big model xml file,trained by CvRTrees, load fails!

i train a CvRTrees, and get the conreponding model writen in xml format, which is too big over 1.6G, but is very good for my project.

now, when i reload the model to predict new data, it fails: runtime error!

early i tried the adaboost. in which condition model file was more than 700M and same thing happened: model loading by load() function failed!

so, it is again and again to retrain the model as long as i need to deal with new data!

is it a bug about load() function of ML algrithms in OpenCV, which can not deal with too big xml file containing model data???

does anybody has any solution!?

add a comment

1 answer

Sort by » oldest newest most voted

answered Jul 26 '13

Spas Hristov

501 ●2 ●3 ●16 http://fb.com/spas.hri...

I have similar problem. I have trained real boost with huge DS (around 33.000 samples, took 9h) and now when I try to load the XML file it crashes. Any idea ?

Comments

can you tell me how exactly did you load the trained model (.xml file) ? I am also using CvRTrees and I have saved the trained model correctly. What's the exact command to load the model?