Hyperparameters for training

[YoussefFathi]

Hello,
would like to congratulate you on such great work!

Are the hyperparameters for the kitti_raw dataset included in params_kitti_raw.yaml the same ones used to reach the results in the paper or should they be changed?

[vincentfung13]

Hi Youssef,

Yes you should be able to reproduce the performance by using params_kitti_raw.yaml, you can change it accordingly if you want to reproduce the results for different settings in the paper (for example the training resolution).

Zijian

[YoussefFathi]

It was mentioned in the MPI paper that for KITTI Raw they cropped all images by 5% from all sides before testing and also calculated the metrics only on non-occluded regions of the scene. Did you do the last part in your method or only cropped by 5% from all sides?

[hyBlue]

@YoussefFathi
Hi, did you figure it out how to evaluate? I'm now testing the author's pretrained model but I can never get the results as in the paper. I evaluated PSNR and it seems around 15. on in the test set. Could you please share the tips if you get the good results?