List your company

Video Action Recognition on Kinetics-400 v1 (test)

79.1Top-1 AccuracyMVFNet_En
69.63672.09374.5577.007Dec 13, 2020

Evaluation results

MethodLinks
MVFNet_EnBackbone=R101+R101, Frames x Crops x Clips=(16+8) x 3 x 10, GFLOPS=188.1 x 302020.1279.193.8
SlowfastBackbone=R101+R101, Frames x Crops x Clips=(16+64) x 3 x 10, GFLOPS=213 x 302020.1278.993.5
TPNBackbone=ResNet-101, Frames x Crops x Clips=32 x 3 x 10, GFLOPS=374 x 302020.1278.993.9
MVFNetBackbone=ResNet-101, Frames x Crops x Clips=16 x 3 x 10, GFLOPS=125.4 x 302020.1278.493.4
SlowfastBackbone=R101+R101, Frames x Crops x Clips=(8+32) x 3 x 10, GFLOPS=106 x 302020.1277.993.2
NL+I3DBackbone=ResNet-101, Frames x Crops x Clips=128 x 3 x 10, GFLOPS=359 x 302020.1277.793.3
SmallBigBackbone=ResNet-101, Frames x Crops x Clips=32 x 3 x 4, GFLOPS=418 x 122020.1277.493.3
MVFNetBackbone=ResNet-101, Frames x Crops x Clips=8 x 3 x 10, GFLOPS=62.7 x 302020.1277.492.9
SlowOnlyBackbone=ResNet-101, Frames x Crops x Clips=16 x 3 x 10, GFLOPS=185 x 302020.1277.2—
MVFNetBackbone=ResNet-50, Frames x Crops x Clips=16 x 3 x 10, GFLOPS=65.8 x 302020.127792.8
ip-CSNBackbone=ResNet-101, Frames x Crops x Clips=32 x 3 x 10, GFLOPS=82 x 302020.1276.792.3
NL+I3DBackbone=ResNet-50, Frames x Crops x Clips=128 x 3 x 10, GFLOPS=282 x 302020.1276.592.6
X3D-MFrames x Crops x Clips=16 x 3 x 10, GFLOPS=6.2 x 302020.127692.3
MVFNetBackbone=ResNet-50, Frames x Crops x Clips=8 x 3 x 10, GFLOPS=32.9 x 302020.127692.4
SlowfastBackbone=R50+R50, Frames x Crops x Clips=(4+32) x 3 x 10, GFLOPS=36.1 x 302020.1275.692.1
TEABackbone=ResNet-50, Frames x Crops x Clips=8 x 3 x 10, GFLOPS=33 x 302020.127591.8
SlowOnlyBackbone=ResNet-50, Frames x Crops x Clips=8 x 3 x 10, GFLOPS=41.9 x 302020.1274.991.5
TEINetBackbone=ResNet-50, Frames x Crops x Clips=8 x 3 x 10, GFLOPS=33 x 302020.1274.991.8
NL+I3DBackbone=ResNet-50, Frames x Crops x Clips=32 x 3 x 10, GFLOPS=70.5 x 302020.1274.991.6
S3D-GBackbone=Inception V1, Frames x Crops x Clips=64 x 3 x 10, GFLOPS=71.4 x 302020.1274.793.4
R(2+1)DBackbone=ResNet-34, Frames x Crops x Clips=32 x 1 x 10, GFLOPS=152 x 102020.1274.391.4
TSMBackbone=ResNet-50, Frames x Crops x Clips=8 x 3 x 10, GFLOPS=33 x 302020.1274.191.2
STMBackbone=ResNet-50, Frames x Crops x Clips=16 x 3 x 10, GFLOPS=67 x 302020.1273.791.6
TSNBackbone=Inception V3, Frames x Crops x Clips=25 x 10 x 1, GFLOPS=80 x 102020.1272.590.2
I3DBackbone=Inception V1, Frames x Crops x Clips=64 x N/A x N/A, GFLOPS=108 x N/A2020.1272.190.3
ECO-RGB_EnBackbone=BNIncep+Res3D-18, Frames x Crops x Clips=92 x 1 x 1, GFLOPS=267 x 12020.1270—