OpenCV SVM training dataset -


lets have dataset of 350 positive images , more 400 negative images. aren't same size. size bigger 640x320.

  1. what should create better dataset? need images smaller? if yes, why?

  2. should apply normalization dataset? should (contrast, noise reduction)?

  3. can create bigger dataset using existing one? if yes, how?

thanks in advance!

  1. optimal size of images can classify object yourself.
  2. yes, classifiers works better after normalization, there options. popular ways center dataset (subtract mean) , normalize range of values in [-1:1] range. other popular way of normalization similar previous normalize standard deviation (preferable in cases).
  3. yes, can create bigger dataset existing on adding distorsions , noise images existing dataset.

Comments

Popular posts from this blog

angularjs - ADAL JS Angular- WebAPI add a new role claim to the token -

php - CakePHP HttpSockets send array of paramms -

node.js - Using Node without global install -