search

JXT210 / NIPNet

2 stars 0 forks source link

Input concat form #2

Open shaokangXu opened 4 days ago

shaokangXu commented 4 days ago

Excuse me, may I ask why the two inputs are spliced in the batchSize dimension before they are fed into the network。

Code: image

JXT210 commented 4 days ago

分别提取固定图像和运动图像的多尺度特征，然后共享一个编码器的权重

shaokangXu commented 7 hours ago

In that case, is the size of BatchSize only set to 1 during training?