search

simonmeister / UnFlow

UnFlow: Unsupervised Learning of Optical Flow with a Bidirectional Census Loss
MIT License
294 stars 57 forks source link

Some question about the result of flownet CS on kitti_trainning_2015 #58

Closed KarelZhang closed 5 years ago

KarelZhang commented 5 years ago

hi, @simonmeister when I evaluate the Unflow-CS on kitti_trainning_2015, I got a wrong result as flow: -- evaluating 'CS': 1/10 -- evaluating 'CS': 2/10 -- evaluating 'CS': 3/10 -- evaluating 'CS': 4/10 -- evaluating 'CS': 5/10 -- evaluating 'CS': 6/10 -- evaluating 'CS': 7/10 -- evaluating 'CS': 8/10 -- evaluating 'CS': 9/10 -- evaluating 'CS': 10/10 (CS) EPE_noc = 23.534972286224367 (CS) EPE_all = 28.98195629119873 (CS) outliers_noc = 86.66220474243164 (CS) outliers_all = 87.88960418701171

But the result of Unflow-C is correct. So do you think where is wrong?

Thanks!!