Training Boost clasifier with CvMLData

asked 2013-09-17 04:25:05 -0500

I was trying to train Boost classifier using CvMLData and reading from .cvs file,but when I try to load .cvs the program trows an error:

OpenCV Error: One of arguments' values is out of range (Storage block size is too small to fit the sequence elements) in cvSetSeqBlockSize

The .cvs file has the following format:


Any idea, what causes this?

edit retag flag offensive close merge delete


Reading CSV files depends on your actual file system language settings, which leads to differences between comma's and dots on different systems. Could this be the problem?

StevenPuttemans gravatar imageStevenPuttemans ( 2013-09-17 07:47:08 -0500 )edit

I found it. The .csv file is just toooooooooo large xD (~850mb) I was able to run it, after some modifications (in module ml, data.cpp, read_csv function replase cvCreateMemStorage() with cvCreateMemStorage(128*1024)). Default MemStorage was 64k, now is 128k, but now I'm out of memmory.... I will find another way

Spas Hristov gravatar imageSpas Hristov ( 2013-09-17 07:59:30 -0500 )edit

How many entries does one need to get 850 mb on CSV files? Thats enourmous!

StevenPuttemans gravatar imageStevenPuttemans ( 2013-09-17 08:04:11 -0500 )edit

Yep, 5000 rows X 20 000 cols

Spas Hristov gravatar imageSpas Hristov ( 2013-09-17 08:21:25 -0500 )edit

Can you tell me which changes did you do to make program run? I have the same problem but I don't know how to reslove it :(

jcv gravatar imagejcv ( 2014-09-21 06:32:09 -0500 )edit

Did you have to recompile the libraries for the changes in data.cpp to work? What changes did you make exactly for the program to run? My .csv is 1.9GB :(

nicogc gravatar imagenicogc ( 2019-06-06 13:16:07 -0500 )edit

This question is from 2013. In computer vision that is like the stone age. Please open up a new question because people will tend to ignore these.

StevenPuttemans gravatar imageStevenPuttemans ( 2019-06-26 04:35:04 -0500 )edit