architecture=I3D2019.0180.7
I3D_RGB+FlowBackbone=3D Inception2021.0380.7
I3DRGB+FlowBackbone=3D Inception2021.0380.7
Two-Stream I3DPretrain=ImageNet+K400, Modalities=RGB+Flow2022.0880.7
Two-Stream LGD-3DPretrain=ImageNet+K600, Modalities=RGB+Flow2022.0880.5
SlowOnly-R101Pretrain=OmniSource, Modalities=RGB2022.0879
R(2+1)DVideo Input Type=Decoded, Input Modality=RGB + Flow2019.0178.7
I3D RGB + DMC-Net (I3D)Video Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Spatio-temporal modeling, Backbone Architecture=I3D + DMC-Net2019.0177.8
BQNBackbone=TSM R502021.0377.6
BQNBackbone=TSM R50, Inference Protocol=3 crops x 2 clips2021.0377.6
PAN_FullBackbone=TSM R502021.0377
PAN FullBackbone=TSM R502021.0377
SCT-SPretrain=ImageNet2021.0876.5
S3DVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Spatio-temporal modeling2019.0175.9
S3D-GPretrain=K4002021.0875.9
S3DPretrain=ImageNet+K400, Modalities=RGB2022.0875.9
LGD-3DPretrain=K6002021.0875.7
FASTER32Pretrain=K400, Modalities=RGB2022.0875.7
LGD-3DPretrain=ImageNet+K600, Modalities=RGB2022.0875.7
pBYOL (p = 4)Backbone=S3D-G, Modality=R, Pretrain=K4002022.1275
I3D RGBVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Spatio-temporal modeling, Backbone Architecture=I3D2019.0174.8
I3DPretrain=ImageNet+K400, Modalities=RGB2022.0874.8
SCEBackbone=ResNet3D-50, Modality=R, Pretrain=K400, Te=162022.1274.7
MF-NetVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Spatio-temporal modeling2019.0174.6
I3DPretrain=K4002021.0874.5
R(2+1)DPretrain=K4002021.0874.5
R(2+1)D-34Pretrain=K400, Modalities=RGB2022.0874.5
BraVeBackbone=ResNet3D-50, Modality=R+F, Pretrain=K6002022.1274.3
OFF2017.1174.2
BraVeBackbone=ResNet3D-50, Modality=R, Pretrain=K6002022.1274
pBYOL (p = 4)Backbone=ResNet3D-50, Modality=R, Pretrain=K400, Te=162022.1273.6
TSMBackbone=R502021.0373.5
TSMBackbone=R502021.0373.5
TEABackbone=R502021.0373.3
TEABackbone=R502021.0373.3
ST-VLMPF+iDT2017.1173.1
BraVeBackbone=ResNet3D-50, Modality=R+F, Pretrain=K4002022.1272.7
STCPretrain=K400, Modalities=RGB2022.0872.6
DI Four-StreamBackbone=ResNeXt1012021.0372.5
DI Four-StreamBackbone=ResNeXt1012021.0372.5
STMN+iDT2017.1172.2
STMBackbone=R502021.0372.2
STMBackbone=R502021.0372.2
pBYOL (p = 4)Backbone=ResNet3D-50, Modality=R, Pretrain=K400, Te=82022.1272.1
BraVeBackbone=ResNet3D-50, Modality=R, Pretrain=K4002022.1272
ConST-CLBackbone=ResNet3D-50, Modality=R, Pretrain=K4002022.1271.9
DMC-Net (I3D)Video Input Type=Compressed, Backbone Architecture=I3D2019.0171.8
TVNetVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Motion representation learning2019.0171
TVNetBackbone=BNInception2021.0371
TVNetBackbone=BNInception2021.0371
ARTNetVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Spatio-temporal modeling2019.0170.9
SCEBackbone=ResNet3D-50, Modality=R, Pretrain=K400, Te=82022.1270.5
ResNeXtPretrain=K4002021.0870.2
Three-Stream TSN2017.1169.4
STP2017.1168.9
Two-Stream TSN2017.1168.5
TSN_RGB+FlowBackbone=BNInception2021.0368.5
TSNRGB+FlowBackbone=BNInception2021.0368.5
ECOPretrain=K400, Modalities=RGB2022.0868.4
CVRLBackbone=ResNet3D-50, Modality=R, Pretrain=K6002022.1268
CORPBackbone=ResNet3D-50, Modality=R+F, Pretrain=K4002022.1268
CVRLBackbone=ResNet3D-50, Modality=R, Pretrain=K4002022.1266.7
SCEBackbone=ResNet3D-18, Modality=R, Pretrain=K400, Frames=162022.1266.6
Two-Stream I3D2017.1166.4
VideoPromptPretrain=CLIP, Modalities=RGB2022.0866.4
L2STM2017.1166.2
TDD+iDT2017.1165.9
SCEBackbone=ResNet3D-18, Modality=R, Pretrain=K400, Frames=82022.1265.7
Two-Stream fusionVideo Input Type=Decoded, Input Modality=RGB + Flow2019.0165.4
LTC+iDT2017.1164.8
TransRankBackbone=R(2+1)D-18, Modality=R+RD, Pretrain=K2002022.1264.2
TECBackbone=ResNet3D-18, Modality=R, Pretrain=K4002022.1263.6
TECBackbone=S3D-G, Modality=R, Pretrain=K4002022.1263.5
TransRankBackbone=ResNet3D-18, Modality=R+RD, Pretrain=K2002022.1263.5
KVMDF2017.1163.3
TransRankBackbone=ResNet3D-18, Modality=R+RD, Pretrain=UCF1012022.1263
CoCLRBackbone=S3D, Modality=R+F, Pretrain=K4002022.1262.9
DMC-Net (ResNet-18)Video Input Type=Compressed, Backbone Architecture=ResNet-182019.0162.8
PWC-Net (ResNet-18) + CoViARVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Motion representation learning, Backbone Architecture=ResNet-182019.0162.2
TECBackbone=R(2+1)D-18, Modality=R, Pretrain=K4002022.1262.2
SCEBackbone=ResNet3D-18, Modality=R, Pretrain=K2002022.1262
CATEBackbone=ResNet3D-50, Modality=R, Pretrain=K4002022.1261.9
iDT2017.1161.7
TransRankBackbone=R(2+1)D-18, Modality=R, Pretrain=K2002022.1260.1
Two-Stream2017.1159.4
Two-streamVideo Input Type=Decoded, Input Modality=RGB + Flow2019.0159.4
ProViCoBackbone=ResNet3D-18, Modality=R, Pretrain=K4002022.1259.4
CoVIARVideo Input Type=Compressed2019.0159.1
TransRankBackbone=ResNet3D-18, Modality=R, Pretrain=K2002022.1258.1
ActionFlowNetVideo Input Type=Decoded, Input Modality=RGB, Temporal Modeling Approach=Motion representation learning2019.0156.4