Problem when trying to analyze data using several Spv learning method

Hi Stéphane,
I came into some problems when I am trying to do the automatic recognition with PkID.
The analyses always failed when using Spv learing methods 1~4, but 5~7 is always good.
I can't find a solution on the manual, so is there a solve for this?
(I'm quite sure I created the learing file properly and made all the settings right, since method 5~7 can work...)

Thank you!

Ted Chang

Hi Ted,

To answer your question I need the following information:
Did you verify your Tanagra and PkID version ? (please indicate them, see menu>help>about)
Did you uncheck variables that are constant and/or redondant before starting analysis ? (some methods do not accept constants)
Did you use original groups or did you change groups organization in the groups panel ?

thanks

Stéphane

Hi Stéphane,
1. Yes, my PkID is 1.2.6, my tanagra is from the zooscan web site.
2. Yes, only the variables mentioned in the PkID manual is checked, also, new variables from new version of Zooprocess is un-checked.
3. I always use the original groups

Actually, after those constant variables is un-checked, the analysis seems good.

Ted C.

thanks for sharing

___________

hardcore sex
live sex cams

Hi,

I have created learnsets and after analysis by PkId (Plankton Identifier) I can extract vignettes with ImageJ but there is no pictures inside files.

If I check Analyse.txt, I can see only results of 1 sample on 4 samples used.

So It the analyse by PkId which don't working correctly but I don't understand why.

Have you an idea ?

Thank you
ESTIENNE Olivier

ChangCY wrote:Hi Stéphane,
I came into some problems when I am trying to do the automatic recognition with PkID.
The analyses always failed when using Spv learing methods 1~4, but 5~7 is always good.
I can't find a solution on the manual, so is there a solve for this?
(I'm quite sure I created the learing file properly and made all the settings right, since method 5~7 can work...)

Thank you!

Ted Chang