Open horford opened 2 months ago
I think it's not hard to modify our code to fulfill multi-resolution training. Maybe we can schedule a meeting and talk about your need?
Has the problem of compatibility with multiple cameras and different image resolutions been solved?
Hi,
I was wondering if there was a workflow or a method of getting this project to train with a mixed dataset of images. E.g from 2 different cameras.
Currently when trying to train I get an error "AssertionError: All images should have the same size." is there a workaround for this? Other than converting/cropping all my images to the same size.
Many thanks, Harry