a**********s
发帖数: 588 | 1 C/C++, 希望比较稳定, 快速, self-contained
我的数据量大概是几千到几万, 自变量的维数是四维到十维...
有什么推荐的? thanks! | z**z
发帖数: 222 | 2 is ransac that complicated that you can't write one yourself? | a**********s
发帖数: 588 | 3 It's easy to write it (and I did it in some vision course), it's not easy
to write it fast and stable, say, if you want to take advantage of multi-
processing and if you want to extend it with better randomization.
【在 z**z 的大作中提到】
: is ransac that complicated that you can't write one yourself?
| z**z
发帖数: 222 | 4 sounds like you need a parallel computing course,
what do you mean by better randomization,
you can't get a better distributed/working sampling if you don't know its
pdf etc.
forgive me if i'm rude .....
【在 a**********s 的大作中提到】
: It's easy to write it (and I did it in some vision course), it's not easy
: to write it fast and stable, say, if you want to take advantage of multi-
: processing and if you want to extend it with better randomization.
| a**********s
发帖数: 588 | 5 Thanks. I used my own ransac (which was done in a vision course), but it
doesn't work well enough on my current project. My own implementation
works only well enough for toy project, I guess, that's why I am asking
for a better one, more stable, and faster.
A generic ransac package usually asks for customized distribution, and I
believe there are a lot of paper out there on how to do randomization for
more robust model fitting.
its
【在 z**z 的大作中提到】
: sounds like you need a parallel computing course,
: what do you mean by better randomization,
: you can't get a better distributed/working sampling if you don't know its
: pdf etc.
: forgive me if i'm rude .....
|
|
