Giters

cleinc / bts

From Big to Small: Multi-Scale Local Planar Guidance for Monocular Depth Estimation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Question about the image size during training and evaluation

kamiLight opened this issue · comments

commented

you use the random crop with 704x352 to train the model, and i want to know how to evaluate the model on 1242x384 images with eigen splits? do you crop to 704x352 during evaluation, if so, random crop or central crop? if not crop just use high resolution, will it cause other issues as change the input size?

commented

@kamiLight Please check our implementation from bts_eval.py. We feed original images without cropping then masking on them to evaluate. The only requirement is that image height and width should be multiples of 32.