List your company
ResearchFormerly WizWand

Action Classification on Kinetics-400 (val)

90Top-1 AccuracyVideoMAE V2-g
74.478.4582.586.55Apr 23, 2021Aug 18, 2021Dec 14, 2021Apr 10, 2022Aug 6, 2022Dec 1, 2022Mar 29, 2023
Updated 8mo ago

Evaluation results

MethodLinks
VideoMAE V2-gResolution=64 x 266^2, Views=2 x 3, TFLOPS=160.302023.039098.4
MTV-HPre-training Data=WTS 280^2, Views=4 x 3, TFLOPS=73.572023.0389.998.3
VideoMAE V2-HViews=5 x 3, TFLOPS=17.882023.0388.697.9
VideoMAE V2-gViews=5 x 3, TFLOPS=38.162023.0388.598.1
CoVeRPre-training Data=JFT-3B, Views=1 x 32023.0387.2—
MaskFeatViews=4 x 3, TFLOPS=45.482023.038797.4
MAE-STViews=4 x 3, TFLOPS=25.052023.0386.897.2
VideoMAEViews=5 x 3, TFLOPS=17.882023.0386.697.1
MViTv2-LResolution=312^2, Views=40 x 3, TFLOPS=42.422023.0386.197
MARmask ratio (p)=50%, Pre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-L, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=276 x 3 x 5, Param (M)=3112022.0785.396.3
MaskFeatPre-training Dataset=Kinetics-600, Supervised Pre-training=false, Architecture=MVIT-L, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=377 x 1 x 10, Param (M)=2182022.0785.196.6
MAE-vPre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-H, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=1193 x 3 x 7, Param (M)=6322022.0785.196.6
Video Swin-LResolution=384^2, Views=10 x 5, TFLOPS=105.352023.0384.996.7
MAE-vPre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-L, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=598 x 3 x 7, Param (M)=3042022.0784.896.2
MaskFeatPre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=MVIT-L, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=377 x 1 x 10, Param (M)=2182022.0784.396.3
VideoMAEPre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-L, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=597 x 3 x 5, Param (M)=3052022.0783.996.3
MARmask ratio (p)=75%, Pre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-L, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=131 x 3 x 5, Param (M)=3112022.0783.996
Video SwinPre-training Dataset=ImageNet-21K, Supervised Pre-training=true, Architecture=Swin-L, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=604 x 3 x 4, Param (M)=1972022.0783.195.9
MTV-BResolution=320^2, Views=4 x 3, TFLOPS=11.162023.0382.495.2
ViViT FEPre-training Dataset=ImageNet-21K, Supervised Pre-training=true, Architecture=ViT-L, Input Size=128 x 224^2, FLOPs×Cr.×Cl. (G)=3980 x 3 x 12022.0781.793.8
ViViT-L FEViews=1 x 3, TFLOPS=11.942023.0381.793.8
MAE-vPre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-B, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=180 x 3 x 7, Param (M)=872022.0781.394.9
MVITSupervised Pre-training=false, Architecture=MViT-B, Input Size=64 x 224^2, FLOPs×Cr.×Cl. (G)=455 x 1 x 5, Param (M)=372022.0781.295.1
MARmask ratio (p)=50%, Pre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-B, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=86 x 3 x 5, Param (M)=942022.078194.4
TimeSformerPre-training Dataset=ImageNet-21K, Supervised Pre-training=true, Architecture=ViT-L, Input Size=96 x 224^2, FLOPs×Cr.×Cl. (G)=8353 x 3 x 1, Param (M)=4302022.0780.794.7
VideoMAEPre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-B, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=180 x 3 x 5, Param (M)=872022.0780.794.7
TimeSformer-LViews=1 x 3, TFLOPS=7.142023.0380.794.7
Video SwinPre-training Dataset=ImageNet-1K, Supervised Pre-training=true, Architecture=Swin-B, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=282 x 3 x 4, Param (M)=882022.0780.694.6
BEVTPre-training Dataset=IN-1K+DALLE, Supervised Pre-training=false, Architecture=Swin-B, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=282 x 3 x 5, Param (M)=882022.0780.6—
En-VidTr-LInput=32 x 2, GFLOPs=392, Latency (ms)=147.22021.0480.594.6
X3D-XXLInput=16 x 5, GFLOPs=1962021.0480.494.6
MotionformerPre-training Dataset=ImageNet-21K, Supervised Pre-training=true, Architecture=ViT-L, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=1185 x 3 x 10, Param (M)=3822022.0780.294.8
MVITSupervised Pre-training=false, Architecture=MViT-B, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=170 x 1 x 5, Param (M)=372022.0780.294.4
SlowFastSupervised Pre-training=false, Architecture=R101+NL, Input Size=(16+64) x 224^2, FLOPs×Cr.×Cl. (G)=359 x 3 x 10, Param (M)=602022.0779.893.9
SlowFast R101-NLViews=10 x 3, TFLOPS=7.022023.0379.893.9
En-VidTr-MInput=16 x 4, GFLOPs=220, Latency (ms)=98.12021.0479.794.2
En-VidTr-SInput=8 x 8, GFLOPs=130, Latency (ms)=73.22021.0479.494
MARmask ratio (p)=75%, Pre-training Dataset=Kinetics-400, Supervised Pre-training=false, Architecture=ViT-B, Input Size=16 x 224^2, FLOPs×Cr.×Cl. (G)=41 x 3 x 5, Param (M)=942022.0779.493.7
TDNViews=10 x 3, TFLOPS=5.942023.0379.494.4
VidTr-LInput=32 x 2, GFLOPs=351, Latency (ms)=110.22021.0479.193.9
En-I3D-TPN-101Input=32 x 2, GFLOPs=541, Latency (ms)=207.82021.0479.194
TAdaConvNeXt-TPre-training Dataset=ImageNet-1K, Supervised Pre-training=true, Architecture=ConvNeXt-T, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=94 x 3 x 4, Param (M)=382022.0779.193.7
SF101 16x8Input=(64+16)x2, GFLOPs=213, Latency (ms)=124.32021.0478.993.5
TPN101Input=32 x 2, GFLOPs=374, Latency (ms)=133.42021.0478.993.9
VidTr-MInput=16 x 4, GFLOPs=179, Latency (ms)=61.12021.0478.693.5
CorrNet101Input=32 x 2, GFLOPs=1872021.0478.5—
ip-CSNSupervised Pre-training=false, Architecture=ResNet152, Input Size=32 x 224^2, FLOPs×Cr.×Cl. (G)=109 x 3 x 10, Param (M)=332022.0777.892.8
NL101Input=32 x 2, GFLOPs=544, Latency (ms)=134.12021.0477.793.3
TPN50Input=32 x 2, GFLOPs=199, Latency (ms)=89.32021.0477.793.3
VidTr-SInput=8 x 8, GFLOPs=89, Latency (ms)=36.22021.0477.793.3
En-I3D-50-101Input=32 x 2, GFLOPs=509, Latency (ms)=192.72021.0477.793.2
I3D NLViews=10 x 3, TFLOPS=10.772023.0377.793.3
SF101 8x8Input=(32+8)x2, GFLOPs=106, Latency (ms)=71.92021.0477.592.3
Vanilla-TrInput=8 x 8, GFLOPs=89, Latency (ms)=32.82021.0477.593.2
I3D101Input=32 x 2, GFLOPs=342, Latency (ms)=118.32021.0477.492.7
NonLocal I3DPre-training Dataset=ImageNet-1K, Supervised Pre-training=true, Architecture=ResNet101, Input Size=128 x 224^2, FLOPs×Cr.×Cl. (G)=234 x 3 x 10, Param (M)=622022.0777.393.3
CorrNet50Input=32 x 2, GFLOPs=1152021.0477.2—
SF50 8x8Input=(32+8)x2, GFLOPs=66, Latency (ms)=49.32021.047792.6
NL50Input=32 x 2, GFLOPs=282, Latency (ms)=53.32021.0476.592.6
TEINetInput=16 x 2, GFLOPs=66, Latency (ms)=49.52021.0476.292.5
TEA50Input=16 x 2, GFLOPs=702021.0476.192.5
CIDCInput=32 x 2, GFLOPs=101, Latency (ms)=82.32021.0475.592.1
I3D50Input=32 x 2, GFLOPs=167, Latency (ms)=74.42021.047592.2