microsoft/MaskFlownet

Does image size matters?

JohnnieXDU opened this issue · 1 comments

Hi, really appreciate for opening the code! Can i have few questions?

Does the image size influence the performance? I mean in the training stage, image patches (896x320 for Kitti) are used. But in testing, do you still use the image patches? Or use the entire image (around 1240x370 for Kitti)?

I know cropping is for augmentation, but If you use small pates for training, and larger images for testing, does this strategy will influence the performance?

Thank you very much!

Hi Johnnie, thanks for your interest in our work!

To fairly evaluate the overall performance, the entire image is used for testing, and we believe that the image size will influence but not that much. For KITTI dataset, as we discussed in the 5.1 section in the paper, input images are resized to 1280×576 (before augmentation and cropping) since the decreased aspect ratio better balances the vertical and horizontal displacement, so, for testing, a similar resize should be applied to keep the aspect ratio the same.

Hope this solve your questions.