I'd like to know if it would make sense to train the Network with pictures with other camera intrinsics. Since I'm planning on working with the new iPad Pro to use the network, I'd like to ask if this would make sense. I was all about to write my own linemod like data generator in swift, but it was too much work with the masked images and stuff, so i decided to use raster triangle instead to generate the data of my own dataset.
I am just not quite sure if i understand it all right, and struggling to grasp the concept of cam intrinsics. Is it just for labeling the images intrinsics or is this the target intrinsics that the network will be able to work with in the end? Or is this just so the transformation and rotation can be determined by the data generator?
This is more of a question than an issue.
I'd like to know if it would make sense to train the Network with pictures with other camera intrinsics. Since I'm planning on working with the new iPad Pro to use the network, I'd like to ask if this would make sense. I was all about to write my own linemod like data generator in swift, but it was too much work with the masked images and stuff, so i decided to use raster triangle instead to generate the data of my own dataset.
I am just not quite sure if i understand it all right, and struggling to grasp the concept of cam intrinsics. Is it just for labeling the images intrinsics or is this the target intrinsics that the network will be able to work with in the end? Or is this just so the transformation and rotation can be determined by the data generator?
Kind regards.