search

megvii-research / RealFlow

The official implementation of the ECCV 2022 Oral paper: RealFlow: EM-based Realistic Optical Flow Dataset Generation from Videos
Other
93 stars 11 forks source link

How many iterations of GRU used in RAFT forward operation of raft-RFAB.pth pretrained model ? #2

Closed ducheng678 closed 1 year ago

ducheng678 commented 1 year ago

I just notice that you set iterations when kitti to 24 and sintel to 32, but have no idea about your RAFT model setting trained by RF-AB dataset.

coolbeam commented 1 year ago

I just notice that you set iterations when kitti to 24 and sintel to 32, but have no idea about your RAFT model setting trained by RF-AB dataset.

Hi, RAFT model setting trained by RF-AB dataset is the same: 12 when training, 24 when testing on kitti and 32 when testing on sintel. Thank you.

ducheng678 commented 1 year ago

I just notice that you set iterations when kitti to 24 and sintel to 32, but have no idea about your RAFT model setting trained by RF-AB dataset.

Hi, RAFT model setting trained by RF-AB dataset is the same: 12 when training, 24 when testing on kitti and 32 when testing on sintel. Thank you.

Get it. Thanks a lot