what techniques might improve vslam performance?
I have implemented the vslam_system and am in the process of testing it with some of my own data sets. I was wondering if there are any plans to improve the computational speed in the future. At the moment the computational performance is very acceptable in the beginning of a run and typically taking about 12.51 ms between frames, but eventually (after approximately 100 frames), it slows down to upwards of 890 ms between frames.
Comments
Would you mind re-titling this as a question? I'm hoping to set a trend to make browsing and answering questions easier. Perhaps something like "Is there a way to improve VSLAM performance?
Asomerville
(
2011-05-12 09:18:07 -0600
)editadd a comment
