Depth for initialization or throughout? #6
Description
Hi @halismai ,
BPVO is a really nice work.
I have been exploring BPVO for its use in a monocular system, and looking into the use of depth for pose estimates - whether it is only used for initialization or throughout (seems to be latter as it is primarily designed for RGB-D).
Is there any part of implementation which I overlooked that estimates depth on frame-frame basis (or similar) so that depth is not explicitly required other than for initialization?
Thanks!