List your company
ResearchFormerly WizWand

Action Recognition on HMDB-51 (average of three splits)

87.2Top-1 AccMaxExp+IDT
60.57667.48874.481.312Nov 24, 2017Aug 9, 2018Apr 25, 2019Jan 9, 2020Sep 24, 2020Jun 10, 2021Feb 24, 2022
Updated 8mo ago

Evaluation results

MethodLinks
MaxExp+IDTPooling Order=TO+, Auxiliary Descriptors=IDT2021.1087.2
SCK(TO+) +IDTPooling Order=TO+, Auxiliary Descriptors=IDT2021.1086.1
MaxExp+IDTPooling Order=SO+, Auxiliary Descriptors=IDT2021.1085.7
SCK(SO+) +IDTPooling Order=SO+, Auxiliary Descriptors=IDT2021.1085.1
SlowOnly-8x8-R101 + FlowPretrain=Kinetics + OmniSource2020.0383.8
EvaNetType=Ensemble2018.1182.3
I3D + PA3DPretrain=ImageNet + Kinetics2020.0382.1
DSP2018.1181.5
DSPBackbone=TS I3D2018.0781.5
ADL+I3DAuxiliary Descriptors=I3D2021.1081.5
Dicrim. Pooling2018.1181.3
Top modelType=Individual2018.1181.3
Full-FT I3DBackbone=I3D2021.1081.3
MaxExpPooling Order=TO+2021.1081.1
Two-stream I3D2018.1180.9
PoTion2018.1180.9
TS I3D2018.0780.9
I3D + PoTionPretrain=ImageNet + Kinetics2020.0380.9
HDPPooling Order=TO+2021.1080.9
Two-stream I3DDim=3D2017.1180.7
I3D two-streamBackbone=3D Inception-v1, Flow=true, Pre-train Data=ImageNet+Kinetics2019.0880.7
Two-Stream-I3DPretrain=ImageNet + Kinetics2020.0380.7
MaxExpPooling Order=SO+2021.1080.3
iTGM-EnsembleType=Ensemble2018.1180.1
HDPPooling Order=SO+2021.1080.1
3D-EnsembleType=Ensemble2018.1179.9
SlowOnly-8x8-R101Pretrain=Kinetics + OmniSource2020.0379
R(2+1)D2018.1178.7
TCM-R50Pretrain=Kinetics, Backbone=ResNet502022.0277.5
Flow-I3DPretrain=ImageNet + Kinetics2020.0377.1
TDNPretrain=Kinetics, Backbone=ResNet502022.0276.3
(no EPN)Pooling Order=SO+2021.1076.1
Fully-supervised SOTABackbone=S3D, Frozen=false2019.1275.9
S3D-GInput=64 x 3 x 224 x 224, Clips=ALL, Backbone=Inception2020.0675.9
S3DPretrain=Kinetics, Backbone=Inception V22022.0275.9
R(2+1)D-RGB & DSNInput=32 x 3 x 112 x 112, Clips=4, M=12020.0675.5
RGB-I3DPre-train dataset=ImageNet+Kinetics, Spatial resolution=224 x 224, Backbone architecture=Inception V12017.1174.8
RGB-I3DPretrain=ImageNet + Kinetics2020.0374.8
I3D-RGBInput=64 x 3 x 224 x 224, Clips=ALL, Backbone=Inception V12020.0674.8
R(2+1)D-RGBInput=32 x 3 x 112 x 112, Clips=10, Backbone=ResNet-34, label=ours2020.0674.8
I3DPretrain=Kinetics, Backbone=Inception V22022.0274.8
(no EPN)Pooling Order=TO+2021.1074.8
MF-NetInput=16 x 3 x 224 x 224, Clips=502020.0674.6
R(2+1)D-RGBInput=32 x 3 x 112 x 112, Clips=102020.0674.5
R(2+1)DPretrain=Kinetics, Backbone=ResNet342022.0274.5
I3D RGBBackbone=3D Inception-v1, Flow=false, Pre-train Data=ImageNet+Kinetics2019.0874.3
DSPBackbone=TS ResNet + IDT2018.0774.3
TSMBackbone=ResNet50, Pre-train=Kinetics, #Frames=82021.0373.5
Ours-cumulativeBackbone=TSM-ResNet50, Pre-train=Kinetics, #Frames=all, #Clusters=82021.0373.4
Ours-slopeBackbone=TSM-ResNet50, Pre-train=Kinetics, #Frames=all, #Clusters=82021.0373.3
TEAPretrain=Kinetics, Backbone=ResNet502022.0273.3
TSMPretrain=Kinetics, Backbone=ResNet502022.0273.2
TSN+TSMAttention=with2018.0972.7
ECOBackbone=BNInception+3D ResNet-18, Flow=false, Pre-train Data=Kinetics2019.0872.4
DSPBackbone=TS ResNet2018.0772.4
TSN+TSMAttention=without2018.0972.2
STMBackbone=ResNet-50, Flow=false, Pre-train Data=ImageNet+Kinetics2019.0872.2
STM Network+IDT2018.0772.2
STMBackbone=ResNet50, Pre-train=ImageNet+Kinetics, #Frames=162021.0372.2
STMPretrain=Kinetics, Backbone=ResNet502022.0272.2
ShuttleNet+MIFS2018.0771.7
SI+DI+OF+DOFBackbone=ResNeXt50, Pre-train=Imagenet, #Frames=dynamic images2021.0371.5
TLE:Bilinear2018.0971.1
SVMP2018.0771
ARTNet with TSNPre-train dataset=Kinetics, Spatial resolution=112 x 112, Backbone architecture=ResNet-182017.1170.9
ARTNet with TSNBackbone=3D ResNet-18, Flow=false, Pre-train Data=Kinetics2019.0870.9
GRP2018.0770.9
ARTNet with TSNInput=24 x 3 x 112 x 112, Clips=25, Backbone=ResNet-182020.0670.9
ARTNetPretrain=Kinetics, Backbone=ResNet182022.0270.9
TSMBackbone=ResNet-50, Flow=false, Pre-train Data=ImageNet+Kinetics2019.0870.7
ST-ResNet+IDT2018.0770.3
ResNeXt-101Dim=3D, Input=64f2017.1170.2
ActionVLAD+iDTBackbone=VGG162018.0969.8
TSNDim=2D2017.1169.4
Temporal Seg. n/w2018.0769.4
TSNPretrain=ImageNet2020.0369.4
Fusion+iDT2018.0969.2
Two-stream+IDT2018.1169.2
ST Multiplier NetDim=2D2017.1168.9
Spatiotemporal Multiplier Nets2018.0968.9
Spatiotemporal Pyramid Net2018.0968.9
STM Network2018.0768.9
TSNBackbone=BN-Inception2018.0968.5
ARTNet w/o TSNPre-train dataset=Kinetics, Spatial resolution=112 x 112, Backbone architecture=ResNet-182017.1167.6
STCBackbone=ResNet101, Flow=false, Pre-train Data=Kinetics2019.0866.8
Spatiotemporal ResNets2018.0966.4
ST-ResNet2018.0766.4
L2STM2018.0766.2
Two-Stream CNN Fusion2018.0965.4
TSMBackbone=ResNet50, Pre-train=Kinetics, #Frames=1, Implementation=our impl.2021.0365.1
TSNBackbone=ResNet50, Pre-train=Kinetics, #Frames=82021.0364.7
Hierarchical Attention Nets2018.0964.3
ResNeXt-101Dim=3D2017.1163.8
Multi-Granular Nets2018.0963.6
Key Volume Mining2018.0963.3
TDDDim=2D2017.1163.2
VideoLSTM+iDT(FV)2018.0963
C3DPre-train dataset=Kinetics, Spatial resolution=112 x 112, Backbone architecture=ResNet-182017.1162.1
AVTSPre-training Dataset=Audioset, Modality=Audio, Backbone=MC3, Frozen=false2019.1261.6
R3D-RGB & DSNInput=8 x 3 x 112 x 112, Clips=3, M=32020.0661.6