search

gallenszl / CFNet

CFNet: Cascade and Fused Cost Volume for Robust Stereo Matching（CVPR2021）
MIT License
155 stars 23 forks source link

About the three stage strategy #30

Closed killwy closed 2 years ago

killwy commented 2 years ago

In your paper,you mentioned that "we switch the activation function to Mish and prolong the pre-training process in the SceneFlow dataset for another 15 epochs".So,should the learning rate change in another 15 epochs?

gallenszl commented 2 years ago

Hello, for the prolonging 15 epochs, the learning rate is start at 0.001 and downscaled by 2 at epoch 10.

killwy commented 2 years ago

ok,thanks for your instant reply