The International Journal of Fuzzy Logic and Intelligent Systems

eISSN 2093-744X
pISSN 1598-2645

Improved VGGNet for ensemble (model 3)

Type Filter size / stride Input size
Conv3×3 / 1 224×224×64 
Batch norm --
Conv3×3 / 1224×224×64
Batch norm--
Pooling2×2 / 2112×112×64
Conv3×3 / 1112×112×128
Batch norm--
Conv3×3 / 1112×112×128
Batch norm--
Pooling2×2 / 256×56×128
Conv3×3 / 156×56×256
Batch norm--
Conv3×3 / 156×56×256
Batch norm--
Conv3×3 / 156×56×256
Batch norm--
Pooling2×2 / 228×28×256
Conv3×3 / 128×28×512
Batch norm--
Conv3×3 / 128×28×512
Batch norm--
Conv3×3 / 128×28×512
Batch norm--
Pooling2×2 / 214×14×512
Conv3×3 / 114×14×512
Batch norm--
Conv3×3 / 114×14×512
Batch norm--
Conv3×3 / 114×14×512
Batch norm--
Pooling2×2 / 27×7×512
FC layer-1×1×4096
FC layer-1×1×4096
FC layer-1×1×5
SoftmaxClassfier1×1×5

ReLU layers are omitted from the table for brevity.

Int. J. Fuzzy Log. Intell. Syst. 2017;17:170~176 https://doi.org/10.5391/IJFIS.2017.17.3.170
© IJFIS
© The Korean Institute of Intelligent Systems. / Powered by INFOrang Co., Ltd