What steps will reproduce the problem?
use wndchrm 1.32-378. Laboratory of Genetics/NIA/NIH
lgchrm12 at /home/kimmeljc/Celegans/WholeWorms/interpolated_ROIs_orlov/01_ROI
run [eckleyd@lgchrm12 01_ROI]$ wndchrm classify -l FacingL7class.fit
worm15_day5_01.tif
compared to
wndchrm 1.32-371_372M. Laboratory of Genetics/NIA/NIH
niairplgchrm14 at /home/eckleyd
run [eckleyd@niairplgchrm14 ~]$ wndchrm classify -l
/home/eckleyd/RealTimeAssification/FacingL7class.fit worm15_day5_01.tif
What is the expected output? (using older version of wndchrm above)
image norm. fact. p(01) p(02.5) p(04) p(06) p(07) p(08) p(11) act. class pred.
class pred. val.
worm15_day5_01.tif 4.82e-24 0.116 0.198 0.272 0.163 0.087 0.069 0.095 * 04 4.881
What do you see instead?
image norm. fact. p(01) p(02.5) p(04) p(06) p(07) p(08) p(11) act. class pred.
class pred. val.
Segmentation fault (core dumped)seg fault(core dumped)
Original issue reported on code.google.com by dmarkeck...@gmail.com on 16 Aug 2012 at 7:23
Original issue reported on code.google.com by
dmarkeck...@gmail.com
on 16 Aug 2012 at 7:23