Skip to main content

Table 1 Gradient estimation model architecture

From: Video person reidentification based on neural ordinary differential equations and graph convolution network

The module

Gradient estimation model

Subsampling module (optional)

[Normalized layer, 3×3 convolutional layer, ReLU layer]×1

[Normalized layer, 4×4 convolutional layer (step size 2), ReLU layer]×2

The ODE module

[Normalized layer, ReLU layer, 3×3 convolutional layer]×2

[Normalized layer]×1

Convolution model

Tacit module

[Linear transformation layer, ReLU layer]×3

Fully connected module

[Normalized layer, ReLU layer, global maximum pooling layer, linear transformation layer]

  1. Note: the step size of unmarked convolution layer is 1 by default. The number after the multiplication symbol in brackets indicates the number of times the submodule in brackets is repeated