search

visinf / irr

Iterative Residual Refinement for Joint Optical Flow and Occlusion Estimation (CVPR 2019)
Apache License 2.0
192 stars 33 forks source link

How to finetune the model? #41

Closed leiwen83 closed 2 years ago

leiwen83 commented 3 years ago

Like if I try to do the quantization finetune over Sintel dataset, is there any parameter needed to be adjust, like learning rate, or something else?

Thx.

hurjunhwa commented 3 years ago

Hi, Probably, you can refer to this script that we used for fine-tuning on Sintel. We just follow the same finetuning scheme from FlowNet2 and PWC-Net. But, depending on your experiment setup, you may find a better set of hyperparameters.

minghuiwsw commented 1 year ago

hi, @hurjunhwa , I noticed you provide the training script for the baseline: pwcnet. But I wonder if it is for flycharis_occ dataset only or it's applicable to other datasets. I mean, if I want to train pwcnet for other datasets to achieve the expected performance in paper, do I need to change the hypermeters in the config file of pwcnet.sh? If yes, can you provide me with those files? Thanks a lot!