error in skimage.feature.hog

[Brian Holt]

Hi Damian,

Here is a link to the pull request I've created to fix the bug
https://github.com/scikits-image/scikits-image/pull/213

You make a good point about the form of the resulting vector.  I would be
quite happy to return a hierarchical (unraveled) ndarray and make it the
responsibility of the user to ravel it if they so wish.  Comments from
others?

Brian

On 11 July 2012 15:20, Damian Eads <dre25 at cam.ac.uk> wrote:

> Thank you for looking into the bug. On another note, it'd be great if
> there was an option to turn off the flattening of the histogram.
>
> Damian
>
> On Tue, Jul 10, 2012 at 10:45 PM, Brian Holt <bdholt1 at gmail.com> wrote:
> > I've had a quick look and I can reproduce the bug. It seems to be a
> > discrepancy between integer division to compute the number of cells and
> then
> > subsampling the array which gives a different dimensions. Hopefully I'll
> > have a better idea tomorrow.
> >
> > On Jul 10, 2012 10:38 PM, "Stéfan van der Walt" <stefan at sun.ac.za>
> wrote:
> >>
> >> On Tue, Jul 10, 2012 at 6:47 AM, Tony Yu <tsyu80 at gmail.com> wrote:
> >> >> Here is how to reproduce the error:
> >>
> >> Damian, I'm also tied up until Thursday, but I'll have a look at it
> ASAP.
> >>
> >> > As for the color/grayscale issue, I've been thinking that it would be
> >> > good
> >> > to have an `@grayscale` function decorator that checks whether an
> image
> >> > is
> >> > rgb/rgba and either raises an error or warns & converts. I haven't
> >> > gotten
> >> > around to implementing it yet, but it should be fairly simple.
> >>
> >> Great idea... The easier we can make it to build robust pipelines, the
> >> better!  The conversion routines have already come a long way, but I
> >> am very much in favour of more such enhancements.
> >>
> >> Stéfan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scikit-image/attachments/20120711/67f01573/attachment.html>