visinf / irr

Iterative Residual Refinement for Joint Optical Flow and Occlusion Estimation (CVPR 2019)
Apache License 2.0
194 stars 32 forks source link

Saved check point for things3d_subset #11

Closed Ecalpal closed 5 years ago

Ecalpal commented 5 years ago

Hi, Why didn't you provide the check points after finetune in things3d_subset? I think if there is one, it will make the whole training process clearer. :)

hurjunhwa commented 5 years ago

Hi, Oh, right I forgot uploading it as well. I will prepare it by the end of next week!

Ecalpal commented 5 years ago

Thanks! I'm looking forward to these files.

hurjunhwa commented 5 years ago

Hi, I just uploaded the model trained on things3d-subset :) When testing on SintelTrainingCleanFull, you will get:

Flow EPE = 1.8804 Occlusion F1 score = 0.7122

Best, Jun

Ecalpal commented 5 years ago

Hi, I just uploaded the model trained on things3d-subset :) When testing on SintelTrainingCleanFull, you will get:

Flow EPE = 1.8804 Occlusion F1 score = 0.7122

Best, Jun

Thanks!